AI Reflex OS
GenAI Adoption Reflex Area
2026-09-18
Table of Contents
Understanding Current GenAI Capabilities
- Which parts of this work can current GenAI handle reliably enough to support recurring use?
- What can GenAI produce or accomplish here when it has the right context, tools, and information?
- Which parts still depend heavily on human direction, correction, or verification?
- How much does performance vary across realistic cases rather than carefully chosen examples?
- What kinds of errors or limitations would matter most in this work?
- Does the capability remain useful after the effort required to prepare context and review the result is included?
- What should we conclude about where current GenAI is genuinely ready to change how this work is done?
Reviewing Capabilities That Have Recently Become Practical
- Which GenAI capabilities have improved enough to make previously weak or impractical work worth reconsidering?
- What can now be done reliably enough that our earlier assumptions may no longer hold?
- Have improvements in reasoning, multimodality, context handling, tool use, or agentic execution changed what is feasible?
- Which previous barriers have actually disappeared, and which remain despite better models?
- Does the improvement change only the quality of the interaction, or the economics of the underlying work?
- What real organizational task would tell us whether this capability has crossed a useful operational threshold?
- Which existing adoption decisions should we revisit if the improvement proves meaningful in practice?
Distinguishing Mature Capabilities From Impressive Demonstrations
- Does this capability work consistently across realistic variations of the task, or only under favorable conditions?
- What evidence shows that ordinary users can reproduce the result rather than relying on an expert demonstration?
- Which parts of the demonstration depend on unusually clean data, detailed instructions, or carefully selected examples?
- How often does the capability fail, and are those failures easy enough to detect and recover from?
- Does the result remain worthwhile after supervision, verification, latency, and cost are included?
- What changes when the capability is used repeatedly at normal organizational scale?
- What evidence would justify treating this as routine capability rather than continued experimentation?
Understanding Which Types of Work GenAI Handles Well
- What characteristics of this work make it particularly suitable or unsuitable for current GenAI?
- How much of the task involves language, information processing, comparison, generation, analysis, or other cognitive work GenAI handles well?
- Is the relevant context available in a form GenAI can actually use?
- Can a knowledgeable person verify the result without recreating most of the work?
- How much does the task depend on tacit knowledge, ambiguous objectives, physical action, or subtle interpersonal judgment?
- Would GenAI meaningfully improve speed, quality, breadth, or capacity rather than merely produce another version of the same work?
- Which recurring work patterns should we treat as the strongest candidates for GenAI support?
- What kinds of documents, data, images, audio, or other material does this work actually involve?
- Which parts of that material can current GenAI interpret reliably enough to be useful?
- Where could combining several input types reveal information or support work that is difficult today?
- How do input quality, document structure, missing context, or very large source sets affect the result?
- When would retrieval, code execution, or other tools be more reliable than simply placing more material into the conversation?
- What information cannot safely or practically be provided to the available GenAI environment?
- Which multimodal interaction would create enough practical value to justify testing in real work?
- What becomes possible when GenAI can search, retrieve current information, use organizational systems, or call other tools?
- Which parts of this work require information that the model itself does not already possess?
- When does tool access improve reliability rather than merely expand what GenAI can attempt?
- Which external actions could GenAI safely prepare, and which could it reasonably execute?
- What permissions, confirmations, or human authorization should apply before an action affects another system or person?
- How would we detect and recover from an incorrect search result, tool call, or partially completed action?
- Which tool-enabled capability changes this work enough to deserve adoption attention?
Assessing How Far GenAI Can Support Complex Multi-Step Work
- How much of this workflow can GenAI complete coherently before meaningful human direction is needed?
- Which steps can be delegated safely, and which depend on judgment, authority, or context that should remain human?
- Where could errors compound across several steps without becoming immediately visible?
- What checkpoints would make failures detectable before they affect later work?
- Would decomposing the work into smaller human-directed interactions produce a better result than greater autonomy?
- How easily can the workflow recover when information changes, an exception appears, or one step fails?
- What evidence would justify allowing GenAI to handle more of this workflow independently?
Recognizing Where Current GenAI Still Requires Strong Human Judgment
- Which parts of this work involve objectives, values, tradeoffs, or consequences that GenAI should not determine on its own?
- What important local, interpersonal, or organizational context may be unavailable to the model?
- Where does evaluating the output require professional expertise that GenAI cannot reliably replace?
- Which plausible-looking errors would be difficult for a nonexpert to notice?
- Who must retain authority and accountability even if GenAI performs most of the underlying analysis or preparation?
- Which unusual or consequential cases should always return to human judgment?
- Where would reducing human involvement create more risk than the additional automation is worth?
- Which available tools perform best on the representative work we actually need to support?
- How do they differ in reliability, reasoning, multimodal capability, context handling, search, tool use, and agentic execution?
- Which tool has access to the organizational information and systems required for the work?
- How do usability, latency, limits, and integration affect whether employees will actually use each option?
- What difference remains after the cost of the model, supporting infrastructure, and human review is included?
- Do different kinds of work justify using different tools rather than choosing one organizational winner?
- Which differences are material enough that employees need guidance about which tool to use when?
Updating Organizational Assumptions About What GenAI Can Do
- Which assumptions about GenAI capability were formed under older technology and may no longer be accurate?
- Which limitations still genuinely constrain us despite recent improvements?
- Where are we expecting capabilities that remain less reliable than our plans assume?
- What new evidence would be strong enough to change our current view?
- Which policies, workflows, training materials, or opportunity priorities depend on assumptions that have changed?
- Has any capability improvement crossed a threshold that materially changes what is feasible or economical for us?
- What should we reconsider now, and what should remain stable despite continuing technology change?
Understanding the Business and Competitive Impact of GenAI
Identifying Where GenAI Is Already Changing Business Activity
- Which business activities are showing durable change from GenAI rather than isolated experimentation?
- What has actually changed in how the work is performed, not merely in which tool people use?
- Where is GenAI improving speed, quality, capacity, customer experience, or access to expertise?
- Which activities are becoming economical to perform more often or at greater scale?
- Where is visible GenAI adoption producing little change in the underlying business activity?
- What evidence separates demonstrated operational change from vendor claims or promotional examples?
- Which changes elsewhere are relevant enough that we should examine the equivalent activity in our own organization?
Understanding Where GenAI Could Improve Productivity
- Which parts of our work consume significant cognitive effort that GenAI could realistically reduce?
- Where is the task frequent enough that a modest improvement would compound into meaningful capacity?
- Does GenAI reduce the full effort required, including context preparation, review, correction, and downstream rework?
- Is the activity on a critical path where faster completion would actually improve the wider workflow?
- What would happen to the time released if the task became substantially faster?
- Could greater output simply create additional review, coordination, or low-value work elsewhere?
- Which productivity opportunities remain attractive after those downstream effects are included?
Understanding How GenAI Could Affect Cost, Capacity, and Revenue Potential
- Which existing costs could GenAI reduce without shifting equivalent effort somewhere else?
- Where could GenAI release genuinely usable capacity rather than small fragments of time that cannot be redeployed?
- What additional work, customers, cases, or analysis could current teams handle if the capability proves reliable?
- Which activities could become economical at a level of personalization, depth, or frequency that was previously too expensive?
- Could GenAI reduce dependence on scarce expertise, external suppliers, or future hiring?
- What new or improved offering could create revenue rather than merely reduce internal effort?
- Which potential economic effects are large and credible enough to influence organizational priorities?
Identifying How GenAI Could Change Decision-Making
- Where is decision quality currently limited by slow research, fragmented evidence, narrow alternatives, or weak preparation?
- How could GenAI improve the information available to decision-makers without taking over the decision itself?
- Which assumptions, alternatives, or tradeoffs could be examined more systematically with GenAI support?
- Could GenAI materially shorten decision preparation without reducing the evidence or judgment required?
- Where could persuasive or overly confident output make a decision look better supported than it actually is?
- Which parts of the decision still require contextual knowledge, authority, values, or accountability that must remain human?
- Which decision processes are important enough that improved preparation could create meaningful organizational value?
Understanding How GenAI Is Changing Customer Expectations
- Which customer expectations appear to be changing because GenAI is making faster, more available, or more personalized service possible?
- What evidence shows that customers actually value these changes rather than providers merely offering them?
- Which current customer experiences may become comparatively slow, generic, or difficult as alternatives improve?
- Where could GenAI raise expectations for explanation, responsiveness, access, or customization?
- Which expectations would be uneconomic, unsafe, or undesirable for us to satisfy even if competitors pursue them?
- How might widespread GenAI availability change what customers are willing to pay for or perform themselves?
- Which expectation changes are material enough that we should reconsider our current customer experience?
Assessing How Competitors Are Using GenAI
- What GenAI-supported changes can we observe in competitors' actual products, services, costs, speed, or operating models?
- Which competitor activities appear to create durable value rather than publicity or experimentation?
- What evidence suggests GenAI is changing their economics or customer proposition?
- Which advantages come from GenAI itself, and which come from proprietary data, workflow integration, distribution, or organizational execution?
- Could the same underlying capability be available to us without copying the competitor's implementation?
- Where might competitors be learning or building capabilities that compound over time even before the financial effect is obvious?
- Which competitive developments should change our priorities, and which should simply be monitored?
Exploring How GenAI Could Change the Industry
- Which parts of the industry's value chain become materially cheaper, faster, or easier as GenAI capability improves?
- Which expertise or services that were previously scarce could become much more widely available?
- Where could GenAI reduce the value of intermediaries whose main role is information processing or coordination?
- What new entrants or business models become viable if cognitive work becomes substantially cheaper?
- Which current sources of differentiation remain defensible when competitors can access similar frontier models?
- How could industry-wide productivity gains change prices, service standards, customer expectations, or competitive intensity?
- Which possible industry changes are sufficiently plausible that we should prepare before they are fully established?
Identifying New Sources of Value Enabled by GenAI
- What valuable activity would we perform more often or more deeply if its cognitive cost fell substantially?
- Which customer or internal need is currently underserved because individualized analysis, preparation, or support is too expensive?
- What could GenAI enable that is qualitatively different from making an existing task faster?
- Which organizational information or expertise could become useful to many more people through GenAI?
- Where could GenAI allow us to combine capabilities that currently sit in different specialist roles or functions?
- Which new possibilities depend on assets or knowledge that are distinctive to our organization?
- What new source of value is promising enough to test even though there is no existing process to optimize?
Understanding the Cost of Adopting GenAI Too Slowly
- Which valuable improvements are we foregoing while we delay building practical GenAI capability?
- Where could competitors accumulate workflow, data, learning, or customer advantages that become harder to close later?
- Which organizational capabilities take long enough to develop that waiting creates a compounding disadvantage?
- Could delayed adoption make future change more disruptive because work, systems, or expectations have already moved elsewhere?
- Where would waiting actually be rational because the technology, economics, or use case remain too immature?
- What is the cost of delay compared with the cost and risk of moving prematurely?
- Which areas justify faster learning now even if large-scale implementation is not yet warranted?
Translating GenAI Developments Into Strategic Implications
- Which GenAI developments materially change assumptions underlying our current strategy?
- Does the development change what customers can expect, competitors can do, or our organization can economically provide?
- Which existing strategic option becomes more or less attractive because of the capability change?
- What new source of advantage or vulnerability does the development create for us specifically?
- Is the effect likely to come from the model itself or from how organizations combine it with data, workflows, people, and other assets?
- What evidence is still missing before we change a consequential strategic decision?
- Which strategic question should leadership reopen now because the underlying conditions have genuinely changed?
Assessing the Current Adoption Baseline
Determining Who Currently Has Access to GenAI
- Which people currently have working access to GenAI tools that are adequate for their actual work?
- Where does nominal access differ from practical access because of devices, authentication, permissions, limits, or usability?
- Which groups can use GenAI but cannot reach the organizational information or systems their work requires?
- Are some employees relying on personal or unapproved tools because the approved environment does not meet their needs?
- Where do differences in role, location, employment type, or technical environment create unequal access?
- Which groups have enough access for experimentation but not for meaningful recurring professional use?
- What does the evidence show about the organization's actual usable-access baseline?
Understanding How Many People Are Actually Using GenAI
- How many people are genuinely using GenAI for work rather than merely having an account or opening the tool?
- How many have moved beyond a first experiment into recurring use?
- What usage is visible in approved-tool telemetry, and what use may be occurring through personal accounts, embedded features, APIs, or other unobserved tools?
- Where do employee reports and centrally visible usage tell different stories?
- Which groups with adequate access are barely using GenAI, and which groups appear more active than official data suggests?
- What evidence distinguishes meaningful professional use from incidental or superficial interaction?
- What is our best-supported estimate of actual participation once the different evidence sources are combined?
Assessing How Frequently People Use GenAI
- How often do people return to GenAI for real work after their initial experiments?
- Which users repeatedly use GenAI when the same kinds of work situations occur?
- Where is usage frequent but shallow, and where is less frequent use tied to substantial work?
- Which groups experimented briefly and then stopped returning?
- What recurring work situations appear to trigger sustained use?
- How much of the observed frequency reflects the nature of the work rather than stronger or weaker adoption?
- What does the pattern of recurrence tell us about whether useful GenAI habits are actually forming?
Identifying What People Currently Use GenAI For
- What kinds of real work are employees currently using GenAI to support?
- How much use remains concentrated in drafting, summarization, and other obvious entry-level applications?
- Where are employees using GenAI for analysis, comparison, research, preparation, challenge, or multi-step work?
- Which uses recur as part of normal work rather than appearing only in isolated experiments?
- What important practices may be occurring through tools or accounts that central telemetry does not capture?
- Which potentially valuable work areas show little GenAI use despite having suitable opportunities?
- What does the current mix of uses reveal about the breadth and depth of adoption?
Comparing Adoption Across Teams and Functions
- Which teams differ materially in the breadth, recurrence, independence, or depth of their GenAI use?
- Are we comparing teams that actually perform sufficiently similar work for the comparison to be meaningful?
- How do access, tool integration, manager support, learning time, and local opportunity differ between higher- and lower-adoption teams?
- Which differences appear to come from the nature of the work rather than from adoption conditions?
- Are higher-use teams also creating greater value, or simply interacting with GenAI more often?
- Which practices or conditions from stronger teams appear transferable elsewhere?
- What should we investigate before treating one team's adoption level as a benchmark for another?
Understanding Current Levels of GenAI Confidence
- How confident are people using GenAI on the kinds of work they actually encounter?
- Where does low confidence prevent worthwhile experimentation even though the user has adequate access and opportunity?
- Where is confidence high despite weak ability to judge, verify, or correct the output?
- Which tasks produce appropriately cautious behavior because errors would be difficult to detect or consequential?
- Can users distinguish situations where GenAI is dependable from situations where stronger verification or human judgment is needed?
- How does confidence compare with demonstrated capability rather than self-reported comfort alone?
- Which confidence problems are most likely to limit useful adoption or create unsafe reliance?
Assessing How Independently People Can Use GenAI
- Can users recognize worthwhile GenAI opportunities without waiting for somebody else to suggest them?
- Can they explain the situation, objective, constraints, and relevant context clearly enough to obtain useful help?
- Can they move beyond a weak first response by diagnosing what needs to change?
- Can they judge when an output is usable, when it needs verification, and when they lack the expertise to assess it?
- Can they adapt a successful interaction to a different task rather than depending on exact templates or prompts?
- Can they recognize when GenAI is the wrong tool or adds more work than value?
- Where do users still depend on champions, specialists, or fixed instructions for work they should eventually handle themselves?
Understanding How Managers Currently Support GenAI Use
- How do managers currently help employees connect GenAI with the team's real work?
- Are employees given legitimate working time to experiment and develop capability?
- How do managers respond when a GenAI experiment fails or produces an uncertain result?
- Do managers create practical permission to use GenAI within organizational boundaries?
- Can managers distinguish a user capability problem from poor task fit, weak tools, policy uncertainty, or workload constraints?
- Where are local expectations, incentives, or workload pressures unintentionally discouraging useful adoption?
- How dependent is current adoption on whether an employee happens to have a supportive manager?
Identifying Existing Examples of Valuable GenAI Use
- Which existing GenAI uses have produced evidence of better speed, quality, capacity, decisions, or newly enabled work?
- Does the benefit remain after preparation, verification, correction, and support effort are included?
- Which examples have worked repeatedly rather than only in one favorable case?
- Can people other than the original user reproduce the result?
- What conditions, expertise, information, tools, or managerial support made the practice successful?
- What limitations or failure modes would another team need to understand before trying it?
- Which examples reveal a transferable opportunity rather than merely an impressive local success?
Creating an Overall Picture of the Current Adoption Baseline
- What does the combined evidence show about access, experimentation, recurrence, breadth, depth, independence, and integration?
- Where do telemetry, employee reports, manager observations, and examples of actual work agree?
- Where do those evidence sources conflict, and what might the disagreement itself reveal?
- Which parts of the organization show meaningful recurring adoption rather than simple availability or experimentation?
- What appears to explain the largest differences between groups?
- Which important aspects of adoption remain poorly measured or unknown?
- What should leadership understand about the current state before choosing the next adoption priorities?
Understanding How People Currently Work
Identifying the Main Responsibilities of Different Roles
- What outcomes is this role actually accountable for producing?
- Which responsibilities consume substantial time, attention, or scarce expertise?
- Which responsibilities depend heavily on judgment, relationships, or legitimate decision authority?
- Where does the role spend significant effort processing, creating, comparing, or interpreting information?
- Which responsibilities are recurring enough that a better working method could compound into meaningful value?
- What dependencies on other people, systems, or information shape how the role can actually work?
- Which responsibilities are most worth examining for GenAI support without assuming that the whole role should change?
- What work does this person repeatedly perform over a normal week, month, or operating cycle?
- Which recurring tasks follow recognizable patterns despite variation between individual cases?
- Where do people repeatedly search, read, draft, compare, classify, analyze, prepare, or review?
- Which tasks consume disproportionate effort relative to the judgment they require?
- What tasks repeatedly create downstream work, handoffs, or verification for others?
- Which recurring activities are suitable enough for GenAI that users could develop stable habits around them?
- Which task patterns deserve closer examination before we design adoption support or workflows around them?
- What information do people need in order to perform this work well?
- Where does that information currently live, and how difficult is it to find and combine?
- Which parts of the work involve repeatedly searching across documents, systems, messages, or other sources?
- Where do people manually reconstruct context that the organization already possesses elsewhere?
- Which information sources are authoritative, and where do conflicting or outdated versions create problems?
- What information could GenAI access safely and reliably, and what would remain inaccessible?
- Where would better GenAI-supported information use materially improve the work rather than merely make search more convenient?
Identifying Repetitive Knowledge Work
- Which cognitive activities recur with similar reasoning or information-processing patterns?
- Where are people repeatedly recreating comparable drafts, analyses, comparisons, summaries, or classifications?
- Which recurring tasks are difficult to automate conventionally because the inputs vary from case to case?
- How much judgment does each task genuinely require, and where is the effort mostly routine cognitive processing?
- What quality or exception risks would appear if GenAI handled more of the repetitive work?
- Would the verification effort remain low enough for GenAI to create a real advantage?
- Which repetitive activities are frequent and valuable enough to justify changing the working method?
Understanding Where Important Decisions Are Made
- Which recurring decisions materially affect customers, resources, commitments, risk, or subsequent work?
- What information, analysis, and preparation currently support those decisions?
- Where do decision-makers receive evidence too late, too narrowly, or in inconsistent form?
- Which alternatives, assumptions, or tradeoffs are often insufficiently explored?
- Could GenAI improve preparation while leaving the actual judgment and authority with the appropriate person?
- What information would need verification before GenAI-supported analysis could safely influence the decision?
- Which decision points would benefit enough from better preparation to justify adoption effort?
Mapping How Work Moves Between People and Teams
- How does this work move from one person or team to another from beginning to completion?
- What information, decisions, or context must travel with each handoff?
- Where is information repeatedly lost, reconstructed, reformatted, or clarified?
- Which queues or handoffs create delays that faster individual work would not solve?
- Where could GenAI reduce coordination effort, prepare the next handoff better, or preserve continuity?
- Which handoffs exist for reasons such as authority, specialist judgment, or control that GenAI should not simply remove?
- Which cross-team workflows deserve deeper examination because GenAI could change the way the work is divided?
Identifying Common Sources of Friction and Delay
- What repeatedly slows this work down or makes it harder than it needs to be?
- Which delays come from information search, preparation, rework, coordination, or waiting for expertise?
- Where do unclear inputs or requirements create avoidable loops?
- Which activities consume effort simply to translate, reorganize, or reformat information between people or systems?
- Which frictions could GenAI realistically reduce under current capabilities?
- Which delays would remain even if every employee became highly capable with GenAI?
- What should be fixed structurally rather than mislabeled as a GenAI adoption opportunity?
- Which systems and tools are essential to completing this work today?
- What information or capability does each tool contribute that GenAI would need to complement?
- Where do employees repeatedly switch systems, copy information, or manually recreate context?
- Which tools already contain embedded GenAI that employees may not think of as separate AI use?
- What important information or actions are unavailable to the organization's current GenAI tools?
- Where would integration create a larger benefit than improving the standalone model?
- Which existing tools should GenAI complement, and which old tool interactions might become unnecessary?
- Where have employees created their own ways of completing work because formal tools or processes do not serve them well?
- What personal spreadsheets, templates, notes, scripts, external tools, or GenAI services are people relying on?
- What underlying problem is each workaround actually solving?
- Which workarounds reveal unmet requirements that the formal environment has failed to address?
- Where does a workaround create privacy, quality, continuity, or dependency risks?
- Could GenAI or a better integration remove the need for the workaround, or would it merely create another workaround?
- What should we learn from existing informal practice before redesigning the official way of working?
Understanding Which Work Patterns Are Most Relevant to GenAI Adoption
- Which recurring work patterns appear across enough roles or teams to matter organizationally?
- Where do similar problems arise even when the job titles or functions are different?
- Which patterns involve substantial information, language, analysis, preparation, comparison, or knowledge work?
- Which patterns combine sufficient frequency, value, and verifiability to support broad GenAI adoption?
- What work remains too dependent on physical execution, tacit context, or consequential judgment for GenAI to add much value?
- Which patterns could support reusable learning or shared methods without forcing identical workflows across teams?
- What should these recurring patterns change about how we design the adoption approach?
Understanding Different User Groups
Identifying Meaningfully Different Groups of GenAI Users
- Which differences between users actually change the adoption problem they face?
- Would segmenting by role, experience, work type, manager, autonomy, or opportunity exposure lead to different support decisions?
- Which apparently different groups face the same underlying barrier and could be supported together?
- Where do broad categories such as "beginner", "skeptic", or "frontline worker" hide important causal differences?
- Are we segmenting people because the distinction is useful or simply because the data are easy to collect?
- Can we support the resulting groups meaningfully without creating unnecessary complexity?
- Which segmentation gives us the clearest link between a user's situation and what we should do differently?
Understanding Differences in GenAI Experience
- What kinds of GenAI experience do different users actually have rather than simply how long they have had access?
- Who has only tried simple one-shot interactions, and who can already work iteratively across more complex tasks?
- Which users have experience with files, data, search, multimodality, tools, or multi-step work?
- Where does self-reported confidence exceed demonstrated ability to evaluate and verify outputs?
- Which users possess transferable capability rather than familiarity with one product or prompt?
- What useful practices or failure knowledge do experienced users already possess?
- How should support differ between people who need initial orientation and those who need opportunities to deepen existing capability?
Understanding Differences in Motivation to Use GenAI
- What outcome would make GenAI worth changing established behavior for each group?
- Which users already see clear value in GenAI, and which see little relevance to their actual work?
- Where is curiosity high but the expected benefit too small to sustain recurring use?
- Which groups face enough learning effort, risk, or workflow friction that non-use is currently rational?
- Are people motivated by saving time, improving quality, reducing frustration, learning, expanding capability, or something else?
- What organizational signals or incentives are strengthening or weakening that motivation?
- What should change in the environment rather than trying to persuade a group whose current reasons for non-use are sound?
Identifying Different Role-Specific Needs
- What recurring situations in this role are genuinely different enough to require role-specific GenAI support?
- Which information, systems, and context does the role need in order to use GenAI effectively?
- What kinds of output does the role produce, and what quality standard must those outputs meet?
- Where does the role rely on specialist judgment or authority that should remain human?
- Which GenAI capabilities would materially improve the role's work rather than simply being technically available?
- Which needs are common enough to address with organization-wide methods, and which depend on local professional practice?
- What support would make GenAI relevant to this role without prescribing use where it adds little value?
Understanding Which Groups Need the Most Initial Support
- Which groups currently combine low experience with enough relevant GenAI opportunity that additional support could make a meaningful difference?
- Who has access but lacks the confidence or capability to begin useful experimentation?
- Which groups face unclear usage boundaries that make otherwise suitable experimentation difficult?
- Where are poor early experiences likely to become lasting negative assumptions about GenAI?
- Which users lack managers, peers, or local champions who can help with ordinary questions?
- What other structural barriers would remain even if these users received excellent training?
- Which groups should receive additional support first because the likely benefit justifies the effort?
Identifying Groups That Are Ready to Experiment More Deeply
- Which users have enough practical GenAI capability to move beyond basic assistance?
- What evidence shows that they can evaluate outputs critically rather than merely use GenAI frequently?
- Which groups encounter work where deeper use, tool integration, or multi-step execution could create meaningful value?
- Do they have managers, time, access, and usage boundaries that support more ambitious experimentation?
- Can they identify and communicate failures as readily as successful results?
- What organizational uncertainty could this group's experimentation help resolve?
- Which advanced experiment would generate learning that is useful beyond the group conducting it?
Understanding Differences in Concerns and Risk Perception
- What specific concern is driving caution or non-use in each group?
- Which concerns arise from genuine differences in data, professional responsibility, customer impact, or regulatory exposure?
- Where are users reacting to poor prior experiences rather than hypothetical risk?
- Which concerns can be resolved through clearer information or policy, and which require an organizational decision?
- Where does perceived risk exceed the actual boundary, and where might users be underestimating real risk?
- How do job-security, monitoring, quality, workload, deskilling, and professional-identity concerns differ across groups?
- What response would address the actual concern without dismissing legitimate professional judgment as resistance?
Identifying Managers With Different Levels of Adoption Readiness
- Which managers have enough firsthand GenAI experience to discuss its use credibly with their teams?
- Who can recognize relevant opportunities without trying to force GenAI into every activity?
- Which managers understand organizational boundaries well enough to support normal experimentation and escalate exceptional cases?
- Who is supportive in principle but lacks the confidence or practical knowledge to coach employees?
- Where are workload expectations, risk attitudes, or local incentives unintentionally suppressing adoption?
- Which managers already create conditions that produce useful and independent GenAI use?
- What different support would help each manager lead adoption without expecting them to become technical specialists?
Understanding Differences Between Occasional and Regular Users
- What recurring situations cause regular users to return to GenAI without external prompting?
- Why do occasional users stop after a small number of experiments?
- Are regular users encountering more suitable work, or are they simply better at recognizing opportunities in similar work?
- How do the two groups differ in their ability to provide context, iterate, evaluate, and verify outputs?
- What role do tool integration, manager support, learning time, and ease of access play in the difference?
- Is occasional use actually appropriate for some roles because suitable opportunities arise infrequently?
