The use case
One piece of real work, live with real users. The unit of delivery, evidence and learning.
An operating model for enterprise AI adoption
The AdoptionTree®Model
Build AI capability that stays.
From one working AI use case to an organization that transforms itself.
Version 1.1 · October 2026By Willem Knaap · Amsterdam
Explainer video
Figure in the film: McKinsey, The state of AI in 2026: On the road to ROI (25 Aug 2026). A self-reported survey of business respondents, not a count of all organizations; the 89% is regular AI use in at least one business function.
01The premise
A working use case is a beginning.
Build the ability to do it again.
01 / 06The thread
One working AI use case is not the goal. The ability to repeat it is. The Adoption Tree® Model takes an organization from its first use case to many, and turns that into a transformed way of working.
AI adoption is the ability to turn AI into trusted, governed, used and measurable ways of working. The Adoption Tree® Model connects a successful use case to the capability needed to sustain and repeat it.
The business domain
One end-to-end workflow with an accountable owner. The unit of transformation and scale.
The shared foundation
The data, controls, patterns, skills and ownership that every next wave inherits.
Based on the whitepaper · Premise and executive summary, pp. 2–3
02The model in one view
Visible value above.
Lasting capability below.
02 / 06The thread
The model in one picture: above ground the use cases people see, below ground the capability the organization keeps. Every use case that goes live feeds the roots, so the next one is designed to start further ahead.
Read it as a tree. The branches are the use cases people see and use. The trunk is the temporary Adoption Office that runs them. The roots are the foundation: six blocks of capability the organization keeps. Every use case that goes live feeds the roots, and the next one grows from them.
Hover or tap a use case to see what it feeds back into the roots.
The foundation enables the use case.
The use case makes the foundation stronger.
The six roots
What the organization keeps, and every next wave inherits.
Mandate & governance
Sponsor, domain owner and decision rights: who may release, scale, stop and accept risk.
Data & knowledge
Approved sources, access, quality and provenance the next wave reuses.
Architecture & integration
An owned foundation with connected systems and replaceable models.
Responsible AI & controls
Intended purpose, oversight, evaluation and release evidence, designed in from the start.
Workforce & coaching
Tasks redesigned with the people who do them; competent use built through coaching.
Measurement & value
Outcomes against a baseline; the value decision, assets reused and ownership transferred.
The six foundation blocks · Whitepaper, pp. 11–12
03The adoption wave
One domain. Six gates.
Evidence at every step.
03 / 06The thread
Six gates are six steps from use case to adoption: a fixed route so nothing is skipped, a team so everything is recorded, and a named role for everyone. Each wave leaves the organization with learnings it keeps.
Each gate is a decision with a named owner and minimum evidence. A wave that cannot show the evidence waits, changes or stops. Select a block to see what happens in it, who decides and what it gives you.
The reference wave · 24 weeks of working time
Select a block
The 24 weeks are a planning assumption, not an observed duration. The blocks overlap within three rules: nothing is built before Gate 1, nothing goes live before Gate 3, and no realization decision comes before Gate 4.
Gate 0 · Mandate · before the clock starts
Set the conditions before the work starts.
In this blockContract the mandate: sponsor, domain owner, internal successor, decision rights, budget, capacity, procurement and access.
What the gate decidesName the sponsor, domain owner and internal successor. Agree the outcome, budget, boundaries, foundation limit and mandate end date.
Why it matters to you
Ownership, budget and the exit date are settled before anyone picks a use case.
- Decided by
- Executive sponsor, on the domain owner’s commitment
- Evidence to proceed
- Signed charter · joint team · baseline · successor named
- After this gate
- The 24-week clock starts, provided the entry conditions hold
Running alongside
- Responsible AI & applicabilityStarts in week 1
- Workforce adoptionStarts at design
- TransferSuccessor named; shadowing starts
Gate 1 · Select · weeks 1 to 4
Choose one consequential use case.
In this blockMap the domain, its current workflow, users and AI activity. Establish the baseline, run the applicability screen and select the first use case.
What the gate decidesSelect a bounded piece of work within the domain, with real users and a measurable outcome. Establish its intended purpose and complete the applicability screen.
Why it matters to you
You commit to one use case with a measured starting point, and you know which rules apply before anything is built.
- Decided by
- Domain owner & product owner
- Evidence to proceed
- Use-case dossier · applicability screen
- After this gate
- Build may start on the parts of the design already approved
Running alongside
- Responsible AI & applicabilityApplicability screen, routing and classification
- Workforce adoptionStarts at design
- TransferThe office carries the rhythm; the successor shadows
Gate 2 · Design · weeks 5 to 8
Design the work and the controls together.
In this blockCo-design the target workflow, human oversight, architecture, data, evaluations, responsible-AI controls, support and learning.
What the gate decidesDefine the target workflow, human oversight, architecture, evaluation thresholds, coaching and support. Each control has an accountable owner. Where it applies, works-council consent is an exit condition.
Why it matters to you
Controls are designed into the work from the start, and each one has a named owner.