- What would need to change for occasional users to become more regular where greater use would genuinely be valuable?
Understanding How Adoption Support Needs Differ Across User Groups
- What specific adoption problem is each user group experiencing now?
- Which groups need basic orientation, and which need real-work practice, coaching, specialist help, or greater autonomy?
- Who primarily needs more relevant opportunities rather than more GenAI instruction?
- Where is the main constraint access, policy clarity, manager support, workflow fit, confidence, or capability?
- Which forms of support can be shared across groups without becoming too generic?
- Where would role-specific or group-specific support materially improve the result?
- How can we vary support according to actual need while avoiding permanent dependence on increasingly elaborate user categories?
Identifying High-Value Organizational GenAI Opportunities
Collecting Potential GenAI Opportunities Across the Organization
- Where are employees already using or considering GenAI to solve real work problems?
- Which recurring frustrations, bottlenecks, or unmet needs are generating potential GenAI opportunities?
- What valuable ideas are emerging informally in teams that central functions may not currently see?
- Which opportunities are being suggested because GenAI genuinely changes the work rather than because the technology is available?
- Where do similar opportunity patterns appear across otherwise different roles or functions?
- How can we capture enough context about each opportunity to understand the work, expected value, and important constraints?
- What lightweight process would keep opportunity discovery open without turning every idea into a formal project?
Identifying Repetitive Work That GenAI Could Support
- Which recurring activities consume substantial cognitive effort despite following similar patterns each time?
- Where does variation between cases make conventional automation difficult but still leave enough structure for GenAI to help?
- Which parts of the repetitive work could GenAI prepare, transform, classify, compare, or review?
- How much human judgment would still be needed after GenAI performed the repetitive portion?
- Would verification and correction remain low enough for the approach to save meaningful effort?
- Which repetitive activities are too low-value, infrequent, or easy already to justify adoption effort?
- Where would reducing repetitive work release capacity that the organization could actually use?
- Where do employees spend substantial time searching for, reading, combining, or interpreting information?
- Which activities require people to reconstruct context from several documents, systems, or sources?
- Where does the volume of available information make important details difficult to find or compare?
- Which work could improve if GenAI helped retrieve, summarize, structure, or synthesize relevant information?
- What organizational information would GenAI need access to for the opportunity to work in practice?
- How difficult would it be to verify that the information used and conclusions produced are accurate enough?
- Which information-intensive activities offer enough value to justify solving the required access and integration problems?
Identifying Decisions That Could Benefit From Better Preparation
- Which consequential decisions currently require substantial research, analysis, or preparation?
- Where do decision-makers receive incomplete, inconsistent, or poorly structured information?
- Which decisions would benefit from examining more alternatives, assumptions, risks, or tradeoffs?
- Where could GenAI help organize evidence or challenge reasoning without becoming the decision-maker?
- What domain knowledge or authoritative information would be required to evaluate GenAI-supported preparation?
- Which decisions have consequences high enough that verification costs could outweigh the benefit?
- Where would better decision preparation materially improve the quality or speed of organizational decisions?
Finding Opportunities to Improve Communication and Coordination
- Which recurring communication activities consume substantial time because information must repeatedly be rewritten, adapted, or clarified?
- Where do handoffs fail because people receive incomplete context or information in the wrong form?
- Which coordination problems could GenAI reduce by preparing, translating, consolidating, or structuring information?
- Where could GenAI help different audiences understand the same underlying material without creating inconsistent messages?
- Which communication problems are symptoms of a broken process that GenAI should not merely make faster?
- What confidentiality, accuracy, or accountability requirements would apply to the communication?
- Which communication or coordination opportunities would improve real workflow performance rather than simply generate more content?
Identifying Opportunities That Could Improve Quality
- Where does the quality of important work vary substantially between people, teams, or cases?
- Which outputs commonly contain omissions, inconsistencies, weak structure, or avoidable errors?
- Where could GenAI provide a useful second pass, comparison, critique, or completeness check?
- Which quality dimensions can GenAI assess reliably enough to support a human reviewer?
- Where could fluent but incorrect GenAI output create false confidence instead of better quality?
- How much additional review would be required before the GenAI-supported result could be trusted?
- Which quality opportunities are valuable enough that improved consistency or completeness would justify the added process?
Identifying Work That Becomes Practical Only With GenAI
- What valuable work is currently omitted, restricted, or performed only occasionally because it requires too much cognitive effort?
- Which analysis, personalization, preparation, research, or review could become economical if GenAI reduced the cost substantially?
- What could we now do for many people or cases that was previously feasible only for a small number?
- Which new activity would create a meaningful outcome rather than simply increasing the volume of information or content?
- What human expertise, organizational context, or verification would still be required to make the newly possible work useful?
- Could the opportunity create a genuinely new capability rather than merely improve an existing process?
- Which newly practical activities are important enough to deserve experimentation despite having no established baseline process?
Comparing Opportunities by Potential Value and Feasibility
- What specific value could each opportunity create through time, quality, capacity, decisions, customer outcomes, or newly enabled work?
- How capable is current GenAI of performing the required work reliably under realistic conditions?
- What data, organizational context, integrations, permissions, and human expertise would each opportunity require?
- How costly would verification, correction, exception handling, and implementation be?
- How repeatable and scalable is the opportunity beyond one expert user or favorable case?
- What risks or consequences would matter if the GenAI-supported work failed?
- Which opportunities combine enough value, feasibility, and learning potential to justify further investment?
Identifying Opportunities That Could Demonstrate Value Early
- Which opportunities can be tested on real work without requiring major infrastructure or irreversible change?
- Where can improvement be measured clearly against the existing way of working?
- Which opportunities matter enough that success would be credible rather than merely impressive?
- Can users judge the quality of the result well enough to learn from both success and failure?
- Which early opportunities would also build transferable capability, evidence, or infrastructure for later adoption?
- Which apparently easy opportunities could create an unrealistically favorable picture of broader GenAI adoption?
- What early test would demonstrate meaningful value while remaining honest about the limits of what it proves?
Creating an Organization-Wide View of GenAI Opportunity Areas
- Which major opportunity patterns recur across functions even when the specific tasks differ?
- Where do multiple opportunities depend on the same data, integration, capability, or organizational change?
- Which opportunities are broadly transferable, and which are genuinely specific to one professional context?
- Where is the organization concentrating too heavily on obvious efficiency uses while overlooking quality, capability, or newly enabled work?
- Which opportunity areas connect most directly to important organizational problems or strategic priorities?
- Where are technically attractive opportunities unlikely to create enough value to matter?
- What portfolio of opportunity areas gives leadership a realistic view of where GenAI could change the organization?
Anticipating Adoption Barriers
Identifying Where Access Could Prevent Adoption
- Which employees may lack access to a GenAI tool capable enough for the work they are expected to perform?
- Where could authentication, devices, permissions, usage limits, or procurement processes make nominal access impractical?
- Which important work requires organizational information or system integrations that current GenAI tools cannot reach?
- Where might approved tools be materially weaker or less usable than the alternatives employees can access privately?
- Which groups could face unequal adoption opportunities because their technical environment differs?
- What access problems are likely to appear only after employees move beyond simple experimentation?
- Which access barriers must be removed before low adoption could reasonably be treated as a user-behavior problem?
Anticipating Uncertainty About What Is Allowed
- Which normal work situations are likely to leave employees unsure whether GenAI use is permitted?
- What categories of information will users find difficult to classify confidently?
- Where could people misunderstand the difference between approved tools, external tools, and personal accounts?
- Which activities become ambiguous because the output could influence a consequential decision or external action?
- Where might employees avoid useful work because obtaining clarification is harder than simply not using GenAI?
- What questions should managers be able to answer without referring every case to a specialist?
- Which practical boundaries need to be clear before employees can experiment independently and safely?
Identifying Where People May Lack Relevant GenAI Skills
- Which users may struggle to describe their work, objective, constraints, and context clearly enough for GenAI?
- Who is likely to accept a weak first answer rather than refine, challenge, or redirect the interaction?
- Where could users lack the domain knowledge needed to judge whether an output is actually good?
- Which groups may struggle to distinguish fluent output from reliable evidence?
- Who needs help recognizing when verification, another tool, or human expertise is required?
- Which capabilities can develop naturally through repeated real-work use, and which need more deliberate support?
- What skill gaps would materially prevent independent professional use rather than merely make users less sophisticated?
Anticipating Difficulty Seeing Personal Relevance
- Which employees may understand that GenAI matters organizationally but still see no useful connection to their own work?
- Where do current examples fail because they are based on job labels rather than recurring responsibilities and problems?
- Which groups may have a narrow mental model of GenAI as mainly a writing or summarization tool?
- What recurring work situations could reveal more relevant forms of assistance?
- Where might employees be correct that current GenAI offers little value for their work?
- What real examples from similar roles would make useful possibilities easier to recognize?
- Which groups need opportunity discovery more than additional instruction about how GenAI works?
Identifying Trust and Quality Concerns
- Which work situations are likely to make employees appropriately cautious about relying on GenAI output?
- Where could previous errors or weak experiences create broader distrust than the evidence justifies?
- Which users may instead trust fluent GenAI output more than its reliability warrants?
- What kinds of errors would be especially difficult to detect in the work concerned?
- Which verification methods could make useful GenAI work acceptable without recreating the entire task manually?
- What concerns can only be resolved through repeated experience rather than reassurance or communication?
- Where is the expected verification burden high enough that distrust may reflect poor task fit rather than an adoption problem?
Understanding Where People May Lack Time to Experiment
- Which employees have so little discretionary capacity that experimentation would simply become additional work?
- Where are managers likely to prioritize immediate output over the temporary learning cost of trying new methods?
- Which real tasks could provide useful experimentation without requiring separate training exercises?
- How much time do users actually need before they can judge whether a GenAI practice is worthwhile?
- Where would one protected period of experimentation be enough, and where is repeated practice necessary?
- What existing activity could be reduced or deprioritized to create legitimate learning time?
- Which time constraints require a management or leadership decision rather than encouragement for employees to experiment more?
Anticipating Weak Managerial Support
- Which managers may lack enough firsthand GenAI experience to discuss adoption credibly with their teams?
- Where could managers discourage experimentation because they are uncertain about policy, quality, or workforce consequences?
- Which workload or performance expectations leave employees little practical permission to learn?
- Where might managers interpret weak early results as evidence that GenAI has no value?
- Which managers may push usage too aggressively and suppress honest reporting of failures or poor task fit?
- What support would help managers distinguish capability, relevance, risk, access, and motivation problems?
- Where is managerial behavior likely to become a stronger adoption constraint than employee willingness?
Identifying Workflow Friction That Could Discourage Use
- Where does using GenAI require employees to leave their normal workflow, copy information manually, or recreate context?
- Which tasks require information the approved tool cannot access conveniently?
- Where could authentication, file handling, tool switching, or latency make the old method easier?
- Which review or approval requirements could consume most of the time GenAI initially saves?
- Where would GenAI speed one step while leaving the overall workflow unchanged or bottlenecked elsewhere?
- Which frictions can be removed through integration, and which reveal that the use case itself is weak?
- What workflow changes would need to accompany GenAI before recurring use becomes practical?
Anticipating Where Early Failure Could Reduce Confidence
- Which early GenAI tasks are difficult enough that beginners are likely to get weak or misleading results?
- Where could users mistake a poorly chosen first task for evidence that GenAI is generally useless?
- Which initial successes might instead create unjustified confidence because the task is unusually easy?
- What kinds of errors would be especially memorable or damaging to user trust?
- Which users are least likely to diagnose why a weak result occurred or try a different approach?
- What support would help employees distinguish task mismatch, weak interaction, model limitations, and missing context?
- How can early experimentation expose both useful capability and realistic limitations without making failure unnecessarily costly?
Assessing Which Barriers Could Have the Greatest Effect on Adoption
- Which barriers affect the largest amount of potentially valuable work rather than simply the largest number of users?
- Which problems completely prevent adoption, and which merely make it slower or less convenient?
- Which barriers are likely to diminish through experience, and which require organizational intervention?
- Where does one barrier create or reinforce several others?
- Which constraints differ sharply between user groups or types of work?
- What evidence suggests that removing a particular barrier would actually change behavior?
- Which barriers represent the highest-leverage targets for the adoption approach?
Defining the Intended Adoption Outcome
Clarifying Which Adoption Problem Needs to Be Solved
- What specific condition in the organization's current GenAI adoption is unsatisfactory?
- Is the main gap access, opportunity recognition, capability, confidence, recurrence, integration, or value?
- What evidence shows that the observed gap is a real problem rather than an unrealistic adoption expectation?
- Which people, teams, workflows, or outcomes are materially affected?
- What underlying causes might be hiding behind the visible symptom of low usage?
- What would remain unchanged if the organization simply increased training or communication?
- What precise adoption problem should the effort solve before we decide how to intervene?
Defining Which People the Adoption Effort Should Reach
- Which people need to change their behavior or capability for the intended outcome to occur?
- Which roles have enough relevant GenAI opportunity to justify being included?
- What groups require access, manager support, or specialist involvement even if they are not the primary users?
- Does meaningful success require broad participation, or deeper adoption in a smaller set of relevant roles?
- Which employees could reasonably remain outside the current scope because GenAI offers little value to their work?
- What differences between groups mean they should not all be expected to progress at the same pace?
- What population is broad enough to achieve the outcome without turning universal participation into the goal itself?
Defining How GenAI Use Should Change Everyday Work
- In which recurring work situations should employees begin considering GenAI naturally?
- What should people be able to do better, faster, more thoroughly, or more often because GenAI is available?
- Which activities should shift from occasional experimentation into normal recurring practice?
- Where should GenAI remain optional because several effective working methods are acceptable?
- What responsibilities for context, judgment, verification, or decisions should remain clearly human?
- Which existing habits or workflow steps would need to change for GenAI use to become genuinely integrated?
- What would an ordinary workday look like if the desired adoption outcome were achieved?
Describing What Successful Adoption Should Look Like
- What observable behaviors would tell us that GenAI adoption is genuinely working?
- How would successful users recognize opportunities, interact with GenAI, and evaluate the results?
- What would managers observe changing in normal work rather than in usage dashboards?
- Which recurring tasks or workflows would look different?
- How would employees describe the usefulness and limitations of GenAI if confidence were well calibrated?
- What signs would distinguish meaningful adoption from high interaction volume or compliance with an adoption campaign?
- What concrete description of success is specific enough to guide action without prescribing one universal way of using GenAI?
Identifying the Business Outcomes Adoption Should Support
- Which organizational problem or priority should better GenAI adoption ultimately help improve?
- Is the intended value mainly better productivity, quality, capacity, customer outcomes, decisions, innovation, or newly enabled work?
- What operational changes would need to occur before the expected business outcome could plausibly follow?
- How much influence can adoption realistically have on that outcome compared with other factors?
- What business result would make the required investment and disruption worthwhile?
- Which attractive benefits should remain secondary because the organization cannot credibly measure or capture them?
- What business outcome should remain the anchor when adoption activities compete for attention?
Defining the Desired Level of Independent GenAI Use
- Which GenAI decisions should ordinary users eventually be able to make without specialist help?
- How independently should employees recognize relevant opportunities in their own work?
- What should users be able to do when the first response is weak or incomplete?
- How capable should they be of evaluating and verifying outputs before using them?
- Where should templates, examples, or shared methods remain useful without becoming permanent dependencies?
- Which situations should still require a manager, domain expert, or specialist even among capable users?
- What level of independence is appropriate for each kind of work rather than assuming the same target for everyone?
Clarifying What Adoption Is Not Intended to Achieve
- Which organizational problems fall outside the purpose of this adoption effort?
- What technology, governance, data, or workflow problems cannot be solved through better user adoption?
- Which work should not be pushed toward GenAI simply to raise usage?
- What workforce or productivity outcomes would be irresponsible to promise before capability and value are demonstrated?
- Which forms of automation or transformation require separate decisions beyond ordinary GenAI adoption?
- Where would expanding the scope dilute resources or confuse the intended outcome?
- What should leadership explicitly exclude so employees and managers do not mistake adoption for a universal AI transformation program?
Defining the Desired Breadth and Depth of GenAI Adoption
- How many people or roles need meaningful GenAI use for the intended outcome to be achieved?
- Which groups need broad awareness without necessarily becoming frequent users?
- Where should GenAI become deeply embedded in recurring work rather than remain an occasional option?
- How much variation in adoption is appropriate because different roles contain different opportunity levels?
- Should success be judged by frequency, recurrence, independence, integration, or another dimension for each group?
- What depth of use would justify the additional support and integration required?
- What combination of breadth and depth would show that adoption has moved beyond isolated enthusiasts without making uniform usage the objective?
Defining Evidence That Would Indicate Meaningful Progress
- What evidence would show that usable access has improved rather than merely license coverage?
- What would demonstrate movement from experimentation toward recurring and independent use?
- How could we tell whether employees are recognizing more relevant work situations themselves?
- What evidence would show that GenAI is becoming integrated into normal work rather than simply used more often?
- Which qualitative evidence from users, managers, work outputs, or workflows should complement telemetry?
- What operational or business indicators would suggest that adoption is creating value?
- Which few progress signals would help us make better decisions without turning them into superficial usage targets?
Creating a Clear Adoption Outcome for the Organization
- Can the intended outcome be stated as a changed condition in work rather than a list of adoption activities?
- Does it make clear whose behavior or capability should change?
- Does it describe how GenAI should become useful in actual recurring work?
- Does it connect adoption with the organizational value it is intended to enable?
- Does it preserve legitimate variation in how different groups use GenAI?
- Does it avoid promising technological or workforce outcomes that remain uncertain?
- What concise outcome statement would be clear enough to guide leadership, managers, support functions, and employees in the same direction?
Setting Adoption Priorities
Identifying Which User Groups Should Receive Attention First
- Which groups have enough relevant GenAI opportunity that improved adoption could create substantial value?
- Where is the current adoption gap large for reasons the organization can realistically address?
- Which groups are blocked by dependencies that should be solved before additional support is provided?
- Where could early progress produce transferable learning for other users?
- Which groups need attention because they are being disadvantaged by unequal access or support rather than because they are easiest to help?
- How much would focusing on enthusiastic early adopters distort our understanding of later users?
- Which groups should receive attention first when value, readiness, need, and learning potential are considered together?
Selecting Which Teams or Functions to Focus On
- Which teams perform enough GenAI-suitable work to justify concentrated adoption effort?
- Where are important business problems already creating strong demand for a better way of working?
- Which teams have the access, management conditions, and capacity required to learn productively?
- Where could a successful practice plausibly transfer to other teams or functions?
- Which functions contain important opportunities but require foundational work before experimentation will be useful?
- What would we miss if we selected only the teams already enthusiastic about GenAI?
- Which initial portfolio of teams gives us both meaningful value opportunities and useful organizational learning?
Prioritizing the Most Valuable GenAI Opportunities
- Which opportunities could materially improve important work rather than simply make an isolated task more convenient?
- How large is the potential value once verification, implementation, support, and downstream effects are included?
- Which opportunities are feasible with the tools, information, and capabilities available now?
- What important dependencies would need to be solved before an otherwise attractive opportunity becomes practical?
- Which opportunities could create strategic learning or reusable capability even if immediate value is uncertain?
- Where are easy opportunities distracting us from harder but substantially more important work?
- Which opportunities deserve near-term action when value, feasibility, risk, and future leverage are considered together?
Deciding Which Adoption Barriers Need Early Attention
- Which barriers would prevent otherwise valuable opportunities from being attempted at all?
- What constraints affect several user groups, teams, or opportunities at the same time?
- Which problems can users solve through experience, and which require organizational decisions or infrastructure?
- What barrier should be addressed first because other adoption activities depend on it?
- Which apparently urgent problems have limited effect on actual adoption and can wait?
- Where would resolving one bottleneck unlock several downstream activities?
- What evidence supports the order in which we intend to address the major barriers?
Balancing Quick Wins With Longer-Term Capability Building
- Which early opportunities can create credible evidence of value without being artificially easy?
- What capabilities, integrations, management practices, or learning infrastructure will matter long after the first successes?
- Could concentrating on quick wins teach employees that GenAI is useful only for simple tasks?
- Which near-term experiments also build transferable skills, reusable knowledge, or organizational confidence?
- What longer-term investments need to begin before their benefits become immediately visible?
- How much resource should remain available for uncertain but strategically important learning?
- What portfolio balances visible progress with the foundations required for deeper adoption?
Prioritizing Areas Where Leadership Support Is Most Needed
- Which adoption constraints require authority, resources, or decisions that employees and managers cannot provide themselves?
- Where does progress depend on leaders clarifying purpose, priorities, ownership, or risk tolerance?
- Which teams need leadership to create legitimate time or capacity for experimentation?
- Where are conflicting organizational signals undermining local adoption?
- What cross-functional barriers need senior sponsorship because no single function can resolve them?
- Where would additional executive visibility add noise or pressure rather than useful support?
- Which leadership interventions would remove the greatest constraint on meaningful adoption?
Deciding Where Limited Support Resources Should Be Used
- Which adoption problems genuinely require scarce specialist expertise rather than self-service, manager, or peer support?
- Where would a small amount of expert help unlock a high-value opportunity or remove a shared barrier?
- Which requests could create reusable knowledge, infrastructure, or methods for several teams?
- What support is currently being consumed by recurring questions that should become easier to resolve locally?
- Which high-risk or consequential uses warrant specialist attention regardless of adoption scale?
- How ready is the requesting team to own the work after central support ends?
- What allocation of limited support creates the greatest organizational leverage rather than rewarding whoever asks first?
Sequencing Adoption Efforts Across the Organization
- What prerequisites must exist before each major adoption activity can succeed?
- Which work can progress independently in parallel rather than waiting for a single organization-wide sequence?
- Where must usable access and boundary clarity precede meaningful experimentation?
- When should real-work experience precede deeper capability building or shared standardization?
- At what point does enough local evidence exist to justify spreading practices or redesigning workflows?
- Which scaling activities depend on data, integrations, manager capability, or measurement that are not yet ready?
- What sequence follows the actual dependencies of our adoption system rather than a generic maturity model?
Identifying Lower-Priority Activities That Can Wait
- Which proposed activities are useful in principle but do not address a current constraint?
- What sophisticated support, measurement, or governance mechanism would be premature at our present stage?
- Which initiatives depend on user capability or workflow maturity that has not yet developed?
- Where are we designing infrastructure for hypothetical scale before proving enough value?
- What activities would consume scarce attention without changing the next important adoption outcome?
- Which ideas can be deliberately parked until specific evidence or dependencies change?
- What should we exclude now so the adoption effort remains focused on the few conditions that matter most?
Defining the First Adoption Priorities to Act On
- Which few priorities address the strongest combination of value, dependency, and adoption constraint?
- What concrete condition should be different when each priority has been addressed?
- Why does each priority need action now rather than later?
- Who has the authority and capability to own the required action?
- What other work must happen first or in parallel for the priority to succeed?
- What evidence should we expect to see if the priority is producing the intended effect?
- How should the initial priorities be communicated so people understand both what matters now and what is deliberately not being pursued yet?
Developing the Adoption Approach
Choosing How Adoption Should Begin
- What starting approach best fits our current access, user capability, opportunity landscape, and organizational readiness?
- Should the first phase concentrate on selected teams, selected opportunities, or broad low-risk experimentation?
- What real work should anchor the initial experience so users learn something transferable rather than merely see an impressive demonstration?
- How much common structure do beginners need before local experimentation becomes more valuable than central direction?
- What would we learn by starting narrowly that we could not learn from an organization-wide launch?
- What risks would a narrow start create by overrepresenting enthusiasts or unusually favorable teams?
- What beginning gives us enough practical learning to improve the approach before making larger commitments?
Deciding How Much Direction Should Come From the Center
- Which adoption decisions create enough shared risk, dependency, or efficiency benefit to require central ownership?
- What should the organization decide once about platforms, boundaries, infrastructure, measurement, or major investment?
- Which parts of adoption depend too heavily on local work context to be prescribed centrally?
- Where would central consistency remove confusion, duplication, or incompatible approaches?
- Where would central approval create unnecessary delay in ordinary low-risk experimentation?
- How will teams know which decisions they are free to make and which need escalation?
- What balance of common direction and bounded local agency best supports useful adoption?
Deciding How Much Freedom Teams Should Have to Experiment
- Which low-risk GenAI experiments should teams be able to conduct without additional approval?
- What enterprise boundaries must remain common regardless of local experimentation?
- How much freedom do teams need to discover opportunities and working methods central functions could not design for them?
- Which experiments create cross-team, customer, data, or system consequences that justify broader review?
- What should teams record or share so experimentation contributes to organizational learning?
- When does a local experiment become consequential enough to require stronger ownership, validation, or governance?
- How can the organization maximize local learning without turning every team into an isolated GenAI environment?
Designing How Learning Should Happen Through Real Work
- Which genuine work situations provide enough relevance, safety, and verifiability for people to learn productively?
- How can users learn by solving real problems instead of practicing disconnected prompt exercises?
- What kinds of early work give users enough success to continue while still exposing realistic GenAI limitations?
- How should employees reflect on what the result reveals about task fit, interaction, verification, and value?
- What role should managers play in creating time and helping users interpret weak or surprising results?
- How can learning from real work be captured without immediately turning individual techniques into standards?
- What progression would move users from supported experimentation toward independent professional judgment?
Determining How Support Should Be Provided
- What kinds of difficulty should users be expected to resolve independently at each stage of adoption?
- Which common questions can be handled through clear self-service guidance or reusable examples?
- When is peer support the best source of situated practical help?
- Which problems require coaching because the difficulty lies in the user's work context rather than a simple rule?
- What questions genuinely require technical, legal, privacy, security, or other specialist expertise?
- How can recurring specialist questions be converted into better tools, guidance, or local capability over time?
- What support model helps users become more independent rather than creating permanent reliance on central experts?
Defining How Communication Should Support Adoption
- What does each audience need to understand about why the organization is pursuing GenAI now?
- Which parts of the message are established facts, organizational decisions, current expectations, or genuine uncertainties?
- How will broad organizational direction be translated into implications for people's actual work?
- What legitimate concerns about roles, workload, monitoring, quality, or job security need direct acknowledgment?
- Which questions require policy or leadership decisions rather than communication designed to reassure employees?
- How should managers adapt the organization-wide message without inventing answers they do not have?
- What communication rhythm keeps employees informed as adoption changes without creating hype around every development?
Deciding How Managers Should Participate
- What local adoption conditions can line managers influence more effectively than central functions?
- How much practical GenAI experience do managers need to recognize opportunities and discuss limitations credibly?
- What should managers do to create legitimate experimentation time and psychological safety?
- How should they help employees diagnose weak results without becoming technical experts?
- What expectations should managers set around quality, verification, workload, and useful rather than compulsory GenAI use?
- Which governance, technical, or workforce decisions should managers escalate rather than improvise locally?
- What manager role is substantial enough to shape adoption while remaining realistic alongside their existing responsibilities?
Connecting the Adoption Approach to Existing Organizational Structures
- Which existing onboarding, learning, management, planning, communication, or improvement routines can carry parts of GenAI adoption?
- Where can GenAI learning and opportunity discovery fit naturally into work employees already perform?
- Which existing communities, support structures, or knowledge systems could be extended instead of duplicated?
- What current processes would slow adoption or distort it if we simply inserted GenAI into them?
- Where is a temporary dedicated GenAI structure necessary because existing mechanisms lack the required capability?
- Which new adoption structures should eventually disappear as normal organizational functions absorb their responsibilities?
- How can we use existing organizational machinery without institutionalizing immature GenAI practices too early?
Sequencing the Main Elements of the Adoption Approach
- Which elements are genuine prerequisites for others rather than merely customary first steps?
- What access and boundary clarity must exist before users can experiment safely?
- When should communication, real-work learning, and manager support begin relative to tool rollout?
- At what point should peer sharing and team-level methods emerge from individual experimentation?
- When is adoption mature enough for formal measurement of integration and value to become useful?
- What evidence should exist before successful local patterns are deliberately scaled or workflows redesigned?
- What sequence allows the approach to evolve through evidence rather than following a fixed rollout plan?
Preparing the Adoption Approach to Evolve With Experience
- Which assumptions about users, barriers, opportunities, tools, and value are we currently making?
- What evidence would tell us that one of those assumptions is wrong?
- Which parts of the adoption approach should be treated as experiments rather than permanent organizational mechanisms?
- How will employee, manager, and team experience reach the people able to change the approach?
- What should cause us to stop, simplify, expand, or replace an adoption activity?
- Which elements should remain stable because they express purpose, decision rights, or core safety principles?
- What review mechanism will keep the adoption approach responsive without creating continuous organizational churn?
Building Leadership Alignment
Understanding How Different Leaders Currently View GenAI
- How does each leader currently understand what GenAI can and cannot reliably do?
- Which leaders see GenAI mainly as a productivity tool, and who sees implications for capability, workflows, customers, or strategy?
- Where are leadership views based on firsthand experience rather than demonstrations, headlines, or assumptions?
- Which leaders may be overestimating current GenAI capability, and which may still be working from outdated limitations?
- How do different business responsibilities explain legitimate differences in leadership perspective?
- Which disagreements reflect different facts or assumptions rather than different organizational priorities?
- What leadership perspectives need to be understood before meaningful alignment is possible?
Identifying Differences in Leadership Expectations
- What does each leader expect GenAI adoption to achieve for the organization?
- How quickly does each leader expect meaningful changes in use, capability, workflows, or business outcomes?
- Where do leaders differ in how much uncertainty, experimentation, or failure they are willing to accept?
- Which leaders expect organization-wide consistency, and which expect teams to discover their own approaches?
- What assumptions are leaders making about employee willingness, capability, and available time to adopt GenAI?
- Where are expected productivity, cost, or workforce effects running ahead of demonstrated capability?
- Which expectation differences would produce contradictory decisions if they remain unresolved?
Building a Shared Understanding of Why Adoption Matters
- What concrete changes in GenAI capability or the external environment make adoption relevant to this organization now?
- Which organizational problems or opportunities provide the strongest shared reason to develop GenAI capability?
- What evidence supports the case for action without requiring leaders to accept exaggerated claims about the future?
- Which expected benefits are sufficiently established, and which remain hypotheses that need testing?
- What is the cost of learning too slowly, and what is the risk of moving faster than the evidence supports?
- Which legitimate leadership concerns need to remain visible rather than being suppressed in the name of alignment?
- What shared rationale is strong enough to guide action even while some future consequences remain uncertain?
Aligning Leaders on the Intended Adoption Outcome
- What changed condition in everyday work do leaders expect successful GenAI adoption to create?
- Which people or roles need meaningful adoption for that outcome to be achieved?
- How much breadth, recurrence, independence, and work integration does leadership actually expect?
- What business outcomes should improved adoption enable rather than merely accompany?
- Where should GenAI remain optional because useful adoption does not require uniform use?
- Which technology, workforce, or transformation outcomes are explicitly outside the current adoption objective?
- What common adoption outcome can leaders consistently use when making priorities and resource decisions?
Aligning Leaders on Adoption Priorities
- Which adoption priorities currently differ between leaders, and what reasoning sits behind those differences?
- Are leaders prioritizing the same combination of user groups, barriers, opportunities, and organizational capabilities?
- Which priorities address important dependencies rather than simply offering the fastest visible progress?
- How should near-term value opportunities be balanced against capability, integration, and learning needed for deeper adoption?
- Which barriers require leadership attention because teams or employees cannot remove them themselves?
- What attractive activities should deliberately wait because they do not address the current adoption constraint?
- Which priority set can leaders support consistently when resources or local demands create pressure to deviate?
Resolving Differences About Speed and Risk
- Where do leaders currently differ about how quickly GenAI adoption should move?
- Which concrete risks are driving calls for caution, and how significant are they in the intended uses?
- Which opportunities, learning effects, or competitive changes are driving calls to move faster?
- What can be tested through bounded experimentation instead of being resolved through abstract debate?
- Where would moving slowly create its own cost through delayed learning or capability development?
- Which activities justify different speeds because their consequence, uncertainty, or readiness differs?
- What pace allows the organization to learn quickly while keeping consequential commitments tied to evidence?
Clarifying Leadership Roles in the Adoption Effort
- Which GenAI adoption decisions genuinely require senior leadership authority?
- What should functional leaders own because it depends on their work, priorities, and professional context?
- What should line managers own because they control local experimentation, workload, expectations, and team practices?
- Which responsibilities belong instead to technology, governance, risk, learning, or other specialist functions?
- Where are several leaders currently assuming that someone else owns an important decision?
- Which overlapping responsibilities need clearer decision rights rather than more coordination meetings?
- How should leadership ownership be made clear enough that teams know where decisions and escalation belong?
Aligning Leaders on Resources and Support
- What access, integration, learning, specialist support, management time, and technical capability does the intended adoption outcome require?
- Which resources already exist but are fragmented, inaccessible, or poorly coordinated?
- Where would additional resources remove a genuine adoption constraint rather than simply add activity?
- What support are leaders implicitly expecting other functions or enthusiastic employees to provide?
- Which resource requests are premature because the underlying opportunity or need has not yet been demonstrated?
- How should scarce specialist capacity be allocated when several valuable adoption needs compete for attention?
- What explicit resource commitments must leaders make for the stated adoption priorities to remain credible?
Agreeing on Consistent Messages for the Organization
- What core facts and organizational decisions should employees hear consistently from every leader?
- How should leaders distinguish what is known, what is expected, what has been decided, and what remains uncertain?
- What should leaders say about why GenAI matters without implying that every role or task must change?
- How should workforce, role, and productivity questions be discussed when future consequences are not yet fully known?
- Which messages about experimentation, quality, accountability, and permitted use need to remain consistent across functions?
- What parts of the message can be adapted to local work without changing the underlying meaning?
- What common language would reduce confusion while still allowing leaders to speak credibly in their own context?
Resolving Remaining Leadership Disagreements Before Moving Forward
- Which unresolved leadership disagreements would materially change what employees, managers, or support functions are expected to do?
- Which disagreements concern facts that can be investigated rather than preferences that need an explicit decision?
- What evidence or small-scale experiment could resolve an important uncertainty?
- Which differences can legitimately remain local without weakening organization-wide coherence?
- What issues require a clear executive decision because contradictory approaches would create risk or confusion?
- Where would delaying action until complete agreement create more cost than proceeding within agreed boundaries?
- What minimum set of decisions must be settled before the next phase of adoption can operate coherently?
Demonstrating Leadership Commitment
Showing Visible Leadership Interest in GenAI Adoption
- How can leaders show through normal organizational behavior that useful GenAI adoption genuinely matters?
- Which leadership questions or decisions would employees recognize as evidence of sustained attention?
- Where should GenAI opportunities and adoption barriers appear in ordinary business discussions rather than special events?
- How can leaders show interest in learning and value rather than simply asking whether usage is increasing?
- What behavior would make leadership attention appear promotional, performative, or disconnected from actual work?
- How can leaders remain visibly involved without turning every local adoption decision into an executive matter?
- What recurring leadership behavior would make the organization's stated commitment credible over time?
Using GenAI Personally in Relevant Leadership Work
- Which real leadership activities provide sensible opportunities for leaders to use GenAI themselves?
- Where could GenAI improve a leader's research, preparation, option generation, critique, or communication?
- Which leadership decisions should GenAI help prepare without replacing the leader's judgment or accountability?
- What firsthand experience would help leaders understand both GenAI's usefulness and its failure modes?
- How can leaders share what they learned from weak outputs, corrections, or unsuccessful uses rather than showing only polished successes?
- Which personal uses are sufficiently relevant to demonstrate genuine adoption rather than staged examples?
- How can leaders use their own experience to make better decisions about what employees actually need?
Making Time for Teams to Experiment With GenAI
- Which teams need protected learning time because useful GenAI practice cannot develop inside their current workload?
- What real work could provide meaningful experimentation instead of requiring separate artificial exercises?
- Which existing activities could be reduced, delayed, or deprioritized to make experimentation legitimate working time?
- How should managers respond when learning temporarily makes work slower rather than immediately more productive?
- How much repeated practice is needed before employees can fairly judge whether GenAI is useful for their work?
- What would prevent protected experimentation from becoming another task added on top of unchanged expectations?
- What concrete leadership decisions would create enough space for learning without unnecessarily disrupting delivery?
Removing Barriers That Leaders Can Directly Influence
- Which adoption barriers persist because an unresolved leadership decision is required?
- Where are priorities, funding, ownership, approvals, or conflicting organizational expectations blocking otherwise useful adoption?
- Which access or integration problems need leadership intervention across several functions?
- What uncertainty about workforce expectations, permitted experimentation, or resource allocation can leadership resolve directly?
- Where can leaders simplify a process that has become unnecessarily restrictive without weakening genuine controls?
- Which apparent barriers actually require specialist or local action rather than executive intervention?
- Which leadership action would remove the greatest constraint on meaningful adoption now?
Asking Teams About GenAI Opportunities in Their Work
- What recurring work problems do teams believe GenAI could realistically improve?
- Which useful GenAI practices are employees already developing without formal leadership visibility?
- What important work is currently skipped or underdone because it requires too much cognitive effort?
- Where do teams encounter information, preparation, quality, or coordination problems that GenAI might change?
- What constraints prevent employees from testing the opportunities they already see?
- How can leaders invite opportunity discovery without making teams feel expected to produce AI ideas regardless of fit?
- What should happen after a team identifies an opportunity so useful ideas do not disappear into a reporting process?
Following Up on Adoption Progress Regularly
- What should leaders review to understand whether meaningful adoption is developing rather than simply whether usage is rising?
- Which evidence about recurrence, independence, work integration, and value deserves leadership attention?
- What recurring access, management, policy, or support problems need escalation beyond the affected team?
- Which unexpected failures or unintended effects should leaders actively ask teams to surface?
- What locally successful practices or insights deserve examination elsewhere?
- How can follow-up avoid turning experimentation into a compliance exercise where teams hide weak results?
- What should leadership be prepared to change when follow-up reveals that the current adoption approach is not working?
Supporting People When Early Experiments Do Not Work
- What does a failed experiment actually tell us about the task, user, tool, context, or expectation?
- Was the work unsuitable for GenAI, or did the interaction fail for a correctable reason?
- How should leaders respond so employees can report weak results without feeling pressure to present every experiment as a success?
- What support would help users diagnose failure rather than simply repeat the same approach?
- When is another experiment justified, and when is stopping the more sensible conclusion?
- What organizational lesson should be captured if several users encounter the same failure?
- How can leaders normalize evidence-based experimentation without normalizing careless or endless experimentation?
Recognizing Useful GenAI-Supported Work
- What GenAI-supported behaviors actually deserve recognition because they improve work or organizational learning?
- How can recognition focus on value, judgment, verification, or useful experimentation rather than sheer usage?
- Which examples show an employee knowing when not to use or rely on GenAI?
- How should unsuccessful experiments that produce valuable learning be recognized?
- What makes a practice worth making visible to colleagues rather than leaving it as an individual success?
- How can recognition encourage wider experimentation without creating pressure for performative GenAI use?
- Which recognition mechanisms fit naturally into existing management and professional practices?
Maintaining Attention After the Initial Launch
- What important adoption issues are likely to disappear from leadership attention after initial rollout?
- Which indicators should remain visible because they reveal whether adoption is deepening or stagnating?
- How can adoption questions become part of normal operating reviews rather than a permanent special initiative?
- What new evidence, barriers, or capability changes should trigger renewed leadership attention?
- When should leaders reduce their involvement because teams and normal organizational structures can now handle the work?
- How can leadership avoid keeping obsolete adoption activities alive simply because they were once important?
- What cadence keeps leadership appropriately engaged without generating continuous GenAI activity for its own sake?
Demonstrating Consistent Commitment Through Leadership Behavior
- Where do current leadership decisions support the stated GenAI adoption goals, and where do they contradict them?
- Are employees genuinely given time, tools, and permission to develop capability, or only told that adoption matters?
- Do leaders respond constructively when experimentation reveals limitations, poor fit, or uncomfortable workforce questions?
- Are resource and performance decisions consistent with the expectation that learning takes time?
- Are leaders willing to change their own workflows, assumptions, or priorities when evidence supports it?
- What leadership behavior would cause employees to conclude that visible GenAI use matters more than useful work?
- How can commitment become evident through normal decisions about priorities, resources, risk, and work rather than repeated messaging?
Communicating Why GenAI Matters
Developing a Clear Explanation of Why GenAI Matters Now
- What has changed in GenAI capability or the external environment that makes organizational attention justified now?
- Which opportunities or pressures are concrete enough to matter to this organization rather than to businesses in general?
- Why is building practical experience important even where the eventual scale or impact remains uncertain?
- What can we credibly say about current capability without extrapolating from demonstrations or future predictions?
- Which choices remain ours rather than being inevitable consequences of the technology?
- What should employees understand about both the opportunity of acting and the risk of moving thoughtlessly?
- What simple explanation captures why the organization is acting now without relying on hype or fear?
Connecting GenAI Adoption to Organizational Priorities
- Which existing organizational priorities could genuinely be advanced through stronger GenAI capability?
- What actual work connects GenAI adoption with those priorities?
- Where would trying to force a GenAI connection distract from a more important existing objective?
- How could teams recognize what the organizational priority means for their own GenAI opportunities?
- Which concrete examples show a credible link between better GenAI use and an outcome the organization already values?
- Where might GenAI adoption compete with current priorities for time, attention, or resources?
- How can leaders explain GenAI as a means of improving important work rather than a separate technology agenda?
Explaining the Practical Value of GenAI Without Exaggeration
- Which benefits have enough evidence behind them that we can communicate them confidently?
- Which potential benefits should still be described as opportunities to test rather than expected results?
- How should productivity claims distinguish individual time saved from organizational value actually captured?
- What limitations, verification needs, or unsuitable tasks should be acknowledged alongside the benefits?
- Which real examples demonstrate practical value without depending on exceptional users or artificial conditions?
- How can communication avoid both inflated promises and language so cautious that employees cannot see why experimentation is worthwhile?
- What claims would remain credible even if future GenAI capability develops differently from current expectations?
Showing How GenAI Can Improve Everyday Work
- Which recurring work situations provide recognizable examples of where GenAI can help?
- How can examples show support for understanding, comparison, preparation, analysis, critique, and problem solving rather than only writing?
- What does the human contribution look like in each example before, during, and after the GenAI interaction?
- Which examples show value while remaining easy enough for employees to understand and test?
- How can examples include necessary verification and limitations rather than presenting a frictionless demonstration?
- Which examples would feel irrelevant because they do not resemble how employees actually work?
- What everyday stories would help people recognize similar opportunities when they encounter them themselves?
Addressing Common Misunderstandings About GenAI
- Which misunderstandings most distort how employees decide whether and how to use GenAI?
- Where do people confuse fluent output with factual reliability or authoritative knowledge?
- Which employees expect GenAI to automate whole tasks that still require significant human context or judgment?
- Where are people underestimating useful capabilities because their mental model reflects older tools?
- What misconceptions exist about what approved tools can access, retain, search, or do?
- Which misunderstandings are best corrected through firsthand experience rather than additional explanation?
- What small set of distinctions would most improve people's practical judgment without turning communication into technical training?
Explaining What GenAI Adoption Means for Employees
- What changes should employees realistically expect in their everyday work during the current phase of adoption?
- What will people be encouraged or expected to learn rather than simply being told to use GenAI more?
- Which responsibilities for quality, judgment, decisions, and accountability remain with employees?
- How much variation in GenAI use is legitimate because roles and opportunities differ?
- What new activities such as context preparation, verification, or AI supervision could appear alongside tasks that become easier?
- Which role or workforce consequences have actually been decided, and which remain uncertain?
- How can employees understand what adoption means now without treating possible future changes as established plans?
Clarifying What GenAI Adoption Does Not Mean
- Does the organization's adoption goal require every employee to use GenAI frequently?
- Does adoption imply that existing work should be automated whenever technology makes automation possible?
- Which decisions, judgments, or responsibilities remain human even when GenAI supports the underlying work?
- Does successful adoption require replacing every established method that can be made slightly faster?
- What assumptions about immediate job reduction, role elimination, or universal workflow redesign should be explicitly rejected if they have not been decided?
- Where is rational non-use entirely compatible with a successful adoption strategy?
- What clear boundaries would help employees distinguish the actual adoption effort from broader fears or expectations attached to GenAI?
Adapting the Adoption Message for Different Audiences
- What does this audience need to understand in order to make better decisions or change relevant behavior?
- Which concerns and work realities make GenAI matter differently to leaders, managers, professionals, and frontline employees?
- What examples are close enough to each audience's work to make the message concrete?
- Which uncertainties need more emphasis for groups whose work or roles may be affected differently?
- How much technical or policy detail does this audience actually need at this stage?
- Which core facts, decisions, and principles must remain unchanged across every version of the message?
- How can communication be locally relevant without producing contradictory versions of what the organization intends?
Preparing Managers to Explain GenAI Adoption to Their Teams
- What facts, decisions, expectations, and uncertainties must managers understand before discussing adoption?
- Which questions about daily work should managers be able to answer from their own knowledge of the team?
- What concerns can managers address through organizational policy, and which require leadership or specialist input?
- How should managers respond when employees ask about future role or staffing consequences that have not been decided?
- Which examples from the team's own recurring work can make the broader adoption message relevant?
- What language would help managers admit when they do not know an answer without weakening credibility?
- How can the conversation invite questions and experimentation rather than becoming a one-way presentation of the adoption strategy?
Reinforcing the Adoption Message Over Time
- Which core parts of the original message still need reinforcement because they shape everyday decisions?
- What should change in communication as users move from initial experimentation toward recurring and integrated use?
- Which new examples, evidence, or failures should update how the organization talks about GenAI?
- Where has employee experience diverged from the claims made during the original rollout?
- What uncertainties have since been resolved, and what new uncertainties have appeared?
- Which old messages have become misleading because capabilities, policies, or organizational priorities changed?
- How can the adoption narrative evolve while preserving continuity about purpose, responsibility, and what the organization values?
Making GenAI Relevant to Individual Roles
Understanding the Work of a Specific Role
- What outcomes is this role actually responsible for producing?
- Which recurring responsibilities consume the most time, attention, or scarce expertise?
- Where does the role search, interpret, create, compare, or communicate substantial amounts of information?
- Which decisions or judgments define the role's professional contribution?
- What preparation, coordination, or follow-up work surrounds those core responsibilities?
- Which recurring frustrations or bottlenecks make the work harder than it needs to be?
- Which parts of the actual work provide the strongest basis for examining GenAI relevance?
Identifying Role-Specific GenAI Opportunities
- Which recurring activities in this role match work that current GenAI can support reliably enough to be useful?
- Where could GenAI improve preparation, analysis, comparison, drafting, review, or access to information?
- Which activities could become possible more often or at greater depth because GenAI reduces the required effort?
- What organizational information, integrations, or professional context would each opportunity require?
- Where would verification or correction consume enough effort to weaken the opportunity?
- Which tasks remain poor candidates because they depend heavily on tacit context, human authority, or interpersonal judgment?
- Which role-specific opportunities are strong enough to justify testing on real work?
Connecting GenAI to Recurring Responsibilities
- Which responsibilities occur often enough that useful GenAI support could become a natural work habit?
- What recognizable moments within those responsibilities should make GenAI come to mind?
- Where could GenAI become part of normal preparation, analysis, review, or follow-up?
- Which recurring responsibility currently consumes substantial effort without requiring equally substantial unique human judgment?
- Where does the same responsibility vary enough that users still need discretion about whether GenAI will help?
- Which responsibilities should remain primarily human despite recurring often?
- What recurring situations would give this role the clearest path from occasional experimentation to sustained useful use?
- Which real situations does this role encounter often enough that employees will recognize them immediately?
- What example solves a problem the role already considers frustrating or unnecessarily difficult?
- Which example demonstrates a broader GenAI capability without depending on artificial or unusually clean conditions?
- Can the example use familiar inputs, constraints, or outputs from the role's normal work?
- Does it show how the user provides context, judges the result, and verifies what matters?
- Which impressive examples should be avoided because they would not transfer to ordinary work in this role?
- What example is simple enough to try while still teaching something useful about professional GenAI use?
Showing Where GenAI Can Reduce Low-Value Effort
- Which parts of the role consume significant effort while making limited use of its distinctive expertise?
- Where are people repeatedly reorganizing, rewriting, extracting, comparing, or preparing information?
- Which recurring steps exist mainly to move information between formats, audiences, or systems?
- What effort could GenAI reduce without shifting an equal or greater burden into review and correction?
- Where could removing low-value effort free people to spend more time on judgment, relationships, or difficult cases?
- Which apparent efficiency opportunities would merely allow unnecessary work to expand?
- What low-value effort is important and frequent enough that reducing it would materially improve the role?
Showing Where GenAI Can Improve the Quality of the Role's Contribution
- Where could GenAI help this role consider alternatives, evidence, risks, or perspectives that are currently easy to miss?
- Which outputs would benefit from an additional structured critique or completeness check?
- Where could GenAI help people prepare more thoroughly for decisions, conversations, or complex work?
- Which aspects of quality depend on domain expertise that must remain with the role holder?
- Could GenAI reduce quality differences between users without lowering the standard of expert work?
- What additional verification would be needed to ensure that apparent improvement is genuine?
- Which quality improvements would matter most to the people who depend on this role's work?
Clarifying Where Human Judgment Remains Essential
- Which decisions in this role require legitimate human authority regardless of how much GenAI supports the preparation?
- Where does the work depend on tacit organizational, professional, or interpersonal context that GenAI may not possess?
- Which outputs can look convincing while containing errors that only an experienced professional would recognize?
- What values, tradeoffs, or stakeholder consequences require a person to make the final judgment?
- Which unusual cases should be recognized as exceptions rather than forced through a standard GenAI-supported method?
- What expertise must the role continue developing in order to evaluate and supervise GenAI-supported work effectively?
- How should responsibility be divided so GenAI expands capability without obscuring who remains accountable?
Connecting Organizational Adoption Goals to Individual Work
- What part of the organization's adoption outcome is genuinely relevant to this role?
- Which expected adoption behaviors make sense given the opportunities and constraints in this work?
- How could changes in this role contribute to the broader business outcomes adoption is intended to support?
- Where would organization-wide expectations need adaptation because this role's work differs materially?
- What useful practices or learning from this role could eventually benefit other teams?
- What would meaningful adoption look like for someone in this role without relying on raw usage frequency?
- How can the organizational goal be translated into concrete work changes that the employee can recognize and influence?
Preparing Role-Specific Conversations About GenAI
- Which recurring responsibilities should a manager and employee examine together for potential GenAI support?
- What questions would help the employee identify their own opportunities rather than simply receive a list of prescribed uses?
- Which current frustrations, quality problems, or underperformed activities deserve discussion?
- What role-specific risks or usage boundaries need to be understood before experimentation?
- Which examples can provide a useful starting point without becoming the only approved ways of using GenAI?
- What concerns about expertise, workload, role change, or expectations are likely to arise in this conversation?
- What small real-work experiment could turn the discussion into practical learning?
Updating Role-Specific Opportunities as GenAI Capabilities Improve
- Which opportunities previously dismissed as unreliable or impractical should now be reconsidered?
- What newly available capabilities change the kinds of inputs, tools, or actions GenAI can support in this role?
- Which current limitations still remain despite broader technology improvement?
- Have any existing GenAI practices become unnecessarily complicated because they were designed around older limitations?
- What new risks or verification needs appear as GenAI becomes more capable or autonomous?
- Which old examples or guidance should be retired because they no longer represent the best available approach?
- What evidence should be required before changing established role practices around a new capability?
Managing Workforce Concerns and Uncertainty
Identifying the Main Concerns Employees Have About GenAI
- What specific concerns are employees expressing about GenAI and its use in the organization?
- Which concerns involve jobs, role changes, skills, workload, monitoring, quality, fairness, or professional identity?
- How do concerns differ across groups because their work, exposure, or organizational position differs?
- Which concerns point to real unresolved organizational decisions rather than misunderstanding or lack of information?
- Where might employees be reluctant to voice concerns because they fear appearing resistant or incapable?
- What can we learn from differences between what employees say publicly, privately, and through their behavior?
- Which concerns most need leadership understanding because they could materially shape trust, experimentation, or adoption?
Understanding Concerns About Job and Role Changes
- What concrete changes to tasks or responsibilities are employees worried GenAI may cause?
- Which parts of those concerns are supported by changes already occurring in the work?
- What role changes has the organization actually decided, and what remains merely possible?
- Where is GenAI more likely to alter the composition of a role than eliminate it entirely?
- What new responsibilities for judgment, verification, AI direction, or exception handling could accompany reduced routine work?
- Which long-term workforce effects remain too uncertain for leadership to predict credibly?
- How can the organization discuss likely change without disguising organizational choices as technological inevitabilities?
Addressing Anxiety About New Skill Expectations
- What GenAI capabilities are employees currently expected to develop, and which expectations are only anticipated?
- Which groups may feel disadvantaged because colleagues have had more access, experience, or time to learn?
- What practical capability is actually required for competent use in this role rather than for becoming an advanced user?
- Which skills can develop through normal real-work practice, and which need structured learning or specialist support?
- How can employees see that weak early performance is part of learning rather than evidence that they are failing a new professional test?
- What new human capabilities such as judgment, verification, problem framing, or domain expertise become more important alongside GenAI?
- How should expectations be communicated so people know what they need to learn now without being overwhelmed by hypothetical future requirements?
Responding to Concerns About Quality and Reliability
- What types of GenAI failure would matter most in the work employees are being asked to support?
- Which quality concerns reflect genuine model limitations or verification problems?
- Where might employees be overgeneralizing from one poor experience to work GenAI could support effectively?
- What review, testing, or source-checking methods can make suitable GenAI use acceptably reliable?
- Which users have enough expertise to detect subtle errors, and where would reliance create hidden risk?
- Which tasks remain unsuitable because the quality burden or consequence of error is too high?
- What would help employees develop calibrated reliance rather than being asked simply to trust GenAI more?
Addressing Concerns About Monitoring or Evaluation
- What data about employee GenAI use is actually collected by the organization or its tools?
- What legitimate organizational purpose does each form of monitoring serve?
- Could raw usage data be misinterpreted as evidence of productivity, capability, commitment, or resistance?
- Will employees be evaluated differently because they use GenAI more or less often?
- What protections or boundaries should apply to prompts, interaction histories, or other sensitive usage information?
- How can the organization make its measurement practices transparent enough that employees do not have to infer hidden purposes?
- What should leadership explicitly say about what GenAI telemetry will and will not be used to judge?
Understanding Fear of Losing Professional Expertise
- Which parts of employees' expertise could weaken if GenAI routinely performs work that previously built or exercised those capabilities?
- Which knowledge or skills must people retain in order to verify, challenge, or recover from weak GenAI output?
- What routine work currently serves as a pathway through which less-experienced employees develop professional judgment?
- Which capabilities become more valuable as employees spend more time directing and evaluating GenAI-supported work?
- Where can GenAI be used to strengthen learning rather than simply bypass the thinking through which expertise develops?
- Which activities should still be performed manually often enough to preserve resilience or competence?
- How can the organization address deskilling as a genuine design issue rather than dismissing it as reluctance to adopt technology?
Discussing How Workloads May Change With GenAI
- Which activities could genuinely require less effort if GenAI use becomes effective?
- What new work may appear through prompting, context preparation, verification, exception handling, or supervision?
- Could faster production create higher output expectations that absorb all released capacity?
- Where might GenAI increase cognitive load by concentrating human work around difficult judgments and exceptions?
- What does the organization intend to do with capacity that is genuinely released?
- Which workload expectations would make experimentation feel like additional work rather than a new way of working?
- How can leaders discuss potential efficiency honestly before the full effect on workloads is known?
Addressing Concerns About Fairness and Unequal Access
- Which employees currently have materially better access to GenAI tools, information, integrations, or learning opportunities?
- How could differences in access affect performance, development, or future expectations between otherwise comparable employees?
- Which roles naturally offer more GenAI-suitable work and therefore more opportunities to build experience?
- Are some groups receiving better manager support or protected learning time than others?
- What inequalities are temporary consequences of phased adoption, and which reflect avoidable organizational choices?
- How should performance expectations account for unequal access or different opportunity levels?
- What action would reduce unfair disparities without assuming that every role requires identical tools or levels of GenAI use?
Communicating Honestly Where the Future Remains Uncertain
- What can the organization currently state with confidence about GenAI's effect on work and roles?
- Which important future questions depend on technology development, business choices, or evidence that does not yet exist?
- What workforce decisions have not been made and should not be implied through vague language?
- How can leaders distinguish current expectations from promises or predictions?
- What is the organization doing now to learn enough to make later decisions responsibly?
- What events, evidence, or decision points should trigger an update to employees?
- How can uncertainty be communicated clearly enough to maintain credibility without amplifying speculation?
Creating Ongoing Dialogue About Workforce Concerns
- Where can employees raise difficult GenAI concerns without being treated as resistant to adoption?
- How can managers create enough psychological safety for people to admit confusion, poor results, or disagreement about a proposed use?
- Which concerns are best addressed in team discussion, and which require private or specialist channels?
- How should recurring concerns be aggregated so leadership can see patterns without turning feedback into surveillance?
- What should managers do when an employee raises a question they cannot credibly answer?
- How will employees see whether their feedback influenced policy, support, workload, tooling, or adoption decisions?
- What ongoing dialogue would help the organization learn from workforce concerns rather than treating communication as a one-time reassurance exercise?
Enabling Practical Access to GenAI
- Which approved GenAI tools can employees actually access for organizational work today?
- What meaningful capabilities does each available tool provide, including files, search, organizational knowledge, multimodality, code, or external actions?
- Which user groups have access to different tools or capability levels, and why?
- Where does an available tool lack something employees need for the work they are trying to perform?
- Which tools are embedded in existing work systems, and which require employees to leave their normal workflow?
- Where are employees turning to personal accounts or unapproved tools because the approved options do not meet their needs?
- What simple tool map would help employees understand which available option fits which kind of work?
Finding Groups That Lack Practical Access
- Which employees lack any usable access to an approved GenAI tool?
- Who technically has access but cannot use the capabilities their real work requires?
- Which groups face device, network, location, authentication, or environment restrictions that make access inconvenient or unreliable?
- Where do differences in licenses, permissions, or integrations create materially unequal opportunities to use GenAI?
- Which employees can experiment with GenAI but cannot use it on the organizational information needed for meaningful work?
- How should adoption expectations differ for groups whose practical access remains limited?
- Which access gaps should be resolved first because they block the greatest amount of potentially valuable work?
Understanding Where Access Processes Create Friction
- What does an employee currently have to do before they can begin using an approved GenAI tool?
- Which approvals, requests, authentication steps, or waiting periods create unnecessary delay?
- Where do employees struggle to find the instructions, permissions, or information needed to obtain access?
- Which access problems generate the same support requests repeatedly?
- At what points do users abandon the process or turn to easier unapproved alternatives?
- Which access controls serve a real security or governance purpose, and which friction has accumulated without enough benefit?
- What could be simplified while preserving the controls that genuinely matter?
Making Initial Access Easy to Understand
- Can a new user quickly tell which GenAI tool they are supposed to use?
- Are the steps for getting access clear enough to follow without specialist help?
- What does the user need to know before signing in, and what information can wait until later?
- Which terminology, account types, or product choices are currently confusing beginners?
- What common setup mistakes prevent people from reaching a usable first interaction?
- Can employees easily distinguish normal access problems from situations that require additional permission?
- What minimum guidance would get a legitimate user from no access to a working tool with the least unnecessary effort?
- What does a new user need to know to begin a useful first interaction safely?
- Which real, low-consequence work situation provides a sensible starting point with the available tool?
- Does the user understand what context or material can be supplied and what should not be entered?
- Can they tell which features, such as file upload, search, organizational knowledge, or tools, are available to them?
- Where should they turn if the tool behaves unexpectedly or the first result is poor?
- What can users safely discover through experimentation instead of being taught in advance?
- What minimum orientation is enough for someone to begin learning through real work rather than waiting until they feel fully trained?
Identifying Technical or Device Constraints That Limit Use
- Which devices, browsers, networks, or work environments support the organization's approved GenAI tools reliably?
- Where do performance, latency, compatibility, or session limits make GenAI impractical during ordinary work?
- Which important capabilities disappear on devices or environments commonly used by particular groups?
- Are security configurations or network restrictions blocking legitimate features employees need?
- Where does the technical environment force users to move information manually between otherwise disconnected systems?
- Which constraints are temporary product limitations, and which require organizational technology changes?
- What technical barriers must be removed before broader adoption can reasonably be expected from the affected users?
Understanding Permission or Account Problems
- Which account or permission problems are preventing employees from using capabilities they should already have?
- Where do different license levels create unexpected differences in what users can do?
- Are identity, authentication, or organizational account configurations causing recurring failures?
- Which employees need access to shared files, knowledge, connectors, or systems before GenAI becomes useful for their work?
- What permission changes can normal support resolve, and what requires a system owner or governance decision?
- Where are permission requests taking long enough that employees abandon a legitimate use?
- What recurring account problems should be fixed at the system level rather than handled one user at a time?
- What organizational information does GenAI need in order to support the work employees actually want to perform?
- Which useful documents, data, knowledge bases, or systems are currently inaccessible to the approved GenAI environment?
- Where are employees manually copying or explaining context that already exists elsewhere in the organization?
- Which information gaps reduce the quality of GenAI enough that users prefer external or manual alternatives?
- What information could be connected safely through retrieval, integrations, or controlled access?
- Which information should deliberately remain inaccessible because the risk of broader access outweighs the adoption benefit?
- Which missing information access creates enough lost value to justify an integration or data-access initiative?
Providing Clear Support When Access Problems Occur
- Where can employees go when they cannot access or use an approved GenAI capability?
- Can the support path distinguish account, permission, product, integration, policy, and device problems without sending users between multiple teams?
- Which common access issues should employees be able to resolve themselves through clear guidance?
- Which problems require fast resolution because they block recurring work rather than occasional experimentation?
- What recurring support cases reveal a deeper flaw in access design or documentation?
- How should unresolved capability requests be routed when the current approved environment genuinely cannot support the work?
- What support experience would prevent a practical access problem from becoming lasting disengagement from GenAI?
Confirming That People Can Actually Use GenAI Before Expecting Adoption
- Does the target group have working access to a GenAI tool that is capable enough for the intended work?
- Can users authenticate, open the tool, and complete a basic interaction without unusual assistance?
- Do they know which approved tool to use for the situations they are expected to encounter?
- Can they reach the documents, data, or organizational context required for meaningful use?
- What technical, account, permission, or integration problems still prevent practical adoption?
- Are managers or leaders judging low usage in groups that have never had genuinely usable access?
- What access conditions should be verified before the organization interprets non-use as an adoption problem?
Clarifying Safe and Practical GenAI Usage Boundaries
- Which GenAI tools are approved for organizational work, and what kinds of use is each approved for?
- Are some tools appropriate for public information but unsuitable for confidential organizational material?
- How should employees decide between several approved tools with different data protections or capabilities?
- What should someone do when an unapproved tool offers a capability that the approved environment lacks?
- When are personal accounts or consumer versions inappropriate even if the underlying product is otherwise approved?
- How can employees check the status of a new or unfamiliar GenAI tool before using it for work?
- What simple rule would let most employees choose an appropriate tool without needing specialist advice each time?
- What categories of organizational information can employees enter into approved GenAI tools during normal work?
- Which information requires additional conditions, such as a particular tool, environment, permission, or form of anonymization?
- What information should not be supplied to the available GenAI systems at all?
- How should users handle documents or conversations that contain a mixture of routine and sensitive information?
- Do the rules differ when GenAI accesses information through an approved organizational connector rather than manual input?
- What realistic work examples would make the information-handling boundaries easier to apply?
- What should an employee do when they cannot confidently classify the information they need to use?
- What kinds of information can employees reasonably treat as routine for ordinary GenAI use?
- Which characteristics make information sensitive in this organization?
- What categories are most likely to be misclassified because sensitivity depends on context rather than obvious labels?
- How do personal data, confidential business information, client material, trade secrets, privileged information, or regulated data differ in practical handling?
- When can apparently harmless information become sensitive because several pieces are combined?
- What concrete examples would help employees recognize the boundary without becoming information-governance specialists?
- What simple decision process should someone follow when they are uncertain whether information is routine or sensitive?
- Which work must remain inside approved organizational GenAI environments?
- What information or activities make external GenAI tools clearly unsuitable?
- When, if ever, can an external tool be used safely with public or non-sensitive information?
- Does a useful capability in an external tool justify use, or should the missing capability instead be raised through an organizational request?
- What risks arise from account terms, data retention, training use, access controls, or lack of organizational oversight?
- How should employees respond when colleagues are already using an external tool because it works better than the approved alternative?
- What boundary protects the organization without becoming so broad that employees avoid legitimate low-risk GenAI use?
Clarifying When GenAI Outputs Require Verification
- Which kinds of GenAI output require independent verification before they can influence work?
- When is ordinary professional review enough, and when is a source, calculation, test, or authoritative record required?
- Which factual claims are especially likely to need checking because the model may be incomplete, outdated, or wrong?
- How should the required verification change with the consequence of an error?
- What should users check when GenAI produces persuasive reasoning but the underlying evidence is uncertain?
- At what point does the verification burden make GenAI a poor fit for the task?
- What practical review expectation can employees apply without treating every GenAI output as equally risky?
Clarifying Which Decisions Must Remain With People
- Which decisions require a person to retain explicit authority and responsibility regardless of GenAI involvement?
- Where can GenAI prepare evidence, alternatives, or recommendations without determining the final outcome?
- Which decisions involve values, rights, professional judgment, or consequences that make sole GenAI determination inappropriate?
- What low-consequence decisions or actions could reasonably be delegated within clearly defined limits?
- How should employees distinguish GenAI assistance from a decision that must be owned by a person?
- When GenAI can act in external systems, which actions require confirmation before execution?
- Where should human ownership be documented so accountability remains clear even when much of the underlying work is AI-assisted?
Understanding Additional Boundaries in Regulated or Sensitive Work
- What additional legal, regulatory, contractual, professional, or safety requirements apply to this kind of work?
- Which GenAI uses require specialist review before employees can proceed?
- What information-handling restrictions are stricter here than in ordinary organizational work?
- Which decisions or outputs require formal human review, documentation, or approval?
- What records need to be retained about GenAI-supported activity?
- Where does general organization-wide guidance leave important questions unanswered for this function?
- What specialized guidance would let employees act confidently without referring every normal case to compliance or legal specialists?
Clarifying Intellectual Property and Confidentiality Considerations
- What copyrighted, licensed, proprietary, or confidential material can be provided to the available GenAI tools?
- What restrictions apply when employees use client, partner, supplier, or third-party content?
- How should users handle material when ownership or usage rights are uncertain?
- What intellectual-property issues arise when GenAI produces material based on external sources or existing organizational content?
- Which outputs may require additional review before publication, reuse, or commercialization?
- What situations genuinely require legal advice rather than a standard employee rule?
- What practical principles can employees use to handle common intellectual-property and confidentiality questions without becoming legal experts?
Identifying Where Employees Need Further Guidance
- Which recurring GenAI situations are employees least confident about handling within current rules?
- What questions are repeatedly being escalated to managers, support teams, legal, privacy, or security specialists?
- Where are different teams interpreting the same boundary differently?
- Which employees are avoiding useful GenAI work because the cost of resolving uncertainty is too high?
- Which recurring questions can be answered through clearer examples or decision aids rather than new policy?
- What genuinely novel or consequential edge cases still require specialist judgment?
- What pattern in employee questions shows that the guidance itself should be improved?
Making Usage Boundaries Easy to Apply in Everyday Work
- Can employees apply the important GenAI boundaries while working without repeatedly consulting lengthy policy documents?
- What small number of questions about the tool, information, purpose, output, and consequence would resolve most normal cases?
- Which examples should show clearly allowed, clearly unsuitable, conditional, and escalation-required situations?
- Where are current rules technically correct but too abstract to guide an employee's real decision?
- How can relevant guidance appear at the point of work rather than only in separate governance material?
- What should employees do when they reach a case that remains genuinely uncertain after applying the normal rules?
- How can the organization maximize independent legitimate use while reserving specialist review for situations that truly require it?
Helping People Recognize Useful GenAI Opportunities in Their Work
Helping People See Beyond the Most Obvious GenAI Uses
- What forms of GenAI use do employees currently think of first when they encounter the technology?
- Which useful forms of analysis, comparison, explanation, preparation, critique, exploration, or problem solving remain outside that mental model?
- How can examples demonstrate broader cognitive support without implying that GenAI should make the final judgment?
- Which current capabilities make GenAI useful for more than generating or summarizing text?
- What examples would help employees see a transferable work pattern rather than one clever use case?
- Which impressive demonstrations should be avoided because they are too remote from ordinary work to improve opportunity recognition?
- What broader mental model would help employees notice GenAI possibilities independently in unfamiliar situations?
Connecting GenAI Opportunities to Recurring Work Situations
- Which situations recur often enough in this role for GenAI support to become a useful habit?
- What recognizable work cues signal that GenAI might be worth considering?
- Where do employees repeatedly face the same kind of information, preparation, comparison, or uncertainty problem?
- Which recurring situations are valuable enough that improving them would matter over time?
- Where should GenAI merely be considered rather than automatically used whenever the situation occurs?
- How can employees describe the underlying situation in a way that transfers across different individual tasks?
- What small set of recurring situations would make the strongest foundation for everyday opportunity recognition?
- Where does this work require people to read, search, combine, or interpret large amounts of information?
- When are employees manually summarizing or comparing several sources before they can proceed?
- Which tasks involve extracting a consistent set of details from changing documents or records?
- Where does information volume make it difficult to identify differences, gaps, patterns, or the main point?
- Can current GenAI access enough of the relevant information to help meaningfully?
- How easily could the employee verify important conclusions produced from that information?
- What recurring information pattern should become a cue to consider GenAI support?
Helping People Recognize Moments of Uncertainty or Decision
- What kinds of uncertainty regularly cause employees to pause, search, or ask others for another perspective?
- Which decisions benefit from generating alternatives, exposing assumptions, or comparing tradeoffs?
- Where could GenAI help structure an unclear situation before a person decides what to do?
- What moments would benefit from having GenAI challenge an initial view or identify missing considerations?
- Which uncertainties are suitable for GenAI-supported preparation but still require a person's judgment and authority?
- Where would limited context or high consequence make GenAI a poor source of decision support?
- How can uncertainty become a cue for useful GenAI assistance without turning GenAI into the default decision-maker?
Identifying Opportunities During Writing and Communication Work
- Which recurring writing or communication tasks require more preparation effort than their underlying message warrants?
- Where do employees need to turn rough information into a clear structure or first draft?
- Which communications need repeated adaptation for different audiences, formats, or levels of detail?
- Where could GenAI help incorporate feedback, compare versions, or improve clarity without replacing human ownership of the message?
- Which communications depend on sensitivity, relationships, authority, or nuance that require especially strong human authorship?
- What verification or approval is necessary before GenAI-assisted communication is sent externally?
- Which recurring communication situations should naturally trigger consideration of GenAI support?
Identifying Opportunities During Planning and Preparation
- Which activities require employees to repeatedly prepare agendas, plans, questions, checklists, scenarios, or background material?
- Where could GenAI help break a broad objective into workable parts?
- Which preparation tasks would benefit from anticipating risks, objections, dependencies, or missing information?
- Where could GenAI generate alternatives that a person can compare before committing to a plan?
- What local context would the model need before its preparation is genuinely useful?
- Which plans involve consequences or uncertainty too high to rely heavily on generic GenAI reasoning?
- What recurring preparation moments provide strong opportunities for GenAI to improve the quality of later human work?
Identifying Opportunities During Meetings and Collaboration
- What work before a meeting could GenAI help prepare more thoroughly or efficiently?
- Where could GenAI help participants understand relevant background, alternatives, or unresolved questions?
- What meeting outputs could GenAI help organize, summarize, or convert into clear actions afterward?
- Where could GenAI help compare different viewpoints without replacing the human discussion needed to resolve them?
- Which collaborative work suffers because context repeatedly has to be reconstructed between participants?
- What confidentiality or consent issues limit using meeting material with GenAI?
- Which recurring collaboration moments are useful enough that GenAI support could become part of normal team practice?
Helping People Recognize GenAI Opportunities in Problems They Already Face
- What recurring problems or frustrations currently consume disproportionate employee effort?
- Which of those problems involve information, language, alternatives, preparation, comparison, or other work GenAI handles reasonably well?
- Where could GenAI help clarify or structure the problem before anyone attempts to solve it?
- Which problems persist because employees lack time or access to another perspective rather than because the underlying issue is technically difficult?
- What organizational context would GenAI need before it could contribute meaningfully?
- Which frustrations are actually caused by broken processes, missing authority, or poor systems that GenAI would only mask?
- What small experiment could test whether GenAI changes the problem enough to be worth continued use?
Using Real Work Examples to Expand Opportunity Recognition
- Which real GenAI examples are close enough to the audience's work that they can recognize the underlying situation?
- What work problem caused the original user to consider GenAI?
- What did GenAI contribute that made the example genuinely useful?
- What human expertise, context, iteration, or verification was necessary for the result?
- What limitation or failure should be shown so the example does not imply effortless success?
- What part of the example transfers to other roles even if the specific prompt or output does not?
- How can the example encourage employees to identify analogous opportunities rather than simply copy the demonstrated interaction?
Helping People Begin Recognizing New GenAI Opportunities Independently
- Are users beginning to consider GenAI when relevant work situations arise without being reminded by an adoption initiative?
- What kinds of situations are they now recognizing that they previously overlooked?
- Can users generalize from one successful practice to a different task with the same underlying pattern?
- Do they know how to test an unfamiliar opportunity without assuming GenAI must be useful?
- Can they distinguish a promising opportunity from work that is too consequential, contextual, or difficult to verify?
- How can managers strengthen recognition by asking useful questions rather than supplying all the use cases?
- What evidence would show that opportunity recognition has become an independent capability rather than continued dependence on examples?
Supporting Early Experimentation
Choosing Real Work That Is Safe to Experiment With
- Which real work is relevant enough that a successful experiment would matter to the user?
- Is the task low enough in consequence that an imperfect result can become learning rather than damage?
- Can the user judge the quality of the result from their own knowledge or reliable evidence?
- Can the experiment be performed without introducing unnecessary privacy, confidentiality, or permission complexity?
- Is the task simple enough for a beginner while still requiring some meaningful interaction with GenAI?
- Which work should be excluded because failure would be difficult to detect or too costly at this stage?
- What real recurring task gives the best balance of relevance, safety, verifiability, and learning value?
Selecting an Initial Task That Can Produce Visible Value
- Which task currently requires enough effort that improvement would be obvious to the user?
- Can the employee understand what a good result should look like before beginning the experiment?
- Is the task narrow enough that the user can reach a useful result without complex setup or integrations?
- Does the task matter enough that success would motivate further experimentation rather than feel like a novelty?
- Can the GenAI-assisted result be compared directly with how the work is normally performed?
- What dependencies could obscure whether GenAI itself contributed to the improvement?
- Which initial task offers a realistic chance of visible value without being engineered to guarantee success?
Trying GenAI on Work That Would Otherwise Be Done Manually
- How is this work normally performed without GenAI?
- Which specific part of the manual effort is GenAI being asked to support?
- What context, source material, or constraints does the model need in order to contribute meaningfully?
- What work must the employee still perform before and after the GenAI interaction?
- Where does GenAI genuinely remove effort rather than merely move it into context preparation or correction?
- Does the reviewed result meet the quality standard required for the task?
- What does the experiment reveal about whether this manual activity is a good candidate for recurring GenAI support?
Comparing GenAI-Assisted Work With the Existing Approach
- How does the full time required with GenAI compare with the current method, including preparation and review?
- Does GenAI change the quality, completeness, or usefulness of the final work?
- What additional verification or correction does the GenAI-assisted version require?
- Which parts of the task become easier, and which become more complicated?
- Does the GenAI-supported method change only the employee's effort, or does it improve the wider workflow?
- What new risks or limitations appear that are absent from the existing approach?
- Is the observed improvement large and repeatable enough to justify another experiment?
Improving Weak Initial Results Through Iteration
- What specifically is weak, missing, generic, inaccurate, or unsuitable in the initial response?
- Does the model need more context, clearer constraints, examples, source material, or a better explanation of the intended outcome?
- What follow-up would most directly address the weakness rather than simply asking for another version?
- Would separating the task into stages make the work easier to evaluate and improve?
- Is the problem with the interaction, or is the task itself a poor fit for the available GenAI capability?
- At what point does continued iteration cost more than the additional improvement is worth?
- What does the process teach the user about how to recover from weak GenAI results in future work?
Trying Different Ways to Use GenAI on the Same Task
- What changes if GenAI generates the work, critiques existing work, or compares several alternatives?
- Would providing more relevant context materially improve the result?
- Does handling the task in stages produce a more reliable outcome than asking for everything at once?
- What happens if the user asks GenAI to expose assumptions or weaknesses instead of simply giving an answer?
- Which interaction pattern produces the best balance of usefulness, user effort, and verification?
- What does the comparison reveal about the contribution GenAI is best suited to make?
- Which lesson from trying different approaches could transfer to other work situations?
Capturing What Was Learned From an Experiment
- What real work situation was tested, and what was the intended improvement?
- What did GenAI do well enough to create practical value?
- Where did the interaction fail, create extra work, or require stronger human judgment?
- What context, expertise, tool capability, or verification condition materially affected the result?
- Which part of the result depended on this particular case, and which lesson appears more general?
- What would the user do differently if the same situation occurred again?
- What is worth recording or sharing so others can learn without treating one experiment as an established practice?
Sharing Early Experiments With Colleagues
- Is this experiment useful enough that colleagues are likely to encounter a similar situation?
- What context do others need to understand why GenAI was used and what problem it was solving?
- What did the user contribute in terms of expertise, information, iteration, or judgment?
- What limitations, failures, and verification requirements should be shown alongside the successful result?
- Which part of the approach is transferable, and which part should not simply be copied?
- How could colleagues test the underlying pattern on their own work rather than reproduce the exact prompt?
- What feedback from other users would help determine whether the discovery is more broadly useful?
Deciding Which Experiments Are Worth Repeating
- Did the experiment create enough improvement in effort, quality, capacity, or another meaningful outcome to justify further attention?
- Was the result dependable enough that another attempt is likely to teach us something rather than merely reproduce randomness?
- How much context preparation, iteration, checking, and correction was required?
- Does the work recur often enough for repeated use to matter?
- What unresolved tool, information, governance, or skill issue currently limits the experiment?
- Would repeating the experiment with another case or user test an important uncertainty?
- Which experiments deserve another cycle, and which should be stopped because the opportunity appears weak?
Turning Successful Experiments Into Continued Use
- What underlying condition made this experiment successful?
- How often does the same work situation recur in normal practice?
- Which parts of the method can be reused, and which need fresh judgment each time?
- What cue could help the user remember to consider GenAI when the situation appears again?
- What verification should remain part of the recurring practice?
- Does the approach need further replication before it is shared or standardized for others?
- What would show that the experiment has become a useful working method rather than an isolated success?
Building Initial GenAI Confidence
Helping Someone Complete Their First Useful GenAI Interaction
- What real, low-consequence task would be meaningful enough to give this person a credible first experience?
- What minimum context or source material does GenAI need before the user begins?
- What outcome would be useful enough for the person to see why further experimentation could be worthwhile?
- Is the task simple enough that interaction difficulty does not overwhelm the learning objective?
- What should the user inspect or verify before treating the result as usable?
- What did the interaction reveal about both GenAI's capability and the user's own role in producing a good result?
- What natural next task would build on this success without assuming the person is already an independent user?
Recovering From a Weak First GenAI Result
- What specifically made the first result disappointing or unusable?
- Was the task itself suitable for GenAI at the user's current level of experience?
- Did the model lack important context, source material, constraints, or examples?
- Could a focused follow-up materially improve the result?
- Were the user's expectations based on capabilities the available tool does not reliably have?
- Would a smaller or more verifiable task provide a better second experience?
- How can the user leave the failed attempt with a more accurate understanding rather than concluding either that GenAI is useless or that they personally failed?
Understanding Why a GenAI Interaction Did Not Work
- Was the objective clear enough for GenAI to understand what useful help would look like?
- Did the interaction omit information that a knowledgeable human would need to perform the same work?
- Was the task too broad or complex to handle effectively in one interaction?
- Did the user expect GenAI to know current, organizational, or specialist information it could not access?
- Is the weak result primarily a user-interaction problem, a model limitation, a tool limitation, or poor task fit?
- What evidence would distinguish those explanations before the user simply tries again?
- What lesson should the user carry forward so the same type of failure becomes easier to diagnose next time?
Learning to Improve Results Through Follow-Up
- What part of the initial result needs the most improvement?
- What additional context, constraint, example, or source would help GenAI respond more usefully?
- How can the user point out a weakness or incorrect assumption instead of restarting the interaction immediately?
- When should the user ask for alternatives, deeper analysis, or a different structure?
- When is it better to begin a fresh interaction because the current conversation is built on a poor premise?
- How can the user tell when additional refinement is producing diminishing returns?
- What general follow-up behavior would help the user treat GenAI as an iterative work process rather than a one-shot answer system?
Building Confidence in Reviewing GenAI Outputs
- What criteria should the user apply when deciding whether the output is actually useful?
- Which parts of the response can be judged from the user's own professional knowledge?
- What assumptions or missing context should the user look for before accepting the result?
- Which claims or recommendations require independent evidence rather than confidence in the model's wording?
- Where might the user's own lack of domain knowledge make reliable review impossible?
- What should the user do when an answer is plausible but they cannot establish whether it is correct?
- How can repeated evaluation build confidence in their own judgment rather than confidence in GenAI by default?
- What part of this output could cause a meaningful problem if it were wrong?
- Which facts, figures, quotations, sources, dates, or external claims require confirmation?
- Does the task depend on current information that the model may not reliably possess without search or retrieval?
- What authoritative source or independent method should be used for verification?
- How much checking is proportionate to the consequence and detectability of an error?
- When is an explicitly uncertain or approximate answer acceptable for the purpose?
- How can the user make verification a normal judgment rather than either checking everything or trusting everything?
Expanding From Simple to More Difficult Tasks
- Which kinds of GenAI work can the user already perform reliably without immediate help?
- What next task introduces only one or two new challenges rather than several at once?
- Does the more difficult work require better context selection, decomposition, evaluation, or verification?
- How can the task be divided so the user can inspect intermediate results rather than lose control of the whole process?
- What additional consequence or uncertainty requires stronger review?
- At what point does the user need coaching, domain expertise, or specialist support rather than independent experimentation?
- What progression would expand capability while keeping confidence tied to demonstrated competence?
Learning How to Respond When GenAI Makes a Mistake
- What kind of mistake did GenAI make, and how was it detected?
- How important would the error have been if the user had not noticed it?
- Did the mistake result from missing context, weak reasoning, incorrect information, tool failure, or a task beyond the model's reliable capability?
- Can the error be corrected through clarification, additional evidence, or a different approach?
- Does the mistake reveal that future outputs in this task need independent verification?
- When should repeated errors lead the user to stop relying on GenAI for this work?
- How can the experience improve the user's judgment about when GenAI is useful rather than simply increasing or destroying trust?
Repeating Successful Uses Until They Feel Familiar
- Which successful GenAI-supported task recurs often enough to justify repeated practice?
- What recognizable cue can help the user remember the method when the situation returns?
- Which context, criteria, or interaction steps remain similar across cases?
- What changes from one case to another and still requires active judgment?
- Is the user becoming faster at starting, refining, and reviewing the interaction?
- Can the user adapt the method without depending on the original example or exact prompt?
- When has the practice become familiar enough to occur naturally without turning into uncritical routine?
- Can the user independently recognize when GenAI is worth considering?
- Can they describe the work, objective, constraints, and relevant context clearly enough to begin?
- Can they diagnose and improve a weak result instead of immediately asking someone else for help?
- Can they judge when output is usable, when it needs verification, and when they cannot evaluate it reliably?
- Can they adapt a useful approach to a new situation rather than relying on fixed prompts or templates?
- Do they know when poor task fit, risk, or uncertainty makes escalation or non-use the better decision?
- What evidence would show that confidence has become appropriately independent and calibrated rather than simply higher?
Learning From Early Users
Identifying People Who Are Already Using GenAI Regularly
- Who is using GenAI often enough across real work to have accumulated meaningful practical experience?
- Which users go beyond basic drafting or summarization into analysis, comparison, research, review, or multi-step work?
- Who can explain not only what worked but also where GenAI failed, required correction, or was the wrong tool?
- Which users have learned to adapt their methods rather than relying on fixed prompts or examples?
- What work, expertise, autonomy, or access conditions make these users different from colleagues who use GenAI less?
- Are experienced users concentrated in particular roles or teams whose circumstances may limit how representative they are?
- Which early users can provide the most useful evidence about opportunities, limitations, and adoption conditions?
Collecting Examples of Useful Early GenAI Use
- What real work situations are early users repeatedly supporting with GenAI?
- What problem or constraint made GenAI worth using in each example?
- What context, source material, or organizational information did the user provide?
- How did the interaction develop beyond the first response?
- What human judgment, verification, or correction was required before the result became usable?
- What practical benefit was observed, and what limitations remained?
- Which examples are documented well enough to become useful organizational learning rather than isolated success stories?
Understanding Why Certain Uses Became Successful
- What underlying work problem made this GenAI use valuable in the first place?
- What characteristics of the task made current GenAI particularly suitable for it?
- What user expertise, interaction capability, or professional judgment contributed to the result?
- What information, tool access, integrations, or organizational context made the practice possible?
- How much did manager support, experimentation time, or local autonomy contribute to success?
- Which elements are essential to the value mechanism, and which are incidental features of this particular implementation?
- What conditions would need to exist before expecting similar success elsewhere?
Identifying Problems Early Users Encounter
- What difficulties recur even among employees who already use GenAI regularly?
- Which problems come from model capability rather than the user's interaction technique?
- Where do missing organizational information, integrations, or permissions limit otherwise useful work?
- What recurring verification, quality, or reliability problems make continued use harder?
- Which workflow frictions create more copying, checking, or coordination than the benefit justifies?
- What policy or support questions continue to require unnecessary escalation?
- Which recurring early-user problems reveal organizational issues that should be solved before wider adoption?
Understanding Workarounds Early Users Have Developed
- What informal methods are experienced users creating to make GenAI useful despite current limitations?
- What specific tool, information, workflow, or governance problem is each workaround compensating for?
- Which workarounds reveal capabilities or integrations that users genuinely need?
- Where does a workaround create new privacy, quality, continuity, or dependency risks?
- Which methods are useful adaptations that others could learn from rather than symptoms of a broken environment?
- Would solving the underlying organizational problem make the workaround unnecessary?
- What should the organization learn from workarounds before deciding whether to formalize, replace, or discourage them?
Identifying Quality Problems in Early GenAI Use
- What types of quality failure appear repeatedly in actual GenAI-supported work?
- Which tasks produce plausible results that are particularly difficult to verify?
- Where are users accepting polished output before checking evidence, assumptions, or omissions?
- Which failures arise because the model lacks relevant organizational or domain context?
- What verification or review methods are experienced users finding effective?
- Which quality problems improve with better working methods, and which reflect persistent capability limits?
- What quality evidence should exist before an early practice is shared or scaled more broadly?
Understanding Where Early Users Are Creating Measurable Value
- Which early uses show credible improvement in time, quality, capacity, decision preparation, or newly enabled work?
- Does the benefit remain after context preparation, interaction, verification, correction, and support effort are included?
- Which improvements can be observed directly rather than inferred from user enthusiasm?
- Where has released time actually been converted into greater output, better work, or additional capacity?
- Which uses create value repeatedly across cases rather than only under favorable conditions?
- Where is GenAI's contribution difficult to separate from user expertise, process improvement, or other simultaneous changes?
- Which early uses provide the strongest evidence that deeper adoption deserves further investment?
Identifying Working Patterns That Others Could Reuse
- Which successful GenAI uses share the same underlying interaction or work pattern?
- What problem, context, and value mechanism does the reusable pattern actually describe?
- Which parts of the original practice must be preserved for the pattern to work?
- What details, such as exact wording, model choice, or local templates, should other users be free to adapt?
- What expertise, information, or verification capability does the pattern assume?
- Has anyone beyond the original user reproduced the benefit under different conditions?
- Which patterns are mature enough to share as candidates for reuse without presenting them as universal best practice?
Using Early User Feedback to Improve Adoption Support
- What kinds of help do experienced users still seek after basic orientation is no longer necessary?
- Which support materials are too generic, outdated, or disconnected from the work they are actually doing?
- What recurring questions should become easier to answer through self-service guidance or better examples?
- Where do users need coaching, peer discussion, or specialist expertise rather than additional formal training?
- What support can be reduced because capable users can now handle the issue independently?
- Which feedback reveals a tool, policy, integration, or management problem that support alone cannot solve?
- How should the support model change based on what early users actually need rather than what the organization initially expected?
Identifying What the Organization Should Change Based on Early Experience
- Which initial assumptions about GenAI adoption have early users shown to be wrong or incomplete?
- What access, tool, integration, or information problems should be corrected before broader adoption?
- Which usage boundaries or guidance are creating unnecessary uncertainty or failing to address real risk?
- What learning or support methods are producing capability, and which are producing little practical change?
- Which early practices deserve further replication before being spread more widely?
- What organizational friction is preventing demonstrated local value from translating into normal work?
- What should the organization change now before scaling the current adoption approach further?
Developing Everyday GenAI Habits
Moving From Occasional GenAI Use to Regular Use
- What causes users to remember GenAI only during training or unusual tasks rather than during normal work?
- Which recurring situations create enough value that people would naturally want to use GenAI again?
- What friction between recognizing an opportunity and beginning the interaction currently interrupts repeated use?
- Are users getting enough practical benefit from GenAI to justify changing an established working habit?
- What cues, integrations, saved context, or reusable methods could make returning to GenAI easier?
- Where would regular use be artificial because suitable GenAI opportunities simply do not occur often?
- What would show that repeated use is becoming a useful work habit rather than compliance with an adoption initiative?
Identifying Recurring Moments When GenAI Should Come to Mind
- Which recurring work situations share characteristics that make GenAI worth considering?
- When do employees repeatedly need to understand unfamiliar or complex information quickly?
- Where do people regularly generate alternatives, compare options, or challenge assumptions?
- Which moments involve turning rough, scattered, or unstructured material into something usable?
- Where does recurring preparation, review, or information synthesis create a natural GenAI cue?
- Which cues are broad enough to transfer between tasks without becoming a rigid list of use cases?
- What situations should employees learn to recognize even when they ultimately decide not to use GenAI?
Using GenAI During Normal Preparation Work
- What recurring activities does this person routinely prepare for before the main work begins?
- Where could GenAI help organize context, surface missing information, or generate useful questions beforehand?
- Which preparation tasks could benefit from alternatives, critique, or scenario thinking rather than simple drafting?
- What organizational context must be supplied before the preparation becomes genuinely useful?
- Which parts of preparation still require the user's own priorities, judgment, or stakeholder knowledge?
- What review is necessary before GenAI-generated preparation influences later work?
- What recurring preparation cue could make GenAI a natural option without turning it into a compulsory step?
Using GenAI During Recurring Writing and Communication
- Which writing or communication situations recur often enough for a reusable GenAI approach to matter?
- Where does repeatedly starting from a blank page consume unnecessary effort?
- Which communications require adaptation for different audiences, formats, or levels of detail?
- Where can GenAI improve structure or clarity while the user retains ownership of meaning and tone?
- Which communications depend too heavily on trust, sensitivity, authority, or personal judgment for routine delegation?
- What checking is needed before recurring GenAI-assisted communication is sent or published?
- How can a useful communication pattern become habitual without making every message sound standardized?
Using GenAI During Analysis and Decision Preparation
- Which recurring analyses involve gathering, structuring, comparing, or interpreting similar kinds of information?
- Where could GenAI help surface assumptions, alternatives, risks, or missing evidence?
- Which recurring decisions would benefit from better preparation even though the final judgment remains human?
- What criteria or authoritative information should guide the GenAI-supported analysis?
- Which facts or conclusions must remain independently verified?
- Where would habitual GenAI use improve analysis, and where might it create routine overreliance?
- What recurring decision-preparation cue should prompt consideration of GenAI support?
Using GenAI Before and After Meetings
- Which recurring meetings involve enough preparation or follow-up work for GenAI support to be worthwhile?
- What background, questions, objections, or options could GenAI help prepare before the meeting?
- How could GenAI organize permitted notes or discussion material into decisions, actions, and unresolved issues afterward?
- Which meeting information is too sensitive or restricted for the available GenAI environment?
- What should participants verify before relying on a GenAI-generated summary or action list?
- Where would GenAI reduce coordination effort without adding another parallel meeting process?
- What simple before-and-after routine could become useful without being imposed on meetings that do not need it?
Continuing Useful GenAI Conversations Instead of Starting Over
- Which recurring work benefits from preserving previous context, assumptions, sources, or decisions across interactions?
- When does continuing an existing conversation save the user from repeatedly rebuilding important context?
- What information should remain visible because it materially shapes later GenAI responses?
- When can old context begin to bias the interaction or make it difficult to tell what assumptions remain current?
- What parts of longer-running work are better stored in a persistent project, reusable context, or another structured resource?
- When should a user deliberately start fresh because the objective, context, or source material has changed?
- How can retained context reduce friction without turning outdated assumptions into invisible dependencies?
Reusing Successful GenAI Approaches in Similar Work
- What underlying work pattern made the previous GenAI approach successful?
- Which new situation is similar enough that reuse is reasonable?
- What context, criteria, or source material must be updated each time?
- Which part of the method is genuinely reusable, and which part depended on the original case?
- How can reuse reduce setup effort while preserving active judgment about whether the method still fits?
- What signs indicate that a familiar approach is being applied mechanically to unsuitable work?
- How can successful methods become easy to recall without turning into rigid prompt formulas?
Building Team Cues That Encourage Regular GenAI Consideration
- Which existing team routines could naturally prompt people to notice useful GenAI opportunities?
- What questions can managers or colleagues ask during normal work without creating a separate adoption process?
- Could project reviews, planning, or team discussions surface useful GenAI experiences and failures?
- How can recurring team cues focus on relevant work rather than pressure everyone to demonstrate AI usage?
- Which reminders would become annoying or performative rather than improving recognition?
- What team cues should disappear once people recognize opportunities independently?
- What lightweight shared practice would make GenAI easier to remember while preserving individual discretion about whether to use it?
Recognizing When GenAI Habits Are Beginning to Fade
- Which previously recurring GenAI uses are appearing less often or disappearing?
- Has the underlying work changed, or are users simply forgetting a still-useful practice?
- Has the perceived value declined because the method creates too much friction, review, or weak output?
- Are access, integration, policy, or manager-support problems interrupting a previously useful habit?
- Has a better non-GenAI method or newer GenAI approach made the old habit unnecessary?
- Which fading habits are worth reinforcing, and which should be allowed to disappear?
- What does declining use reveal about whether the practice ever created enough durable value to remain part of normal work?
Sharing Useful Practices Across Teams
Identifying GenAI Practices Worth Sharing
- Which GenAI practices have produced useful results repeatedly rather than in a single favorable example?
- What meaningful value does the practice create for the work?
- Can another competent user understand and reproduce the approach?
- What expertise, information, tools, or organizational conditions does the practice depend on?
- What failures, verification requirements, or limits need to be understood before reuse?
- Which successful practices are still too local, immature, or variable to promote broadly?
- What evidence makes a practice useful enough to share as a candidate for adaptation elsewhere?
Capturing Enough Context to Make an Example Understandable
- What work situation or problem caused the user to turn to GenAI?
- What outcome was the user trying to achieve?
- What source material, information, constraints, or examples were provided?
- How did the interaction develop after the initial response?
- What human expertise, correction, or judgment materially shaped the final result?
- What limitations, verification requirements, or failure cases appeared?
- What does another team need to know before deciding whether the example is relevant to its own work?
Explaining Why a GenAI Practice Was Useful
- What changed in the work because this practice was used?
- Did it improve time, quality, capacity, decision preparation, or enable work that was previously impractical?
- What effort would normally have been required without GenAI?
- How much of the result depended on the user's expertise or organizational context?
- Does the benefit remain after verification, correction, and interaction effort are included?
- What tradeoffs or new responsibilities did the practice introduce?
- Why is the underlying value mechanism potentially relevant to another team?
Showing How the Practice Worked in Real Work
- What was happening in the actual workflow when GenAI became useful?
- What did the user provide to GenAI at the beginning?
- How did the user refine, challenge, or redirect the interaction?
- What tools, sources, or organizational information were involved?
- What did the user verify or decide independently before using the result?
- How did the GenAI-supported output enter the rest of the real workflow?
- What would another user misunderstand if they saw only the final prompt or output?
Selecting the Right Channels for Sharing Practices
- Which people are likely to encounter the kind of work this practice supports?
- Where do those employees already exchange useful working knowledge?
- Does the practice need discussion and explanation, or can a searchable written example communicate it adequately?
- When would a live demonstration, team discussion, community, or reusable resource create the most learning?
- Which channels would bury useful practices among too much general GenAI content?
- How will employees find the practice later when the relevant situation actually occurs?
- What sharing channel fits both the maturity of the practice and the audience that could benefit from it?
Avoiding Presenting Local Examples as Universal Solutions
- Which parts of this practice depend on the original team's work, expertise, data, systems, or environment?
- What assumptions behind the success may not hold in another context?
- Does the practice require tools or organizational information that other teams do not possess?
- Which types of work would make the approach inappropriate or uneconomic?
- What limitations should be visible alongside the successful example?
- Which parts should other teams preserve, and which should they deliberately adapt?
- How can the example be framed as evidence to test rather than a company-wide solution to copy?
Helping Other Teams Adapt a Practice to Their Own Context
- What problem is the receiving team actually trying to solve?
- Does the same underlying value mechanism apply to its work?
- What information, tools, constraints, or quality standards differ from the original environment?
- Which parts of the shared practice are essential and which can be redesigned locally?
- What small test would show whether the adapted practice works under the new conditions?
- What verification or risk controls need to change for the receiving team's context?
- What would demonstrate that the adaptation creates value rather than simply reproducing another team's method?
Collecting Feedback From Teams That Reuse a Practice
- Did the shared practice create the expected benefit in the new environment?
- What had to be changed before it became useful?
- Which assumptions from the original example failed to transfer?
- What new limitations, risks, or better methods did the receiving team discover?
- Was the practice understandable enough for independent adaptation without help from its originator?
- What parts of the shared documentation should change based on the reuse experience?
- What does repeated reuse now tell us about which elements are genuinely transferable?
Updating Shared Practices When Better Approaches Emerge
- What capability, tool, workflow, or organizational condition has changed since the practice was documented?
- Does the original method still solve the problem efficiently under current GenAI capability?
- Have users discovered a simpler, more reliable, or easier-to-verify approach?
- Which instructions exist only because of limitations that no longer apply?
- Have new risks, integrations, or usage boundaries changed how the practice should work?
- Who should review and approve material changes to an important shared practice?
- How can users see clearly that an updated approach has replaced the previous one?
Removing Examples That Are No Longer Useful
- Which shared examples no longer reflect how effective GenAI-supported work is currently done?
- What examples depend on obsolete models, tools, restrictions, or workarounds?
- Which practices have been replaced by simpler or more reliable methods?
- What material is still being discovered by users even though it should no longer guide current work?
- Does an outdated example retain enough historical or learning value to archive rather than delete entirely?
- Who is responsible for deciding when a shared practice should be deprecated or retired?
- What should be removed from active guidance so obsolete knowledge does not compete with current practice?
Building Peer Support for GenAI Adoption
Identifying People Willing to Help Colleagues With GenAI
- Who is already helping colleagues solve practical GenAI problems informally?
- Which people have enough real-work experience to explain both useful methods and important limitations?
- Who can translate GenAI use into the language and work context of their colleagues?
- Which potential helpers are trusted because of their domain credibility rather than simply their technical enthusiasm?
- Are willing helpers distributed across roles and teams or concentrated in a small group?
- What responsibility, authority, or specialist knowledge should peers not be expected to provide?
- How can willing colleagues contribute to adoption without turning voluntary enthusiasm into an unlimited support obligation?
- Where do employees naturally turn when they encounter a small GenAI question during real work?
- Do users know which questions are appropriate to ask a colleague rather than a formal support function?
- What social or organizational factors make employees reluctant to admit that they need help?
- How can requesting practical help remain quick enough that people do not abandon the GenAI use instead?
- Which recurring questions could be answered through lightweight peer channels?
- What questions should immediately move to policy, technical, legal, or other specialist support?
- What simple help path would make peer support accessible without creating another administrative process?
Connecting Less Experienced Users With More Experienced Peers
- Which users would benefit most from seeing how an experienced colleague handles similar work?
- How important is matching by role, task, domain expertise, or GenAI experience?
- What should the less-experienced user learn to do independently rather than handing the task to the expert?
- When is a short demonstration enough, and when would repeated coaching be useful?
- How much peer-support time is reasonable before the underlying need should be addressed through another mechanism?
- What could make the relationship feel evaluative, burdensome, or socially uncomfortable?
- How can the connection build the user's capability so dependence on the more experienced peer decreases over time?
Creating Opportunities for Colleagues to Solve GenAI Problems Together
- Which recurring GenAI problems are difficult enough that comparing several approaches would be useful?
- What real work can colleagues examine together instead of discussing GenAI only in general terms?
- Where could different domain perspectives reveal why one approach succeeds and another fails?
- How can participants compare interaction methods without treating one person's prompt as the correct answer?
- What should the group learn about the task, tool, verification, or workflow after solving the problem?
- Which discoveries are important enough to capture for others?
- How can collaborative problem solving strengthen distributed capability rather than producing another recurring meeting with little practical value?
Using Team Discussions to Answer Common GenAI Questions
- Which GenAI questions are appearing repeatedly across several team members?
- What can the team resolve from its own practical experience and organizational guidance?
- Which answers require confirmation because they involve policy, risk, or specialist judgment?
- Could discussing one recurring issue prevent several separate help requests?
- What does the pattern of questions reveal about weaknesses in tools, guidance, or team methods?
- Which answer should be recorded so the team does not repeatedly rediscover it?
- How can GenAI discussion remain part of useful team learning without displacing more important operational work?
Encouraging People to Share Unfinished GenAI Work
- What useful learning remains hidden when colleagues share only polished GenAI successes?
- Which weak responses, failed interactions, or uncertain outputs would help others understand realistic GenAI use?
- How can people show where they became stuck without feeling that their capability is being evaluated?
- What kind of feedback would help the user diagnose the problem rather than simply replace their work?
- Which unfinished material contains confidential or sensitive information that should not be shared broadly?
- What can colleagues learn from seeing how another user iterates, verifies, or abandons an approach?
- How can sharing work in progress normalize responsible learning without reducing professional quality standards?
Connecting People Across Teams Who Face Similar GenAI Challenges
- Which teams are independently encountering the same GenAI problem or opportunity?
- What common mechanism links their challenges despite differences in local work?
- Who has already tested approaches that could save another team from repeating the same discovery process?
- Which context differences make direct reuse inappropriate?
- What lightweight connection would allow useful exchange without requiring a permanent formal community?
- Which recurring cross-team problems justify a broader discussion or reusable organizational solution?
- How can these connections reduce duplicate experimentation while preserving legitimate local adaptation?
Knowing When a Peer Question Needs Specialist Support
- Can this question be answered safely from established guidance and practical peer experience?
- Does it involve legal, privacy, security, regulatory, intellectual-property, or other formal interpretation?
- Is the problem actually technical enough to require system or product expertise?
- Could the answer affect consequential decisions, permissions, or production workflows beyond the local user?
- Does the repeated appearance of the question suggest an organizational problem rather than an individual support need?
- Does the peer know where and how to escalate without becoming the intermediary responsible for resolving it?
- How can escalation remain simple enough that users do not rely on informal guesses instead?
Avoiding Dependence on a Small Number of Helpful Individuals
- Which colleagues are currently receiving a disproportionate share of GenAI questions?
- Are users learning to solve problems independently, or simply asking the same expert to perform the GenAI work for them?
- What recurring knowledge held by a few helpers should become broadly accessible guidance or examples?
- Which common questions should managers, teams, or self-service resources be able to handle instead?
- How can more people develop enough capability to provide ordinary peer help?
- Is the amount of peer-support work visible and reasonable within helpers' normal responsibilities?
- What would make peer support resilient if the most enthusiastic or experienced individuals became unavailable?
Sustaining Peer Support as General GenAI Capability Improves
- Which basic peer-support needs should decline as users become more independent?
- What more advanced questions are emerging as employees attempt deeper GenAI-supported work?
- Where is peer learning still valuable because local experience changes faster than formal guidance?
- Which support activities can now be reduced or retired without leaving users stranded?
- Are experienced peers still spending time on issues that should have become self-service?
- How should peer support shift from basic tool help toward shared problem solving, workflow learning, and transfer of practices?
- What would show that peer support is evolving with adoption rather than becoming a permanent substitute for broader capability?
Building Team-Level GenAI Working Methods
Identifying Recurring Team Work That GenAI Can Support
- Which activities does the team perform repeatedly enough that a shared GenAI-supported method could create meaningful value?
- Where do several team members independently perform similar information, preparation, analysis, or communication work?
- Which recurring tasks create inconsistent quality or unnecessary reinvention across the team?
- What team decisions or handoffs depend on recurring preparation that GenAI could help improve?
- Which activities require enough common coordination that purely individual GenAI methods create problems?
- What work should remain individually flexible because different approaches produce equally good outcomes?
- Which recurring team activities provide the strongest case for developing a shared GenAI working method?
Comparing How Team Members Currently Use GenAI
- Which team members already use GenAI for the same or similar work?
- How do their approaches differ in context provision, interaction, verification, or tool choice?
- Which differences are associated with meaningfully better outcomes?
- What useful individual practices are currently invisible to the rest of the team?
- Which variations reflect legitimate personal working style rather than quality differences?
- Where does inconsistent GenAI use create downstream coordination or review problems?
- What can the team learn by comparing current methods before deciding whether anything should be standardized?
Identifying GenAI Approaches That Consistently Work Well
- Which GenAI approaches have produced useful results repeatedly across several real cases?
- Can more than one team member reproduce the benefit without depending on the original user's expertise?
- What information, tool access, or work conditions make the approach succeed?
- What verification or quality controls are necessary each time?
- Where does the method break when the case becomes unusual or more consequential?
- Which parts remain too variable to justify a shared team method?
- What evidence shows that an approach is reliable enough to become a candidate for team-wide use?
Agreeing on Shared GenAI Working Patterns
- Which parts of this recurring work benefit enough from consistency to justify a shared method?
- What outcome, inputs, boundaries, review requirements, or handoffs should the common pattern specify?
- Which conversational wording or personal interaction choices should remain flexible?
- What minimum quality and verification expectations should apply regardless of individual style?
- Do team members need to use the same tool, or only tools that meet defined capability and governance requirements?
- How simple can the shared method remain while still improving coordination and reliability?
- What team agreement would make work more dependable without freezing experimentation or personal judgment?
Developing Reusable Team-Level GenAI Resources
- What recurring team work would become easier with a reusable GenAI resource?
- Does the team need a prompt, example, checklist, context package, project, agent, template, or another type of reusable support?
- What organizational or task information can remain stable across cases?
- What context must still be supplied or judged afresh each time?
- How should the resource communicate verification, limitations, and situations where it should not be used?
- Who will keep the resource current as tools, models, and team practices change?
- How will team members know when the resource is appropriate rather than applying it automatically to every similar-looking task?
Integrating GenAI Into Team Preparation and Coordination
- What preparation or coordination work does the team repeatedly perform together?
- Where could GenAI consolidate information or identify open questions before a discussion or handoff?
- Could GenAI reduce repeated status reconstruction, preparation, or information translation between team members?
- What shared inputs need to remain accurate and current for the method to work?
- Which coordination decisions still require explicit human discussion rather than AI-generated synthesis?
- Where would introducing GenAI create another parallel process instead of simplifying the existing one?
- How can GenAI become part of the team's normal coordination workflow only where it removes real friction?
Defining How Important GenAI-Assisted Work Should Be Reviewed
- Which GenAI-assisted outputs need another person's review before they are used?
- How should the review level change with consequence, uncertainty, and detectability of error?
- What facts, sources, calculations, assumptions, or decisions must reviewers check explicitly?
- Who has the domain expertise and authority required to perform the review?
- What evidence should accompany the output so reviewers do not have to reconstruct how it was produced?
- Where would mandatory review consume enough effort to eliminate the value of using GenAI?
- What proportionate team review standard provides confidence without adding unnecessary control?
Building GenAI Into Team Handoffs and Shared Work
- Where does important context currently get lost when work passes between team members?
- Could GenAI help structure handoff material so the next person receives the information they actually need?
- Which shared documents, records, or workspaces could preserve useful context across the workflow?
- How can GenAI help distinguish completed work, unresolved questions, decisions, and required next actions?
- What human judgment or authorization must accompany the handoff rather than being inferred from AI-generated material?
- What confidentiality, access, or provenance requirements apply to shared GenAI-supported work?
- How could GenAI make the workflow more continuous without obscuring who owns each stage?
Updating Team Methods as Experience Grows
- What evidence suggests that the current team GenAI method is no longer the best way to perform the work?
- What have users learned from repeated success, failure, and exception handling since the method was introduced?
- Have new GenAI capabilities or integrations removed assumptions that shaped the original method?
- Which instructions or review steps can now be simplified without weakening quality?
- Where has excessive standardization become unnecessary as user capability improves?
- What new risks or failure patterns require stronger controls than before?
- What review process would let the team evolve its methods deliberately without changing them after every product update?
Avoiding Unnecessary Standardization of GenAI Use
- Which aspects of this work genuinely need consistency because they affect quality, safety, accountability, or coordination?
- Where can team members use different GenAI interaction styles and still produce equally acceptable outcomes?
- Are we standardizing something because variation causes a real problem or because a single method is easier to administer?
- Could a fixed prompt or procedure make the team slower to adopt better approaches?
- Which boundaries, inputs, verification rules, or handoffs need to be common even if interaction methods do not?
- Can a shared practice be offered as a recommended pattern without making it mandatory for every case?
- What is the minimum standardization needed to make the team's GenAI-supported work dependable?
Supporting Managers in Leading Adoption
Helping Managers Understand Their Role in GenAI Adoption
- Which adoption conditions can this manager influence directly through priorities, workload, expectations, coaching, and team routines?
- What responsibility does the manager have for connecting GenAI to the team's actual work?
- Which decisions about tools, governance, technical architecture, or formal policy belong elsewhere?
- How should the manager create practical permission for useful experimentation without pushing GenAI into unsuitable work?
- What should the manager notice and escalate when access, tooling, policy, or specialist support becomes the real constraint?
- How can the manager support quality and accountability without becoming the team's GenAI expert?
- What clear division of responsibility would help the manager lead locally without absorbing work that belongs to central or specialist functions?
Helping Managers Build Their Own GenAI Experience
- Which real management tasks would give this manager useful firsthand experience with GenAI?
- What should the manager learn about GenAI's strengths, limitations, and failure modes through personal use?
- Which interactions would help the manager understand why employees sometimes need iteration, context, or verification?
- How can the manager experience both useful results and realistic failures rather than only polished demonstrations?
- What level of capability is enough to recognize opportunities and discuss GenAI credibly with the team?
- Which outputs should the manager verify carefully so personal use also builds sound judgment?
- What evidence would show that the manager understands GenAI well enough to lead adoption without needing to become the most advanced user?
Preparing Managers to Discuss GenAI With Their Teams
- What should the manager be able to explain about why the organization is pursuing GenAI and what is expected now?
- Which parts of the message are established decisions, current expectations, or genuine uncertainties?
- What team-specific opportunities can the manager connect to recurring work rather than generic use cases?
- Which questions about quality, workload, roles, monitoring, or acceptable use are employees likely to raise?
- What should the manager answer directly, and what should be escalated to leadership or specialist functions?
- How can the manager avoid making promises about future roles, productivity, or staffing that have not been decided?
- What conversation would leave employees clearer about both useful experimentation and the boundaries within which it should happen?
Helping Managers Identify Opportunities in Team Work
- Which recurring team activities involve enough information, preparation, analysis, communication, or review to justify examining GenAI?
- Where does the team repeatedly encounter bottlenecks, rework, information overload, or work that is underdone because of limited capacity?
- Which opportunities would improve important work rather than merely increase visible GenAI use?
- What local knowledge from employees closest to the work should shape opportunity discovery?
- Which tasks remain poor candidates because they depend heavily on tacit context, authority, relationships, or difficult-to-detect errors?
- What small real-work experiment could test whether a promising opportunity creates enough value?
- Which opportunities should the manager pursue locally, and which require technical, governance, or cross-functional support before they can proceed?
Helping Managers Create Time for Experimentation
- Where can useful GenAI experimentation happen inside real work rather than as an additional side activity?
- Which lower-value activity could be reduced, delayed, or deprioritized to create legitimate learning time?
- How much repeated practice is needed before employees can fairly judge whether GenAI helps with the work?
- Which delivery pressures make experimentation unrealistic at the moment?
- How should the manager respond when learning initially slows work rather than producing immediate productivity gains?
- What experimentation would consume team capacity without addressing a meaningful opportunity?
- How can the manager protect enough learning time without turning GenAI experimentation into a permanent extra workload?
Supporting Managers in Coaching Less Confident Users
- What is actually limiting this employee: unfamiliarity, weak interaction skills, difficulty judging outputs, poor early experience, or something else?
- What real, low-consequence task could let the employee practice with enough relevance to build justified confidence?
- How can the manager help the employee diagnose a weak result rather than simply provide the correct prompt or answer?
- What questions would encourage the employee to add context, challenge the output, or try a different approach independently?
- When would a more experienced peer be better placed to provide practical help?
- What mistakes should be treated as normal learning, and which reveal that the task or tool is unsuitable?
- What would show that coaching is increasing the employee's independence rather than creating ongoing dependence on the manager?
Helping Managers Respond to Concerns and Resistance
- What specific concern is the employee expressing rather than labeling the behavior broadly as resistance?
- Is the issue relevance, access, confidence, workload, quality, policy uncertainty, professional responsibility, job change, or another cause?
- What evidence suggests the concern is based on a genuine problem in the work or adoption environment?
- Which questions can the manager answer from established organizational decisions, and which remain genuinely uncertain?
- What concern should the manager avoid dismissing simply because it slows adoption?
- What practical change would address the cause if the employee's reluctance is rational?
- How can the manager distinguish healthy skepticism, poor GenAI fit, and unresolved barriers from unwillingness to engage with a genuinely useful change?
Helping Managers Remove Team-Level Adoption Barriers
- What is currently preventing the team from using GenAI productively where relevant opportunities exist?
- Which barriers can the manager remove directly through priorities, time, expectations, coordination, or local working methods?
- Where are access, integration, or information problems making useful GenAI work unnecessarily difficult?
- Are employees hesitating because practical permission or quality expectations remain unclear?
- What workflow friction could the manager simplify without requiring a larger organizational change?
- Which barriers require escalation to technical, governance, leadership, or specialist functions?
- Which single change under the manager's control would remove the greatest amount of adoption friction for the team?
Helping Managers Reinforce Useful GenAI Behaviors
- Which employee behaviors are producing better work, stronger judgment, or reusable learning rather than merely more GenAI activity?
- How can the manager reinforce recognizing worthwhile opportunities instead of encouraging use on every task?
- What should the manager recognize when employees verify important outputs, expose failures, or decide not to use GenAI?
- How can useful recurring practices be made visible to colleagues without turning one person's method into a mandatory standard?
- What team conversations could reinforce learning from both successful and unsuccessful use?
- How can the manager avoid converting recognition into pressure to demonstrate visible GenAI usage?
- What normal management routines could help valuable behaviors persist after the novelty of the adoption effort fades?
Helping Managers Adapt Their Approach as Team Adoption Matures
- Which forms of manager support are still necessary, and which are becoming unnecessary as users gain independence?
- Where can the manager step back because employees now recognize opportunities and solve ordinary GenAI problems themselves?
- Which users or workflows are ready for deeper experimentation, integration, or shared team methods?
- What new issues are emerging around quality, workflow design, capacity, role changes, or coordination as use becomes more established?
- Which practices should now be handled through normal team routines rather than special adoption activity?
- When does the manager need to revisit an established practice because tools, capability, or work conditions have changed?
- How should the manager's role shift from encouraging initial use toward overseeing useful work, team methods, and adaptation?
Coordinating Adoption Across the Organization
Mapping GenAI Adoption Activities Across Teams
- What GenAI adoption activities are currently happening across functions, teams, and central support groups?
- Where are teams experimenting, developing learning resources, redesigning workflows, or building reusable solutions independently?
- Which initiatives depend on the same tools, data, specialists, or governance decisions?
- What local activity is invisible to the rest of the organization despite potentially useful learning?
- Where are several initiatives addressing the same problem through different approaches?
- What important adoption need appears to have no clear activity or owner?
- What organization-wide view would provide enough visibility to coordinate work without requiring central control over every local initiative?
Clarifying Ownership for Different Adoption Activities
- Who owns shared GenAI platforms, access, integrations, and major technical decisions?
- Who owns organization-wide usage boundaries, risk requirements, and specialist escalation?
- What should central adoption or learning functions provide rather than line managers or employees?
- Which opportunity, experimentation, and workflow decisions are best owned within functions or teams?
- Where are important responsibilities currently shared so ambiguously that decisions are delayed or duplicated?
- Which activities require organization-wide ownership even if execution remains distributed?
- What division of ownership would make it clear who decides, who supports, who executes, and who escalates each type of adoption issue?
Identifying Duplicate Adoption Efforts
- Which teams are independently creating similar training, examples, guidance, templates, integrations, or support mechanisms?
- Where are several groups repeatedly solving the same technical or governance problem?
- What experiments are being repeated because previous learning is difficult to discover?
- Which duplication is useful because different contexts genuinely require independent testing?
- What repeated work could become a shared service, resource, component, or knowledge asset?
- Where would centralizing a local activity create more overhead or reduce relevance?
- Which duplicate efforts should be combined, connected, or simply made visible to one another?
Identifying Dependencies Between Adoption Initiatives
- Which initiatives depend on usable tool access, data, integrations, or governance decisions before they can succeed?
- What learning activity assumes capabilities or permissions that are not yet available?
- Which scaling efforts depend on local practices that have not yet been replicated or validated?
- Where does workflow redesign depend on manager support, technical changes, or specialist input?
- Which shared initiatives are competing for the same scarce expertise or organizational attention?
- What work can progress independently without waiting for unrelated adoption activity?
- What sequence emerges when initiatives are ordered by their real dependencies rather than by a generic rollout plan?
Connecting Central Support With Local Team Needs
- What support is currently being designed or delivered centrally?
- Which recurring problems are teams actually encountering in their real work?
- Where is central support too generic to solve the practical issue teams face?
- Which local questions recur often enough that a shared organizational response would create value?
- What should teams and managers be expected to solve themselves as capability develops?
- How can evidence from local work reach the central function able to improve tools, guidance, learning, or policy?
- What feedback loop would keep central support responsive without making every local need a central project?
Balancing Organization-Wide Consistency With Local Flexibility
- Which elements need organization-wide consistency because variation would create security, legal, fairness, interoperability, or coordination problems?
- What common platform, data, quality, or accountability requirements should apply across teams?
- Which opportunity, workflow, interaction, or learning decisions depend strongly enough on local context to remain flexible?
- Where would local freedom create unnecessary duplication or incompatible solutions?
- Where would a central standard prevent teams from discovering better methods for their own work?
- Can shared principles or interfaces provide enough coherence without prescribing the complete working method?
- What is the minimum level of consistency needed to protect the organization while preserving useful local agency?
Sharing Adoption Learning Across Functions
- Which discoveries from one function are likely to help people dealing with similar work elsewhere?
- What information about the situation, context, value, failures, and verification is needed before another function can interpret the lesson?
- How can unsuccessful experiments and limitations travel across boundaries as readily as polished successes?
- Which underlying mechanisms are transferable even when the exact local implementation is not?
- What learning should remain local because the work, risk, or context differs too much?
- Which existing knowledge-sharing mechanisms can carry GenAI learning without creating unnecessary new meetings or forums?
- How can cross-functional learning expand other teams' options without turning an early practice into an organization-wide standard prematurely?
Coordinating Limited Adoption Resources
- Which specialist skills, funding, licenses, integration capacity, and management attention are currently scarce?
- What requests genuinely require those scarce resources rather than peer, manager, or self-service support?
- Which opportunity would create the greatest value or unblock the largest number of other initiatives?
- Where would specialist involvement create a reusable component or learning benefit for several teams?
- Which requesting teams are sufficiently ready to own and sustain the work after central support ends?
- What activities should be delayed because their dependencies or expected value do not justify scarce resources yet?
- What transparent allocation criteria would direct limited support toward organizational leverage rather than visibility, seniority, or first-come demand?
Resolving Conflicts Between Different Adoption Approaches
- Where are teams adopting approaches that cannot coexist easily across shared systems, workflows, or governance boundaries?
- What legitimate differences in work or context explain each local approach?
- Which variation is harmless and does not require an organization-wide resolution?
- Where does inconsistency create real risk, unfairness, confusion, duplication, or interoperability problems?
- What evidence exists about the relative strengths and limitations of the competing approaches?
- Can a shared outcome, interface, or boundary allow several local methods to continue?
- Which conflicts require a central decision because local variation would impose costs on the rest of the organization?
Maintaining a Coherent Organization-Wide Adoption Direction
- Do current GenAI initiatives still contribute to the organization's intended adoption outcome?
- Are functions making locally sensible decisions that nevertheless conflict when viewed across the organization?
- Where has adoption become fragmented across tools, messages, metrics, governance, learning, or workflow initiatives?
- Which local innovations reveal that the broader adoption approach itself should change?
- What common purpose, decision rights, and core boundaries should remain stable across different local approaches?
- Which established activities no longer contribute enough value to remain part of the adoption system?
- How can the organization preserve a coherent direction while allowing the methods used to reach it to evolve with evidence?
Measuring Ongoing GenAI Use
Defining What Meaningful GenAI Use Should Be Measured
- Which dimensions of adoption matter for the decisions we need to make: access, experimentation, recurrence, breadth, depth, independence, or work integration?
- What kinds of GenAI interaction count as meaningful professional use rather than incidental tool activity?
- Which measures reflect the intended adoption outcome rather than whatever telemetry is easiest to collect?
- How should measurement account for roles where valuable GenAI opportunities occur only occasionally?
- What usage metrics could encourage employees to create visible activity instead of useful work?
- What qualitative evidence is needed to understand what the numbers cannot show?
- What measurement definition would help us diagnose adoption accurately without turning GenAI usage itself into the objective?
- How many people have genuinely usable access rather than merely an assigned license?
- How many people with usable access have actually tried GenAI on real work?
- How many continue using it after initial experimentation?
- Where is the gap between access and use largest, and what explanations are plausible?
- What use may be occurring through other approved, embedded, personal, or unobserved tools that the access data do not capture?
- Where could high license coverage create a misleading impression that adoption is already established?
- What evidence beyond entitlement data is needed before leadership draws conclusions about actual use?
Measuring How Frequently People Use GenAI
- How often are employees using GenAI for real work over a meaningful period?
- Which users return daily, weekly, occasionally, or only after isolated experiments?
- What recurring work situations are responsible for the observed frequency?
- How much variation in frequency is explained by differences in the work itself?
- Where is frequent use concentrated in shallow or low-value interactions?
- Where is infrequent use entirely appropriate because the relevant work occurs rarely?
- What can frequency tell us about recurrence while avoiding the assumption that more interactions automatically mean better adoption?
Understanding How Many People Use GenAI Regularly
- What pattern of recurring use should count as regular for the kinds of work we are studying?
- How many users return to GenAI when relevant situations recur rather than according to an arbitrary activity threshold?
- Is recurring use spreading beyond the original enthusiasts and early adopters?
- Which groups show stable recurrence, and which continue cycling between experimentation and inactivity?
- Do regular users rely on GenAI for valuable work or mainly for easy, low-consequence tasks?
- What conditions distinguish regular users from occasional users performing comparable work?
- What does the size and composition of the regular-user population reveal about whether useful habits are becoming established?
Measuring the Breadth of Work Supported by GenAI
- How many different recurring work situations are employees using GenAI to support?
- Is use concentrated in a narrow set of obvious activities such as drafting and summarization?
- Where is GenAI beginning to support analysis, comparison, preparation, review, research, or multi-step work?
- Which high-potential work patterns remain largely untouched?
- How does appropriate breadth differ between roles whose work contains very different GenAI opportunities?
- Are users expanding into new activities because their capability is increasing or because they feel pressure to use GenAI more broadly?
- What does the range of meaningful uses reveal about how far adoption has progressed beyond basic experimentation?
Understanding Whether People Return to GenAI After Initial Use
- What proportion of first-time users return to GenAI for another real work situation?
- How soon and under what circumstances do they return?
- Where does use drop sharply after an initial experiment?
- What successful experiences appear to create recurring voluntary use?
- What poor results, friction, policy uncertainty, or lack of relevance causes users not to return?
- How do return patterns differ between groups facing comparable work opportunities?
- What does recurrence tell us about perceived usefulness that first-use or license data cannot reveal?
Comparing GenAI Use Across Teams
- How do teams differ in recurrence, breadth, depth, independence, and workflow integration rather than simply interaction volume?
- Are the teams being compared similar enough in their work for the differences to be meaningful?
- How much do access, integrations, manager support, workload, and local learning conditions explain the variation?
- Are high-use teams also obtaining better work outcomes?
- Which practices or environmental conditions distinguish teams whose use appears more meaningful?
- Where would comparing raw adoption levels create unfair or misleading conclusions because GenAI opportunity differs substantially?
- What lessons from team differences deserve further investigation before they are generalized elsewhere?
Tracking How GenAI Use Changes Over Time
- Is adoption changing because more employees are starting, existing users are returning more often, or work integration is deepening?
- Which teams or user groups are accelerating, plateauing, or declining?
- Are users expanding from basic assistance into more substantial forms of work?
- Which previously common practices are disappearing, and is that a problem or a sign that better methods have emerged?
- What product changes, access decisions, training, management actions, or workflow changes coincide with important shifts?
- Are usage trends sustained long enough to distinguish durable change from temporary campaigns or novelty?
- Which trend matters most for the next adoption decision, and what additional evidence is needed to interpret it?
Combining Usage Data With Qualitative Evidence
- What can the available telemetry establish with reasonable confidence?
- What important aspects of purpose, quality, value, confidence, and workflow integration remain invisible in the logs?
- What are employees and managers reporting about why GenAI is or is not being used?
- What do actual work examples reveal that aggregated interaction counts cannot?
- Where does qualitative evidence contradict the apparent story told by telemetry?
- What could explain those disagreements, including shadow use, compelled use, weak tool fit, or different work opportunities?
- What combined interpretation is best supported when telemetry, surveys, interviews, manager observations, and work evidence are considered together?
Measuring How Deeply GenAI Is Integrated Into Recurring Work
- Which recurring tasks now include GenAI as a normal part of the work rather than an occasional experiment?
- Do employees recognize the relevant situation and use GenAI without external prompting?
- Has the workflow itself changed, or has GenAI simply been added as an optional side activity?
- How much does the practice depend on one enthusiast, champion, or manual workaround?
- Are GenAI outputs connected to the systems, handoffs, and quality controls that surround the work?
- Where has integration become unnecessary because users discovered that GenAI adds little value?
- What evidence would show deepening integration without creating a target that employees must use GenAI in every eligible situation?
Assessing Whether GenAI Is Creating Value
Defining What Value From GenAI Should Mean
- What organizational outcome is this GenAI use expected to improve?
- Is the expected value lower effort, better quality, faster completion, greater capacity, stronger decisions, customer benefit, or newly enabled work?
- What benefit matters to the people performing the work as well as to the organization?
- Which outcomes can be measured directly, and which require credible qualitative evidence?
- What downstream effect must occur before an individual productivity improvement becomes organizational value?
- Which indicators, such as usage volume or output quantity, should not be treated as value by themselves?
- What definition of value is specific enough to test whether continuing or expanding the GenAI use is worthwhile?
Establishing a Baseline for Comparison
- How is this work currently performed without the GenAI-supported method?
- What time, effort, cost, quality, throughput, or other relevant outcome is typical today?
- How much normal variation exists between cases, users, or periods?
- What baseline evidence already exists in systems, work records, or quality reviews?
- Where will a carefully constructed estimate be more realistic than precise historical data?
- What other changes are occurring that could make a simple before-and-after comparison misleading?
- What comparison point is credible enough to judge whether GenAI has produced a meaningful improvement?
Assessing Time and Effort Saved Through GenAI
- How much total employee effort does the work require with GenAI compared with the previous method?
- Does the estimate include context preparation, interaction, verification, correction, and exception handling?
- How consistent are the savings across users and realistic cases?
- Does the GenAI-supported process create new work for reviewers, colleagues, or downstream teams?
- Is the time saved concentrated enough to be usable, or fragmented into small intervals that are difficult to redeploy?
- What actually happens to the released time after the task becomes faster?
- Does the net reduction in effort create enough usable capacity to matter organizationally?
Assessing Whether GenAI Improves Quality
- Which dimensions of quality matter most for this work?
- Does GenAI reduce important omissions, errors, inconsistency, or variation between users?
- Does it improve completeness, structure, evidence use, clarity, or another relevant aspect of the output?
- What new quality failures does GenAI introduce that were less common in the previous method?
- How do qualified reviewers assess GenAI-assisted work compared with an appropriate baseline?
- Does the quality improvement persist across different users, cases, and normal work conditions?
- Is the improvement large and dependable enough to justify the additional GenAI interaction and review required?
Assessing Whether GenAI Improves Speed
- How much sooner is the complete piece of work finished when GenAI is used?
- Which specific stages become faster and which remain unchanged?
- How much time is added back through context preparation, review, correction, or escalation?
- Does faster completion of one step reduce the end-to-end cycle time?
- Are downstream users actually receiving what they need sooner, or does the work simply wait at the next bottleneck?
- Could higher production speed create more volume, review burden, or coordination without improving the final outcome?
- What amount of cycle-time improvement would be operationally meaningful enough to count as value?
Assessing Whether GenAI Improves Decision Preparation
- Does GenAI help decision-makers receive more complete or better-organized evidence?
- Are relevant alternatives, assumptions, tradeoffs, and risks surfaced more consistently?
- Does the preparation become faster without sacrificing reliability?
- Are decision-makers receiving information that is easier to interrogate and use?
- What errors, unsupported claims, or biases could GenAI introduce into the preparation?
- Can qualified decision-makers distinguish an improvement in preparation from a more polished presentation?
- What evidence shows that GenAI is strengthening the inputs to human judgment rather than simply increasing the amount of material produced?
Assessing Whether GenAI Enables Work That Was Previously Impractical
- What valuable activity is now being performed that was previously omitted, restricted, or uneconomic?
- What made the work impractical before GenAI was available?
- How much does GenAI reduce the expertise, effort, time, or cost required?
- What measurable or observable value does the newly enabled work create?
- Can the activity be repeated reliably across enough cases to justify continued attention?
- What verification, infrastructure, or additional human work is needed to sustain it?
- Is the new capability important enough that it should become part of normal operations rather than remain an occasional experiment?
Comparing GenAI Benefits With the Effort and Cost Required
- What benefits have been demonstrated strongly enough to include in the value assessment?
- What licensing, infrastructure, integration, model, or vendor costs are required?
- How much user, manager, specialist, and support time does the GenAI-supported method consume?
- What verification, correction, monitoring, and maintenance burden remains?
- Do the economics improve as users gain capability or the practice becomes better integrated?
- What risks or opportunity costs should be considered alongside the direct financial costs?
- Does the total benefit justify continuing, expanding, redesigning, or stopping the GenAI-supported activity?
Distinguishing GenAI's Contribution From Other Changes
- What process, staffing, workload, training, management, or technology changes occurred alongside GenAI adoption?
- Could any of those changes plausibly explain part of the observed improvement?
- Do we have a comparison group, staggered rollout, repeated measure, or other evidence that strengthens attribution?
- Is the measured change close enough to the GenAI-supported workflow to make the causal link credible?
- Which part of the improvement can reasonably be attributed to GenAI and which remains uncertain?
- How much attribution rigor is appropriate given the size and consequence of the investment decision?
- How should the result be communicated so the organization neither dismisses real benefit nor overstates causality?
Building an Overall View of Value Created Through GenAI
- Which GenAI-supported activities show the strongest evidence of meaningful value?
- What kinds of value appear most consistently across teams and use cases?
- Which high-usage practices create little observable improvement in the underlying work?
- Where do benefits differ because of user expertise, workflow conditions, manager support, or available integrations?
- What individual time savings have actually been converted into organizational capacity or outcomes?
- Which claimed benefits remain supported mainly by anecdotes, perceptions, or weak attribution?
- What balanced conclusion should leadership draw about where GenAI is creating value, where evidence is still developing, and where continued investment is difficult to justify?
Diagnosing Resistance and Low Adoption
Identifying Where Adoption Is Lower Than Expected
- Which teams, roles, or user groups show materially less adoption than the intended outcome would suggest?
- Is the gap in initial experimentation, recurrence, breadth, depth, independence, or integration?
- Has adoption always been low, or did users experiment and later disengage?
- What work opportunities actually exist for the affected group?
- Are we comparing their behavior with a realistic target for the work they perform?
- Could apparently low adoption represent rational non-use rather than a problem?
- Where is the gap significant enough that understanding its cause would change an organizational decision?
Distinguishing Lack of Access From Lack of Use
- Does the affected group have genuinely usable access to a capable approved GenAI tool?
- Can employees reach the features, organizational information, and integrations needed for relevant work?
- Are authentication, account, device, permission, or capacity problems creating hidden barriers?
- Which employees have practical access but still do not experiment or return?
- What unobserved use may be happening through other tools or personal accounts?
- Are we treating license assignment as proof that the environment is ready for adoption?
- How much of the apparent adoption gap remains after practical access problems have been accounted for?
Understanding Whether People See Enough Personal Relevance
- Can employees identify recurring situations in their actual work where GenAI could provide meaningful help?
- Are existing examples close enough to their responsibilities to make the opportunity recognizable?
- Do users still think of GenAI mainly as a drafting or summarization tool?
- Have they tested GenAI on work that genuinely matters to them rather than on generic training exercises?
- Does the role contain enough suitable GenAI opportunity to justify more frequent use?
- Are employees correctly concluding that verification, context requirements, or weak task fit make GenAI unattractive?
- What evidence would distinguish weak opportunity recognition from an actual lack of relevant opportunity?
Identifying Skill or Confidence Problems
- Can users describe their work and provide enough context to begin a useful GenAI interaction?
- Can they improve a weak response instead of abandoning the attempt?
- Do they know how to judge quality, verify important information, and recognize the limits of their own expertise?
- Are users avoiding GenAI because previous failures lowered their expectations?
- Is confidence lower than actual capability, or is confidence high despite weak practical judgment?
- Which specific capability gap is preventing useful independent work?
- What evidence would show that support aimed at skill or confidence is addressing the true constraint rather than a different adoption problem?
Understanding Whether Usage Boundaries Are Discouraging Use
- Do employees understand which tools, information, purposes, outputs, and actions are allowed?
- Where do users perceive the rules as broader or more restrictive than they actually are?
- Which recurring situations remain ambiguous enough that employees prefer not to use GenAI?
- How difficult is it to get an answer when ordinary guidance does not resolve the case?
- Which restrictions genuinely prevent a valuable use because of legitimate risk?
- What questions recur often enough to show that the guidance itself is not sufficiently usable?
- How much low adoption would remain if employees could make normal low-risk usage decisions confidently without escalation?
Identifying Poor Early Experiences With GenAI
- What happened when disengaged users first tried GenAI on real work?
- Was the initial task suitable for a beginner and for the available technology?
- Did missing context, unrealistic expectations, or weak tool capability produce the poor result?
- Did users encounter plausible errors that damaged trust more broadly?
- Was anyone available to help them understand why the interaction failed?
- Have later improvements in tools or capability made the original experience obsolete?
- What new experience would test whether the early failure still represents the current reality?
Understanding How Managers Influence Low Adoption
- Do employees receive practical permission from their manager to experiment with GenAI during normal work?
- Is learning time protected, or does every experiment compete with unchanged delivery expectations?
- Can managers connect GenAI to relevant team work rather than simply encouraging generic usage?
- How do managers respond when an experiment fails or an employee raises a legitimate concern?
- Are local performance expectations rewarding short-term output in ways that discourage learning or workflow change?
- Is the manager's own uncertainty, skepticism, or enthusiasm distorting the team's adoption behavior?
- What evidence suggests that changing managerial conditions would materially change adoption in this team?
Identifying Workflow Friction That Makes GenAI Inconvenient
- How many extra steps does an employee need to take to use GenAI in the relevant workflow?
- Does the user need to copy, reformat, or manually reconstruct context that already exists elsewhere?
- Can the approved tool access the systems and information the task requires?
- How much verification or correction is added after the GenAI interaction?
- Does GenAI speed one activity while leaving the wider workflow or downstream bottleneck unchanged?
- Is the current manual method simply easier and more dependable under real conditions?
- What workflow or integration change would have to occur before recurring GenAI use becomes rational?
Identifying Organizational Policies or Structures That Discourage Adoption
- Which policies or approval processes create unnecessary friction for ordinary legitimate GenAI use?
- Where do organizational boundaries make support, information, or specialist expertise difficult to access?
- Are employees being asked to experiment while performance systems reward only immediate delivery?
- Which resource or budget arrangements prevent teams from acting on demonstrated opportunities?
- Where do conflicting messages from leadership, governance, technology, and local management create hesitation?
- Which adoption problems would persist even if employees were highly capable and motivated GenAI users?
- What organizational condition needs to change before individual adoption interventions can reasonably be expected to work?
Determining the Main Causes Behind Resistance or Low Adoption
- What competing explanations for low adoption are supported by the evidence?
- Which observed behavior is merely a symptom, and what underlying condition best explains it?
- How do access, relevance, capability, confidence, tool fit, policy, workflow, workload, and managerial conditions interact?
- Which causes affect many users, and which are specific to particular groups or kinds of work?
- Where is apparent resistance actually rational non-use under the current conditions?
- What evidence would falsify our preferred explanation before we choose an intervention?
- Which cause should be addressed first because changing it is most likely to alter meaningful adoption?
Reinforcing Successful Adoption Behaviors
Identifying the GenAI Behaviors Worth Reinforcing
- Which GenAI behaviors consistently improve the quality, speed, capacity, or usefulness of real work?
- Are users recognizing worthwhile GenAI opportunities independently rather than simply increasing how often they use the tool?
- Do employees provide relevant context, iterate when needed, and evaluate outputs critically before using them?
- Are people verifying important information and retaining human judgment where consequences justify it?
- Which behaviors help successful methods become reusable or contribute useful learning to others?
- What visible GenAI activity creates little value or should actively not be encouraged?
- Which few behaviors would most strengthen durable adoption if they became normal practice?
Recognizing People Who Apply GenAI Usefully
- What did this person accomplish through GenAI-supported work that created genuine value?
- Did they show good judgment about when GenAI was useful and when it was not?
- How well did they evaluate, verify, or correct the output before relying on it?
- Did they improve a recurring way of working rather than produce only an impressive one-off result?
- Did their experience generate useful learning that colleagues or the organization can reuse?
- How can recognition avoid rewarding interaction volume, novelty, or visible enthusiasm instead of sound practice?
- What form of recognition would reinforce valuable behavior without turning GenAI use into a competition?
Reinforcing Opportunity Recognition During Everyday Work
- Which recurring situations should increasingly cause employees to consider whether GenAI could help?
- Are users recognizing underlying patterns such as information overload, comparison, preparation, or uncertainty across different tasks?
- What existing planning, review, or management routines can reinforce these cues naturally?
- Can managers ask questions that improve opportunity recognition without implying that GenAI should always be used?
- Are employees beginning to transfer successful patterns to unfamiliar work situations independently?
- Which reminders can now be reduced because the recognition habit has become established?
- How can the organization reinforce consideration of GenAI while preserving the judgment to decide that it adds little value?
Encouraging Managers to Discuss Successful GenAI Use
- Which successful GenAI-supported practices are useful enough for managers to surface during normal team conversations?
- What work problem did each example solve, and why did the approach create value?
- What human expertise, context, verification, or judgment made the practice successful?
- Which limitations or failures should be discussed alongside the successful result?
- What part of the practice could colleagues reasonably adapt to their own work?
- How can managers discuss successful use without turning team meetings into promotional showcases for GenAI?
- What regular conversation would help useful learning circulate without creating a separate adoption process?
Building GenAI Into Existing Team Routines
- Which established team routines could reinforce useful GenAI behavior without adding new meetings or bureaucracy?
- Where could GenAI naturally support recurring planning, preparation, review, knowledge sharing, or follow-up?
- Which existing routine could include reflection on useful GenAI practices or lessons from failed experiments?
- What small change would make a proven GenAI practice easier to repeat at the right moment?
- Which team routines should remain unchanged because GenAI would add little value?
- How much consistency is useful without turning individual interaction methods into rigid procedures?
- What existing routine could sustain a valuable practice after dedicated adoption attention declines?
Sharing New Examples That Keep GenAI Relevant
- What newly emerging GenAI uses reflect capabilities or working methods employees may not yet recognize?
- Which user groups need fresh examples because their existing examples no longer match the work they are trying to improve?
- Does the example solve a recognizable problem rather than simply showcase a new feature?
- What new capability or working pattern does the example help employees understand?
- What limitations, verification needs, or prerequisites should be shown alongside it?
- Which examples remain too experimental or context-dependent to promote broadly?
- How can new examples refresh opportunity recognition without creating hype around every capability change?
Refreshing Support Material as Users Become More Capable
- Which basic guidance no longer helps experienced users and can be simplified or removed?
- What more advanced questions are appearing as employees attempt deeper GenAI-supported work?
- Which examples, prompts, or instructions were designed for earlier capability levels and now need replacement?
- Where can support become less prescriptive because users have developed stronger independent judgment?
- What guidance remains essential for new or less-experienced users?
- How should materials distinguish enduring principles from tool-specific techniques that may change quickly?
- What support structure allows content to evolve with user capability without continuously expanding in size?
Making Successful GenAI Behaviors More Visible in Everyday Work
- Where are employees using GenAI effectively without colleagues having any opportunity to observe how they work?
- Which parts of successful GenAI use are worth making visible beyond the final output?
- How can teams see useful examples of context setting, iteration, verification, or deciding not to use GenAI?
- What existing work forums can surface these behaviors without requiring formal demonstrations?
- How can employees share their methods voluntarily without feeling monitored or evaluated?
- What would make visibility feel performative and encourage superficial GenAI behavior?
- How can observation of real colleagues help useful working behaviors spread naturally?
Sustaining GenAI Use After Initial Enthusiasm Fades
- Which GenAI uses continue because they solve recurring problems rather than because the technology is still novel?
- Where has usage declined because employees no longer see enough practical value?
- Have access, workflow, manager support, or quality problems reappeared as the initial adoption effort faded?
- Which established examples or working methods have become stale as tools and capabilities changed?
- What recurring work situations can sustain GenAI consideration without repeated campaigns?
- Which support activities can now be reduced, and which need refreshing because the underlying need remains?
- What would show that GenAI has moved from novelty-driven activity to durable usefulness?
Keeping Successful Adoption Behaviors Active Over Time
- Which valuable GenAI behaviors have become stable parts of normal work?
- Where are old working habits returning even though the newer method remains demonstrably better?
- What organizational routines, team methods, or manager behaviors currently reinforce the successful practice?
- Which practices need periodic review because tools, workflows, or capability assumptions may change?
- Are employees still applying judgment and verification, or has a once-careful practice become automatic and less reliable?
- When should an established behavior be updated or retired rather than reinforced further?
- What will keep valuable GenAI-supported work effective and adaptable rather than merely persistent?
Scaling Successful Adoption Patterns
Identifying Adoption Patterns That Are Ready to Scale
- Which GenAI-supported practices have produced useful results repeatedly across realistic cases?
- Has the pattern worked for more than one user rather than depending on its original inventor?
- What evidence shows that the benefit is stable enough to test in another environment?
- Which tools, information, expertise, manager support, or workflow conditions does the pattern require?
- Are its failure modes understood well enough for another team to detect and handle them?
- Is the required support and verification burden manageable beyond the original setting?
- Which patterns are mature enough for deliberate replication without yet assuming organization-wide scalability?
Confirming That a Successful Pattern Produces Real Value
- What specific value does the pattern create in the work?
- Does the benefit remain after context preparation, verification, correction, and support are included?
- Is the improvement repeated across enough cases to rule out a favorable one-off result?
- Does it improve an outcome that matters, such as quality, cycle time, capacity, decisions, or newly enabled work?
- What baseline or comparison supports the value claim?
- What hidden costs or downstream burdens could weaken the apparent benefit?
- Is the demonstrated value strong enough to justify the effort and risk of testing the pattern elsewhere?
Distinguishing the Transferable Core From Local Details
- What underlying mechanism makes this practice successful?
- Which parts depend on the original team's specific data, systems, role design, expertise, or workflow?
- What information, tool access, or verification capability is genuinely essential?
- Which exact prompts, templates, or interaction choices are incidental rather than necessary?
- What aspects can a receiving team safely adapt without losing the value mechanism?
- What conditions must remain consistent for the pattern to work?
- What is the simplest description of the transferable core that another team could test intelligently?
Identifying Other Teams That Could Benefit From the Pattern
- Which teams face the same underlying work problem even if their tasks look different on the surface?
- Where does the same value mechanism plausibly apply?
- Which teams already have the tools, information, expertise, and manager support the pattern requires?
- What differences in quality standards, regulation, workflow, or user capability could prevent transfer?
- Which teams have enough recurring demand for the pattern to matter?
- Who is ready to test the practice without requiring disproportionate central support?
- Which next environment would provide the strongest test of whether the pattern genuinely transfers?
Adapting the Pattern to a New Team or Function
- What is materially different about the receiving team's work environment?
- Which part of the original pattern still solves the same underlying problem?
- What local examples, context, tools, or source material need to change?
- Which new constraints or risks affect how the practice should work here?
- What additional support or capability does the receiving team need before testing it?
- How can the team adapt the method without removing the mechanism that made it successful?
- What outcome would show that the practice has been successfully adapted rather than merely copied?
Preparing Support for the First New Adopters
- What do new adopters need to understand about why the pattern works?
- Which example best shows the practice without encouraging blind replication of the original implementation?
- What access, information, tools, or permissions must be ready before they begin?
- Who can answer practical questions during the first attempts?
- Which common mistakes, limitations, or failure conditions should users expect?
- How much structured support is necessary before users can operate the practice independently?
- What would help the receiving team stop depending on the original team as quickly as reasonable?
Testing the Pattern in a New Environment
- What assumptions about transferability are we actually testing in the new team?
- Which elements should remain consistent with the original pattern so the comparison is meaningful?
- What local adaptations should be allowed from the beginning?
- Which value, quality, reliability, and support outcomes should be observed?
- What failure would indicate that an essential condition from the original environment is missing?
- How much realistic variation in cases and users should the test include?
- What evidence should determine whether the pattern is refined, expanded, or stopped?
Comparing Results Across Different Teams
- Does the pattern create comparable value across the teams that have tested it?
- What differences in results appear between users, workflows, or environments?
- Which local conditions best explain those differences?
- How much support, verification, or adaptation does each team require?
- Which elements of the pattern remain consistently useful across settings?
- Where has local adaptation improved the original method rather than weakened it?
- What does the accumulated evidence now suggest about how broadly the pattern can reasonably scale?
Documenting What Is Needed for Reliable Reuse
- What prerequisites should another team confirm before trying this pattern?
- What part of the method is essential to producing the intended value?
- Which implementation details are optional or context-dependent?
- What verification, quality, exception, or accountability controls need to be preserved?
- What common failure modes should users be prepared to recognize?
- What example would make the pattern understandable without turning the documentation into a rigid recipe?
- What is the minimum documentation needed for reliable reuse without creating a lengthy manual that quickly becomes outdated?
Expanding the Pattern After Successful Replication
- What evidence now supports moving beyond limited replication?
- Which additional teams or work environments are the strongest next candidates?
- How much central, specialist, or peer support would broader expansion require?
- What parts of adoption can now become self-service because the pattern is sufficiently understood?
- Which risks, coordination problems, or verification burdens could increase materially with scale?
- How will learning from later adopters continue to update the pattern rather than freeze the first successful version?
- What evidence should cause expansion to pause even after earlier replications succeeded?
Identifying Workflows That Should Change
Identifying Workflows Where GenAI Is Only Being Added on Top
- Where has GenAI become an additional step without changing the surrounding workflow?
- Which manual steps remain because the process was designed before GenAI existed?
- Are employees performing the old task and then repeating or checking much of the same work after GenAI?
- Which approvals or review steps remain even though their original purpose may no longer apply?
- Does GenAI accelerate one activity while leaving queues, handoffs, or bottlenecks elsewhere unchanged?
- What new work has been added for context preparation, verification, or coordination?
- What would we question if we redesigned this workflow from the outcome backward instead of inserting GenAI into the existing sequence?
- Where do employees repeatedly read, classify, extract, compare, summarize, or transform large amounts of information?
- Which information-processing steps follow recognizable patterns across many cases?
- Where are people manually moving or reformatting information between systems?
- What parts of the processing require genuine human judgment rather than routine cognitive effort?
- Can current GenAI perform enough of the repetitive work reliably to change the workflow?
- What verification burden would remain after more processing is delegated?
- Which information-heavy workflows offer enough recurring value to justify deeper redesign?
Identifying Repeated Handoffs and Rework
- Where does work pass between more people or teams than the outcome appears to require?
- What context is repeatedly lost or reconstructed during handoffs?
- Which outputs are rewritten, reformatted, or reanalyzed before the next person can use them?
- What causes the repeated clarification or rework?
- Could GenAI prepare more complete handoff material or preserve context across stages?
- Which handoffs exist because of legitimate authority or expertise rather than historical process design?
- What handoff or rework pattern should be reconsidered if GenAI changes how information can be prepared or transferred?
Identifying Workflows With Slow Decision Preparation
- Which important decisions are delayed mainly by gathering, organizing, or analyzing information?
- Where does decision preparation require repeated manual work across similar cases?
- Which evidence or analysis frequently arrives too late to influence the decision?
- Could GenAI improve the speed or completeness of preparation without weakening verification?
- What assumptions, alternatives, or risks could be surfaced earlier?
- Which decision authority and human judgment should remain unchanged regardless of faster preparation?
- Which decision workflows would benefit enough from better preparation to justify redesign?
Finding Communication Work That Could Be Simplified
- Where does the workflow require the same underlying information to be repeatedly rewritten for different audiences?
- Which recurring messages consume substantial effort without adding much new judgment?
- Where do communication handoffs exist mainly because information must be translated, summarized, or reformatted?
- Could GenAI prepare or transform this material while preserving human ownership of consequential communication?
- Which communications depend on relationships, sensitivity, authority, or nuance that should remain strongly human-led?
- Would simplifying communication remove work, or merely increase the volume of messages produced?
- Which communication-heavy workflows could become materially simpler rather than merely faster?
Identifying Quality Bottlenecks That GenAI Could Help Address
- Where does inconsistent or poor-quality work create repeated correction or downstream rework?
- Which quality problems are detected too late in the workflow?
- What common errors, omissions, or inconsistencies recur across cases?
- Where does quality assurance depend on scarce experts whose time limits throughput?
- Could GenAI provide an effective first-pass check or flag cases that need closer human review?
- Which quality judgments still require professional expertise or accountable human decisions?
- Where could redesign improve both quality and workflow efficiency instead of simply adding another layer of review?
Understanding Where Human Judgment Adds the Most Value
- Which steps involve consequential decisions, ambiguity, or tradeoffs that require strong human judgment?
- Where does professional expertise or tacit organizational context materially change the right answer?
- Which human activities are primarily mechanical even though they currently sit inside an expert role?
- Where can GenAI prepare evidence, alternatives, or routine analysis so people focus on the parts requiring judgment?
- What human capabilities become more important if routine processing is reduced?
- Which decisions or actions require legitimate human authority even if GenAI performs much of the preparation?
- How could the workflow concentrate human attention where it creates the greatest value?
Mapping How GenAI Currently Fits Into the Workflow
- At which points in the current workflow is GenAI actually being used?
- Who initiates each GenAI interaction, and what are they trying to accomplish?
- What information, documents, or system context does GenAI receive at each point?
- What outputs does it produce, and how are those outputs used afterward?
- What manual work occurs immediately before and after the GenAI step?
- Where are employees compensating for missing integrations, weak context, or unreliable output?
- What does the full map reveal about where the workflow could be simplified, resequenced, or redesigned?
Identifying Parts of the Workflow That No Longer Make Sense
- Which steps exist primarily because earlier tools could not perform work that GenAI can now support?
- What manual processing, duplication, or translation is no longer necessary?
- Which approvals or reviews still serve a meaningful control purpose, and which survive mainly through inertia?
- What outputs are being produced because the old process required them even though nobody now needs them?
- Where are employees repeating work that GenAI has already performed sufficiently well?
- What steps should remain because they preserve judgment, accountability, quality, or resilience?
- What could be removed entirely rather than accelerated with GenAI?
Prioritizing Workflows for Deeper Redesign
- Which workflows combine high recurring effort with strong evidence that GenAI can materially change the work?
- Where is GenAI already producing useful local results that the surrounding workflow prevents the organization from capturing fully?
- Which workflows affect enough people, customers, cases, or resources for redesign to matter?
- What redesign is feasible with current tools, information, integrations, and organizational capability?
- Which workflows carry consequences high enough that redesign requires stronger evidence and controls?
- Where would changing the full workflow produce more value than continuing to optimize individual tasks?
- Which workflows should be examined first when value, feasibility, risk, and learning potential are considered together?
Redesigning Work Around GenAI
Defining the Outcome the Redesigned Work Should Produce
- What should this workflow ultimately deliver for the customer, employee, decision-maker, or organization?
- Which current delays, errors, costs, or unnecessary activities should redesign remove?
- What quality, speed, capacity, or service outcome matters most?
- What level of improvement is realistic given current GenAI capability?
- Which responsibilities, safeguards, or professional standards must remain unchanged?
- What newly enabled work could become possible if the workflow were redesigned effectively?
- What clear outcome should guide decisions about human roles, GenAI use, sequencing, and controls?
Separating Work Best Done by People From Work GenAI Can Support
- Which activities are well suited to GenAI because they involve recurring information processing, generation, comparison, or analysis?
- Which activities require human authority, professional judgment, tacit context, or interpersonal understanding?
- Where should GenAI prepare material that a person evaluates or decides on?
- Which normal cases could GenAI handle more extensively while people focus on exceptions?
- What work should remain fully human because failure would be too difficult to detect or too consequential?
- How might the division change as capability and reliability improve?
- What allocation of work produces the strongest overall outcome rather than simply maximizing automation?
Redesigning the Sequence of Human and GenAI Activities
- What should happen first if the workflow is designed around the desired outcome rather than the historical process?
- Where should GenAI enter the workflow to create the greatest leverage?
- Which human decisions or checks need to occur before GenAI proceeds to the next stage?
- What activities can happen in parallel rather than sequentially?
- Where can unnecessary handoffs or repeated context reconstruction be removed?
- At what points should human judgment interrupt or redirect the GenAI-supported path?
- What new sequence best combines speed, quality, accountability, and the comparative strengths of people and GenAI?
- What information does GenAI need at each stage to produce a useful and reliable result?
- Which sources should be treated as authoritative when information conflicts?
- How current must the information be for the workflow to remain dependable?
- What organizational context is required that the model cannot reasonably infer?
- Which information cannot or should not be made available to the GenAI system?
- Where would missing, stale, or poorly structured information create hidden failure risk?
- What minimum context and information environment is necessary before the redesigned workflow is viable?
Designing Human Review Into the Workflow
- Which GenAI outputs require human review before the workflow can proceed?
- How should review intensity vary with consequence, uncertainty, and detectability of error?
- Who has the expertise and authority needed to perform each review?
- What facts, assumptions, evidence, or quality dimensions should reviewers check?
- Which reviews can be reduced or removed because they duplicate other controls?
- Where would mandatory review consume enough effort to eliminate the value of GenAI?
- What review design preserves accountability while allowing the workflow to capture genuine speed or capacity benefits?
Planning How Exceptions and Difficult Cases Should Be Handled
- What kinds of cases are likely to fall outside the assumptions of the normal GenAI-supported path?
- How will users or systems recognize that an exception has occurred?
- Which signals should cause the workflow to stop, escalate, or revert to a different method?
- Who should own difficult or unusual cases?
- What context and history must accompany the case when it is escalated?
- Can GenAI help identify exceptions without being solely responsible for deciding that a case is exceptional?
- How can exception handling remain robust without burdening ordinary cases with unnecessary complexity?
- Which specific GenAI capabilities are required for the redesigned workflow to function?
- What other systems, applications, or integrations must GenAI or users be able to access?
- Which documents, data, or knowledge sources are essential?
- What permissions, connectors, or technical capabilities are currently missing?
- Which dependency would create the greatest operational risk if it became unavailable?
- What parts of the redesign can work without deep technical integration?
- What prerequisites need to be solved before the organization commits to the redesigned workflow?
Redesigning Handoffs Between People and GenAI-Assisted Work
- What information, decisions, and unresolved issues need to pass at each handoff?
- How can context be preserved so the next person or system does not have to reconstruct the work?
- Who remains responsible for the result as work moves between people and GenAI-supported stages?
- What output format would make the next stage easiest to perform and review?
- Where do current handoffs create unnecessary rework, ambiguity, or delay?
- Could GenAI prepare a clearer handover without obscuring what has been verified or decided?
- What handoff design reduces friction while keeping accountability and decision authority visible?
Testing the Redesigned Workflow in Real Work
- Which part of the redesigned workflow should be tested before the organization commits to the full design?
- What real cases provide enough variation to expose both strengths and failure modes?
- What existing workflow or outcome should serve as the comparison baseline?
- Which users should participate so the test does not depend only on enthusiasts or experts?
- What quality, cycle-time, effort, exception, and verification outcomes should be observed?
- What unexpected workarounds, failures, or downstream effects appear during the test?
- What evidence would justify continuing, changing, or abandoning the redesign?
Refining the Workflow Based on Practical Experience
- Which parts of the redesigned workflow performed as expected in real use?
- Where did employees routinely deviate from the intended design, and why?
- What steps introduced unnecessary effort, delay, or confusion?
- Which GenAI interactions were less reliable than the design assumed?
- What review, escalation, or context requirements turned out to be excessive or insufficient?
- What new opportunities or dependencies became visible only after the workflow was used?
- What should change before the redesigned workflow is adopted more broadly?
Adjusting Roles and Responsibilities
Identifying Tasks That Change Because of GenAI
- Which tasks are now being performed differently because GenAI has become part of the work?
- What parts of those tasks have become faster, easier, or more automated?
- Which steps are disappearing entirely rather than merely becoming quicker?
- What new responsibilities for context, direction, verification, or exception handling have appeared?
- How is the role's distinctive human contribution changing as routine work shifts?
- Which task changes are temporary while users and workflows are still learning?
- What changes are stable enough that role expectations should now be updated?
Identifying Tasks That Require Less Human Effort
- Which recurring tasks now require demonstrably less human effort under normal conditions?
- How much net effort is actually reduced after preparation, supervision, verification, and correction are included?
- What human input remains necessary even when GenAI performs most of the task?
- Are the savings consistent across users, cases, and realistic workload variation?
- Which part of the remaining human effort is now concentrated in difficult judgments or exceptions?
- Is the released capacity substantial and predictable enough to affect planning or staffing assumptions?
- Which reductions have been demonstrated strongly enough to treat as durable rather than anticipated productivity?
Identifying New Tasks Created by GenAI-Enabled Work
- What new work has appeared because employees now direct, evaluate, or supervise GenAI-supported activity?
- Who needs to maintain reusable prompts, context, agents, templates, or other shared GenAI resources?
- What new monitoring, verification, exception handling, or quality-control tasks are required?
- Has GenAI created additional coordination work between roles or systems?
- Which new tasks produce enough value to become explicit responsibilities?
- What temporary work exists only because the organization is still transitioning to a more mature workflow?
- Who should own genuinely new responsibilities once GenAI-enabled work becomes normal?
Clarifying Where Human Judgment and Accountability Remain
- Which decisions must remain explicitly human regardless of how much preparation GenAI performs?
- Who is accountable for the quality and consequences of GenAI-assisted work?
- What professional expertise remains necessary to interpret, challenge, or override GenAI output?
- Which review responsibilities need to be explicit rather than assumed?
- Where could redesigned workflows make responsibility ambiguous because several people and systems contribute?
- What should role descriptions or working agreements make clear about authority and accountability?
- How can human responsibility remain visible even when the person performs less of the underlying production work?
Reassigning Responsibilities in Redesigned Work
- Which responsibilities no longer fit the role boundaries created by the old workflow?
- Where has GenAI shifted preparatory, analytical, or coordination work between roles?
- Which tasks could move to another role because expertise requirements have changed?
- Who should own GenAI-supported preparation, verification, exception handling, and final decisions?
- What handoffs or role boundaries can now be removed?
- Which changes affect formal authority, accountability, employment expectations, or other matters requiring explicit agreement?
- How should new ownership be communicated so people understand both what has changed and what has not?
Adjusting Collaboration Between Different Roles
- Which roles now need to collaborate differently because GenAI changes how information or expertise is produced?
- What information should move directly between roles that previously depended on intermediate processing?
- Where can GenAI reduce coordination effort or make specialist expertise more widely accessible?
- What new dependencies on reviewers, technical owners, or domain experts are being created?
- Which old boundaries are becoming less useful because GenAI bridges part of the previous capability gap?
- Where could broader role overlap create confusion about authority or responsibility?
- What collaboration model best supports the redesigned work while keeping ownership clear?
Updating Managerial Expectations for GenAI-Enabled Work
- What performance expectations are justified by demonstrated GenAI-supported capability in this role?
- Should speed or volume expectations change, and what evidence supports the change?
- How should quality expectations evolve when employees have access to stronger preparation or review tools?
- What level of GenAI capability is reasonable to expect from people performing this work?
- How should managers evaluate outcomes without rewarding raw GenAI usage or penalizing appropriate non-use?
- What expectations could turn productivity gains into unsustainable workload or cognitive pressure?
- What changed expectations are stable and important enough to communicate formally?
Identifying New Capability Requirements for Roles
- What GenAI capabilities does this role now need in order to perform effectively?
- Which abilities to define problems, select context, evaluate outputs, and handle exceptions have become more important?
- What domain expertise must remain strong so employees can supervise GenAI-supported work intelligently?
- Does the role need stronger information, data, source, or evidence literacy?
- Which capabilities can develop through ordinary repeated work?
- Which new requirements need structured learning, coaching, or professional development?
- How should role-development priorities change without overinvesting in tool-specific techniques that may quickly become outdated?
Deciding How Freed Capacity Should Be Used
- How much usable capacity has actually been released rather than merely estimated from nominal time savings?
- Is the capacity concentrated enough to be redirected meaningfully?
- What important work is currently underperformed because people lack time?
- Could the capacity improve quality, customer service, innovation, relationships, decision-making, or other valuable activity?
- What low-value work should not simply expand because employees can now produce more of it?
- What role should employees and managers have in deciding how released capacity is used?
- Which use of the capacity would create the greatest organizational value rather than merely filling the available time?
Supporting People Through Changes in Roles and Responsibilities
- What concrete changes in tasks, responsibilities, expectations, or collaboration are employees experiencing?
- Which changes are creating the greatest uncertainty or concern?
- What has been formally decided, and what remains experimental or subject to review?
- What new capability, practice, or support do employees need to succeed in the changed role?
- Which old responsibilities are genuinely being removed rather than simply supplemented with new GenAI-related work?
- How can employees contribute practical knowledge to the redesign of their own work?
- What support will help the transition improve the role without allowing expectations to move faster than proven GenAI capability and workflow reliability?
Developing Practical GenAI Capability
Describing Work Situations Clearly to GenAI
- What does GenAI need to understand about this work situation before it can help meaningfully?
- Which background details materially change what a useful response would look like?
- What constraints, responsibilities, or boundaries should be made explicit?
- Which details can be left out because they do not affect the task?
- What would a knowledgeable colleague need to know before helping with the same problem?
- Which assumptions would be risky to leave for GenAI to infer on its own?
- How can the situation be described naturally enough that the user can reuse the skill across different kinds of work?
Providing Enough Context for Useful Results
- What information is GenAI currently missing that prevents a more useful response?
- Which documents, examples, data, or prior work would materially improve the interaction?
- What does GenAI need to know about the intended audience, stakeholder, or use of the result?
- Which previous decisions, definitions, or organizational constraints should shape the response?
- What information is authoritative enough that GenAI should rely on it rather than general knowledge?
- At what point would adding more context create little additional value or make the task harder to manage?
- What is the smallest useful context package that gives GenAI enough grounding without overwhelming the interaction?
Defining the Intended Outcome of a GenAI Interaction
- What does the user actually need to accomplish through this interaction?
- What should the result enable the user to do next?
- Which characteristics would make the output genuinely useful rather than merely plausible?
- What level of detail, structure, evidence, or format best supports the intended use?
- Which quality dimension matters most for this task?
- What should GenAI deliberately avoid optimizing for because it would weaken the real objective?
- How can the intended outcome be stated clearly enough to guide the interaction without overengineering the request?
Developing Results Through Conversation
- What needs to be clarified before asking GenAI for a finished result?
- Which parts of the work are easier to develop separately before combining them?
- What feedback should the user give when the response is useful in some respects but weak in others?
- Where should GenAI be asked to challenge assumptions, expose uncertainty, or revisit earlier reasoning?
- When would generating alternatives be more useful than continuing to refine one answer?
- What should the user inspect before allowing the interaction to move to the next stage?
- How can the user treat conversation as a way of developing the work rather than expecting one prompt to produce the final answer?
Working With Existing Documents and Source Material
- Which documents or sources should GenAI use as the basis for this work?
- What material should be treated as authoritative when sources differ?
- What information should GenAI ignore because it is outdated, irrelevant, or outside the requested basis?
- Which parts of the material need interpretation, comparison, or transformation rather than simple summarization?
- How can the user preserve important distinctions and meaning from the original sources?
- Where would unsupported additions or general knowledge weaken the task?
- What instructions would keep the result grounded in the provided material while still allowing useful analysis?
Breaking Complex Work Into Manageable Interactions
- What are the major components of the work, and which of them can be handled separately?
- Which part should be addressed first because later stages depend on it?
- What context, assumptions, or decisions need to carry forward between stages?
- Which intermediate results should be reviewed before the next stage begins?
- Where would a single large request make errors, omissions, or weak reasoning difficult to detect?
- What checkpoints would let the user redirect the work before problems compound?
- What sequence of interactions gives GenAI enough structure while keeping the user in control of the overall task?
Comparing and Challenging GenAI Outputs
- What assumptions does this output depend on?
- What reasonable alternative interpretation or conclusion could also fit the situation?
- Which part of the reasoning is least well supported?
- What evidence actually supports the main claims or recommendations?
- What relevant perspective, constraint, or counterargument may be missing?
- How would a knowledgeable critic challenge this response?
- What needs to change before the user should rely on the output for the intended purpose?
Using GenAI to Support Analysis and Decision Preparation
- What decision, judgment, or analysis is this interaction meant to prepare?
- Which facts or source material should be established before options are compared?
- What alternatives deserve consideration rather than allowing GenAI to converge too quickly on one answer?
- Which criteria, objectives, or constraints should guide the analysis?
- What risks, tradeoffs, dependencies, or second-order effects should be surfaced?
- What uncertainty could materially change the conclusion?
- How can GenAI strengthen the preparation while leaving the actual decision and accountability with the appropriate person?
Reusing Successful GenAI Working Patterns
- What previous GenAI interaction produced a reliably useful result?
- What underlying work pattern made that interaction successful?
- Which elements of the approach can be reused in similar situations?
- What context, criteria, or source material must change for the new case?
- Where could reuse become mechanical and reduce the quality of the interaction?
- How can the useful pattern be captured in a way that is easier to remember than an exact prompt?
- What other recurring work situations could benefit from the same underlying approach?
Adapting GenAI Methods Independently to New Work
- What familiar GenAI working pattern resembles this unfamiliar work situation?
- Which parts of the previous method still make sense here?
- What new context, objective, or stakeholder changes how the interaction should work?
- Which additional constraints or risks make the new situation meaningfully different?
- What small test could show whether the adapted method is useful before relying on it more heavily?
- What can the user learn from the first attempt even if the result is weak?
- Can the user now design an appropriate GenAI interaction from the work itself rather than depending on a predefined example?
Responding to Adoption Problems
Matching the Response to the Actual Cause of the Problem
- What adoption problem are we actually trying to change?
- What evidence supports the current diagnosis rather than another plausible explanation?
- Which intervention directly addresses the mechanism causing the problem?
- What commonly suggested response would treat only the visible symptom?
- Does the intervention fit the affected user group, work, and organizational context?
- What unintended effect could the proposed response create?
- What specific change should occur if the diagnosis and intervention are both correct?
Fixing Practical Access Problems
- What exactly is preventing users from accessing or using the required GenAI capability?
- Which people, roles, or work situations are affected?
- Is the constraint technical, procedural, licensing-related, permission-related, or caused by missing information access?
- What part of the problem can be resolved immediately without weakening necessary controls?
- Which issue requires involvement from technology, security, procurement, or another function?
- How should affected users be informed while the problem is being resolved?
- What system-level change would stop the same access problem from recurring?
Clarifying Confusing Usage Boundaries
- Which specific usage rule or boundary is creating uncertainty?
- What recurring work situations make the current guidance difficult to apply?
- Is the problem that employees misunderstand the guidance or that the guidance itself is ambiguous?
- What realistic examples would make the boundary clearer?
- Which questions genuinely require a policy owner or specialist rather than another explanatory document?
- How should clarification reach everyone affected rather than only the employees who happened to ask?
- What simpler wording or decision rule would let users act independently in similar situations next time?
Providing More Relevant Role-Specific Examples
- Which roles or groups are failing to connect existing GenAI examples with their own work?
- What recurring responsibilities or problems matter most to those users?
- Which real situations from comparable work would demonstrate useful GenAI support credibly?
- What examples are simple enough to test without reducing the role to trivial tasks?
- What limitations, verification requirements, or human responsibilities should the example show?
- Who can provide or validate examples that will feel credible to the affected users?
- How can examples help employees recognize their own opportunities rather than simply consume demonstrations?
Providing Targeted Support Where Capability Is Weak
- What specific GenAI capability is preventing the user from working independently?
- Which users actually share this capability gap?
- Can the missing skill be developed through guided real-work practice rather than broad training?
- Would peer coaching, examples, or a short intervention be enough?
- What general support would be unnecessarily broad for the diagnosed gap?
- What behavior would demonstrate that the targeted capability has improved?
- When should the additional support be reduced so users do not become dependent on it?
Addressing Managerial Barriers to Adoption
- What manager behavior or local management condition is limiting useful adoption?
- What is causing that behavior: uncertainty, workload pressure, skepticism, risk concern, incentives, or unclear expectations?
- Is the manager responding rationally to a real organizational problem that needs to be fixed elsewhere?
- What knowledge, support, authority, or experience would help the manager change the relevant behavior?
- What leadership expectation or decision needs to be clarified?
- When does a local management problem require escalation beyond the team?
- How can managerial behavior improve without turning support for GenAI into pressure for visible compliance?
Changing Learning Approaches That Are Not Working
- What learning activity is failing to produce the intended capability or behavior?
- Are employees struggling because they cannot connect the material to real work?
- Is the content too generic, theoretical, product-focused, or detached from actual work situations?
- Does the approach contain too much instruction and too little meaningful practice?
- What evidence shows that participants are not becoming more capable or independent?
- What alternative learning method better matches the actual gap?
- How can the revised approach be tested on a small enough scale to learn quickly before replacing the broader program?
Removing Workflow Friction That Discourages GenAI Use
- Which step makes a potentially useful GenAI practice slower or harder than the existing method?
- Can the friction be removed through a simple process change before investing in deeper technical integration?
- What information or context is being copied, reformatted, or re-entered repeatedly?
- Which review or approval steps could be simplified without weakening necessary control?
- Does the surrounding workflow need redesign rather than another user-level tip or training intervention?
- What change would make the GenAI-supported method practically easier than the alternative?
- What real-work test would show whether removing the friction actually changes adoption and value?
Responding to Trust or Quality Concerns
- What specific quality problem or failure is reducing trust?
- How often does it occur under realistic conditions?
- Is the problem caused by poor task fit, missing context, weak user technique, inadequate tools, or a persistent model limitation?
- What verification or review method could reduce the risk without making the use uneconomic?
- Should the use case be narrowed to cases where GenAI is more dependable?
- What evidence would justify greater reliance after the intervention?
- When is the correct response to stop using GenAI for this task rather than continuing to rebuild confidence?
Checking Whether the Intervention Actually Improves Adoption
- What specific barrier or behavior was the intervention intended to change?
- What observable change should occur if the intervention works?
- Did the affected users actually experience the predicted improvement?
- Did the original constraint decline, or did usage rise while the underlying problem remained?
- What unintended consequences or new barriers appeared?
- Is the evidence strong enough to draw a conclusion rather than simply needing more normal usage experience?
- Should the intervention continue, be redesigned, be replaced, or stop before additional adoption activity is added?
Building GenAI Into Organizational Practices
Integrating GenAI Into Employee Onboarding
- What should a new employee understand about how GenAI is actually used in this organization?
- Which approved tools and access should be available early enough to support meaningful work?
- What practical usage boundaries should new employees understand from the beginning?
- Which simple role-relevant tasks can introduce GenAI naturally during real work?
- What belongs in formal onboarding, and what would be better learned later through experience?
- How can onboarding explain human responsibility, verification, and escalation without overwhelming beginners?
- How can appropriate GenAI use become part of learning how work is done here rather than a separate technology module?
Integrating GenAI Into Ongoing Professional Development
- Which existing professional-development activities should now include GenAI because it materially affects the work?
- What role-specific GenAI capability matters more than generic product familiarity?
- How can development use real professional tasks rather than disconnected AI exercises?
- Which human skills become more important as GenAI takes on more routine information work?
- What capability needs periodic refresh because tools, workflows, or expectations continue to change?
- How should development differ between beginners, experienced users, and people doing more advanced GenAI-supported work?
- How can professional development remain stable around enduring capabilities while adapting to changing technology?
Building GenAI Into Recurring Planning Activities
- Which recurring planning activities would genuinely benefit from GenAI support?
- Where could GenAI help clarify objectives, dependencies, assumptions, or missing information?
- How could it broaden the range of options or scenarios considered before commitment?
- Where could GenAI challenge a plan or surface risks that are otherwise easy to overlook?
- Which planning judgments must remain with the accountable people involved?
- What inputs, evidence, or GenAI-generated claims need verification before influencing a plan?
- How can useful GenAI support become part of planning without adding another mandatory layer to every planning activity?
Building GenAI Into Meeting and Decision Preparation
- Which recurring meetings or decisions involve enough preparation work for GenAI to create meaningful value?
- What relevant context, evidence, or previous decisions could GenAI help assemble before the meeting?
- How could GenAI surface questions, objections, assumptions, or unresolved issues that deserve attention?
- Where could it compare alternatives or organize tradeoffs before people discuss them?
- What information should not be provided to the available GenAI environment?
- What needs human verification before GenAI-supported material enters the meeting or decision process?
- Which preparation practices are mature enough to become normal without turning GenAI into a required participant in every meeting?
Integrating GenAI Into Standard Work Templates and Resources
- Which existing templates or resources could become more useful if they reflected proven GenAI-supported ways of working?
- Where could a standard resource reduce repetitive context preparation or common user mistakes?
- What reusable organizational context, criteria, or verification guidance belongs in the resource?
- Which parts should remain tool-neutral so the resource survives changes in products and models?
- How can templates avoid freezing today's prompt wording or technical limitations into long-term practice?
- What does the user actually need at the moment of performing the work rather than in a separate AI guide?
- Which standard resources are mature enough to evolve, and which GenAI practices are still too unstable to institutionalize?
Building GenAI Into Review and Quality Practices
- Where could GenAI add useful quality assurance without simply creating another review layer?
- Which checks involving completeness, consistency, comparison, or known criteria can GenAI support reliably?
- What professional or consequential judgments should remain explicitly human?
- How could GenAI help reviewers identify omissions, contradictory information, or unusual cases?
- Where could a polished GenAI critique create false confidence in work that remains weak?
- What evidence or source support should a reviewer require before accepting GenAI-generated quality findings?
- How can existing review practices be redesigned so GenAI improves quality while unnecessary duplicate checking disappears?
Integrating GenAI Into Knowledge-Sharing Practices
- What useful organizational knowledge is currently difficult to find, understand, or reuse?
- Where could GenAI help employees search, summarize, compare, or organize established organizational knowledge?
- Which recurring knowledge products could become easier to maintain or more useful?
- How should source provenance remain visible so users can distinguish generated synthesis from authoritative material?
- What information should remain excluded or tightly controlled?
- How could GenAI help employees reuse prior work without propagating obsolete assumptions or low-quality material?
- Which existing knowledge-sharing practices would materially improve if GenAI were embedded into them?
Updating Established Working Methods to Reflect GenAI
- Which established working methods were designed around assumptions that GenAI capability has now changed?
- What steps exist mainly because earlier tools could not perform work that GenAI can now support?
- Where could GenAI strengthen the method without removing valuable human judgment or accountability?
- Which human steps still serve an essential purpose even if they take time?
- Which established methods remain effective enough that changing them would add little value?
- What real-work test should occur before an updated method becomes standard?
- Which working methods are worth revisiting first because the mismatch between old assumptions and current capability is greatest?
Embedding GenAI Learning Into Existing Organizational Routines
- Where do employees and teams already reflect on how work was performed and what should improve?
- Could project reviews, retrospectives, or team discussions capture useful GenAI successes and failures naturally?
- How can closeouts record transferable working patterns rather than merely saving prompts?
- Which recurring forums could surface changes in capability, workflow friction, or support needs?
- How can GenAI learning remain lightweight enough that people actually contribute to it?
- Which routines would become overloaded if too much adoption content were added?
- Where can organizational learning about GenAI become part of normal improvement activity rather than a permanent separate program?
Making GenAI-Enabled Work Part of Normal Organizational Practice
- Which GenAI-supported methods have enough evidence and repeatability to become normal ways of working?
- Can ordinary intended users perform them independently without relying on the original enthusiasts?
- What existing organizational routines, systems, and accountabilities can support the method?
- Which practices still depend too heavily on individual expertise, workarounds, or unusual local conditions?
- What documentation, training, monitoring, or specialist support remains necessary once the practice becomes normal?
- How will the organization keep an established GenAI-supported method adaptable rather than freezing today's implementation?
- What evidence would show that useful GenAI-supported work no longer requires a special adoption initiative to sustain it?
Reviewing the Adoption Approach
Comparing Adoption Progress With the Intended Outcome
- What changed condition did the organization originally intend GenAI adoption to produce?
- What does current evidence show about progress toward that outcome?
- Which user behaviors, capabilities, workflows, or organizational practices have actually changed?
- Where has adoption developed more broadly or deeply than expected?
- Where is progress materially behind the intended outcome, and why?
- Has the original outcome become outdated because GenAI capability, business priorities, or organizational learning changed?
- What does the comparison imply about what the next phase of adoption should emphasize?
Reviewing Which Parts of the Adoption Approach Worked Well
- Which adoption activities are most strongly associated with meaningful progress?
- What support do users and managers describe as genuinely useful in real work?
- Which interventions increased independent capability rather than ongoing dependence on support?
- What manager behaviors or local conditions helped useful adoption develop?
- Which communication practices improved clarity, relevance, or willingness to experiment?
- Which successful elements work only under particular conditions, and which appear transferable?
- What should clearly continue because the evidence shows that it is still creating value?
Identifying Adoption Activities That Produced Little Value
- Which adoption activities consumed substantial effort without changing useful behavior or work outcomes?
- What training or guidance was completed but rarely applied afterward?
- Which resources, communities, or support mechanisms attract little meaningful use?
- What communication generated attention but little practical change?
- Did these activities fail because the idea was weak, the execution was poor, or the underlying need did not exist?
- Should the activity be simplified, redesigned, replaced, or stopped?
- What scarce resources could be redirected if low-value activities were removed?
Comparing Results Across Different User Groups
- Which user groups have made meaningful progress, and on which dimensions of adoption?
- Which groups remain constrained despite receiving similar support?
- What access, work relevance, manager environment, or opportunity differences explain the variation?
- Which differences are appropriate because the underlying work differs?
- What adoption methods appear effective only for particular groups or conditions?
- Where does unequal progress reflect avoidable inequality in access, learning, or organizational support?
- What should change in the next cycle instead of applying the same intervention more broadly?
Reviewing Whether Support Matches Current User Needs
- What forms of support are users actually relying on now?
- Which earlier support needs have declined as users became more capable?
- What new questions are appearing as GenAI use becomes deeper, more integrated, or more autonomous?
- Are experienced users receiving enough help with advanced workflows, integrations, verification, or changing capabilities?
- Are beginners still able to find the basic help they need?
- Which support is too generic to address the problems users now face?
- How should the support model evolve so routine needs become easier while scarce expertise focuses on higher-leverage problems?
Reviewing the Effectiveness of Adoption Communication
- Do employees understand why GenAI adoption matters to the organization now?
- Are current expectations for employees and managers clear enough to guide real behavior?
- Which messages are being misunderstood or interpreted differently across teams?
- What legitimate concerns remain unanswered because communication cannot resolve the underlying issue?
- Are managers translating organization-wide messages consistently into local work?
- Has the communication kept pace with what employees are actually experiencing?
- What should change so future communication remains credible, relevant, and clear about uncertainty?
Assessing Whether Current Measures Show What Matters
- Which dimensions of adoption and value are currently being measured?
- Do the measures distinguish access, experimentation, recurrence, independence, integration, and actual work outcomes?
- Are raw usage counts receiving more attention than their diagnostic value justifies?
- What important behavior, quality, shadow use, or workflow change remains invisible to current telemetry?
- How credible are the measures used to claim productivity or business value?
- Which metrics could encourage superficial usage or other undesirable behavior?
- What should the measurement system change so it better supports decisions rather than simply reporting activity?
Reviewing How Adoption Resources Are Being Used
- Where are specialist, technical, learning, management, and financial adoption resources currently being spent?
- Which activities consume the largest amount of capacity?
- What support or infrastructure appears to create the greatest organizational leverage?
- Where are resources underused because the intended need has not materialized?
- Which teams receive unusually high support, and is the difference justified by value, risk, or learning potential?
- What recurring needs could now become self-service or locally owned?
- How should scarce resources be reallocated based on the adoption problems and opportunities that now matter most?
Identifying Adoption Activities That Should Stop, Start, or Change
- Which current activity should stop because it no longer solves an important adoption problem?
- What important capability, barrier, or opportunity is not currently addressed?
- Which existing intervention is directionally useful but needs redesign?
- What should remain stable because it continues to work?
- What evidence supports stopping, starting, or changing each element?
- Which changes can occur immediately, and which need testing first?
- What combination of removal, simplification, expansion, and redesign would create the strongest next version of the adoption approach?
Defining the Next Improvement Cycle for GenAI Adoption
- What are the most important lessons from the current adoption cycle?
- Which high-value problems remain unresolved?
- What new opportunities or risks have emerged since the previous cycle began?
- What few priorities should define the next improvement cycle?
- Which interventions should be treated as tests rather than permanent additions?
- What evidence should be collected so the next review can make stronger decisions?
- What clear improvement objective would give the next cycle direction without turning adaptation into continuous churn?
Adapting to New GenAI Capabilities
Monitoring GenAI Developments That Could Matter to the Organization
- Which changes in GenAI capability could materially affect work the organization performs?
- What reliable sources should be monitored for changes in models, tools, integrations, autonomy, cost, or controls?
- How often is formal capability review useful without reacting to every product release?
- Which updates are incremental improvements that require no organizational response?
- What kinds of change could invalidate an important assumption behind current adoption, workflow, or governance decisions?
- Who needs visibility into different kinds of capability change?
- What monitoring approach keeps the organization informed while separating horizon scanning from operational change?
Distinguishing Meaningful Capability Changes From Minor Updates
- What real work becomes possible, easier, safer, or more economical because of the change?
- Does the new capability materially improve reliability under realistic conditions?
- Does it remove a barrier that previously prevented a worthwhile GenAI opportunity?
- Is the capability available in tools and environments the organization can actually use?
- What evidence demonstrates a significant change rather than a compelling demonstration?
- Is the improvement relevant across the organization or only to a narrow group of work?
- Has the change crossed a threshold large enough to justify testing or changing established practice?
Assessing How New Capabilities Affect Existing GenAI Opportunities
- Which existing GenAI opportunities become materially easier, deeper, or more valuable because of the change?
- What limitation that constrained the opportunity previously has actually disappeared?
- Which workflows could now delegate more work to GenAI or reduce earlier workarounds?
- Have any existing methods become unnecessarily complex because they were designed around older limitations?
- What new risks or dependencies appear alongside the improvement?
- Which previously lower-priority opportunities deserve renewed attention?
- What established use should be retested before the organization changes its current assumptions?
Identifying New Organizational GenAI Opportunities
- What valuable work becomes possible now that was not practical before this capability emerged?
- Which previously uneconomic, unreliable, or inaccessible tasks are worth reconsidering?
- What functions or work patterns are most directly affected by the new capability?
- Which user groups possess the expertise and context needed to test the opportunity meaningfully?
- What new customer, operational, analytical, or service possibilities appear?
- What must be learned through real work before the opportunity receives broader attention?
- Which new opportunities are significant enough to explore rather than merely technically interesting?
Reviewing Usage Boundaries After Capability Changes
- Does the new capability change what information GenAI can access, retain, combine, or expose?
- Are current employee-facing usage rules still sufficient for the changed capability?
- What new actions can GenAI now execute rather than merely recommend?
- Which permissions, approvals, verification requirements, or human decision points need reconsideration?
- Do new integrations, memory, tool access, or autonomy materially change the risk profile?
- What existing guidance could become misleading because it reflects older capability assumptions?
- Which boundaries need clarification before the capability is promoted or embedded more widely?
Updating GenAI Learning and Support
- What genuinely new capability do users need to understand in order to use it well?
- Which roles or user groups are affected enough that an update is worthwhile?
- What old guidance has become incomplete, overly restrictive, or unnecessarily complex?
- Which real-work examples would demonstrate the new capability without overstating what it can do?
- What support resources, team methods, or professional-development materials now need revision?
- What can experienced users discover safely through experimentation instead of formal retraining?
- How can learning stay current without requiring the entire workforce to retrain whenever a model or product changes?
Revisiting Existing GenAI-Enabled Workflows
- Which established workflows were designed around capability limitations that may no longer hold?
- Could the new capability remove manual preparation, handoffs, review, or workaround steps?
- What new system integrations or forms of tool use have become practical?
- Which human checks might reasonably change, and which should remain despite improved capability?
- What new failure modes or autonomy risks appear in the redesigned workflow?
- Should the workflow be retested under current conditions before any structural change is made?
- Which existing workflows offer the strongest combination of potential improvement and outdated assumptions?
Communicating Relevant New Capabilities to Employees
- Which new capability materially affects work employees actually perform?
- What practical difference does it make compared with the previous tool or working method?
- Which groups need to know now, and who can wait until the capability becomes relevant to them?
- What real example would make the change understandable without relying on technical feature language?
- What limitations and failure modes should be explained alongside the improvement?
- How can communication distinguish meaningful capability change from routine product marketing?
- What is a reasonable next experiment for employees rather than a blanket expectation to change how they work immediately?
Testing Important New GenAI Capabilities in Real Work
- What representative real task would provide a meaningful test of the new capability?
- What existing method, previous model, or human-only approach should serve as the comparison?
- Which users are representative enough to test transfer beyond technical enthusiasts?
- What quality, reliability, security, or accountability concerns should the test deliberately expose?
- What improvement would be large enough to justify changing the current approach?
- What difficult cases, exceptions, or failures should be included rather than testing only favorable conditions?
- What evidence would justify broader promotion, and what result would indicate that the capability should remain experimental?
Retiring Outdated GenAI Practices as Better Approaches Become Available
- Which current prompts, workflows, controls, examples, or support practices exist mainly because of limitations that no longer apply?
- What newer approach solves the same problem more simply, reliably, or economically?
- Is the old method still useful in any specific context?
- Which shared resources, templates, training materials, or team methods need to be updated or deprecated?
- Who still depends on the old practice, and what would make migration difficult?
- How should replacement be communicated so users can tell clearly which method is now current?
- What should be retired from active use so organizational GenAI practice can improve instead of accumulating obsolete layers?