- Decided by
- Product owner & control owners
- Evidence to proceed
- Target workflow · controls · evaluation & learning plan
- After this gate
- Build proceeds on the approved design
Running alongside
- Responsible AI & applicabilityControls designed into delivery
- Workforce adoptionRole redesign; coaching prepared
- TransferThe office carries the rhythm; the successor shadows
Gate 3 · Release · weeks 9 to 12
Is it feasible? Release to bounded production.
In this blockBuild or configure, integrate, test and document. Train coaches, and prepare support and fallback.
What the gate decidesThe use case works as built and runs under its required controls, so it is released to a defined workflow and user group. Monitoring, support, human oversight and rollback are ready.
Why it matters to you
Nothing reaches real users until it works under its controls. The release and the residual risk are signed separately.
- Decided by
- Domain owner & residual-risk owner
- Evidence to proceed
- Signed release pack · live dashboard · support
- After this gate
- Go-live, for a defined group
Running alongside
- Responsible AI & applicabilityRelease evidence
- Workforce adoptionCoaches trained; support prepared
- TransferFrom Gate 3, internal owners chair, sign and present
No gate · bounded production · weeks 13 to 16
Run it live, for a defined group.
In this blockRelease to a defined group and workflow. Coach users. Monitor quality, trust, overrides, incidents, cost and outcomes.
No gate hereThis block builds the evidence that Gate 4 will judge: what production produced.
Why it matters to you
The value decision will rest on what happened in real use.
- Decided by
- No gate decision in this block
- Output
- Live dashboard · support model · production review
Running alongside
- Responsible AI & applicabilityMonitoring in production
- Workforce adoptionUsers coached in the live workflow
- TransferInternal owners chair; decisions jointly prepared
Gate 4 · Prove · weeks 17 to 20
Is it viable? Decide on measured evidence.
In this blockCorrect failure modes and compare with the baseline. Update assessments, standardize reusable assets and select the next use case.
What the gate decidesCompare outcomes with the baseline and check the business case against what production produced. Decide whether to scale, change or stop, and which reusable assets become maintained defaults.
Why it matters to you
Time saved counts only when the owners decide what it becomes. Reusable assets get an owner, so the next wave can build on them.
- Decided by
- Domain owner & value owner
- Evidence to proceed
- Value evidence · reuse register · next-use-case dossier
- At this gate
- The realization decision: scale, change or stop
Running alongside
- Responsible AI & applicabilityAssessments updated
- Workforce adoptionCompetent use and workload evidence
- TransferReusable assets get named maintainers
Gate 5 · Transfer · weeks 21 to 24
Show that internal owners can run it.
In this blockInternal owners chair the rhythm, rehearse incidents and changes, and approve the roadmap and the external exit.
What the gate decidesA named internal owner, not the Adoption Coach, produces and signs a real release pack without external drafting, and a control owner accepts it. The same owners chair the next wave’s mandate and selection decisions while the external team observes in silence.
Why it matters to you
You end the wave owning the work, and the next wave is designed to start further ahead.
- Decided by
- Sponsor, on the verifier’s report
- Evidence to proceed
- Operating manual · transfer scorecard · next-wave plan
- After this gate
- The office leaves and the coach stays. Wave two starts on what this wave left behind.
Running alongside
- Responsible AI & applicabilityInternal owners operate the controls
- Workforce adoptionTwo trained coaches, or a coach and a trained deputy, stay
- TransferThe exit test; the office observes in silence
Working time and elapsed time
Working time24 weeks
Elapsed, regulated enterpriseplan 9 to 12 months
Gate logic, parallel tracks and reference calendar · Whitepaper, pp. 13–14, 22–23, 28–29
04The capability that stays
The office is temporary.
Internal ownership is the outcome.
04 / 06The thread
AI keeps moving faster. The capability built here moves with it, because it is owned inside the organization and stays after the Adoption Office has handed over.
A jointly staffed office runs one domain wave. Internal successors are named on day one and take over the rhythm as the wave goes on; an internal Adoption Capability of trained coaches stays.
Read the complete transfer modelExit path · dependency is measured and designed to decline
Named internal roles and routines take over. The Adoption Coach stays. The mandate per domain is bounded at two to three waves, with the end date written into Gate 0.
The handover, step by step
Select a step
Step 1 · At Gate 0
The handover gets a name and a date on day one.
Gate 0 names the internal successor and writes the mandate’s end date into the charter. The successor shadows the office lead. From Gate 1, an internal Adoption Coach works next to the external one, and the internal share rises.
Who decides: your named roles, at every gate. The weekly rhythm: the office carries it while your successor shadows. Making, challenging, substituting and exiting are still supported by the office.
Why it matters to you
The exit is designed before the work starts. Successors know the date by which they must be ready, and the sponsor gets a scheduled decision.
Step 2 · From Gate 3
Your owners take the chair while support is still in the room.
From Gate 3, internal owners chair the meetings, sign and present. Decisions are prepared jointly with the office, and external support reduces as the wave goes on.
Who decides: your named roles, at every gate. The weekly rhythm: your owners chair it and the office supports. Making, challenging, substituting and exiting are prepared jointly.
Why it matters to you
Your people take real decisions with support at hand, before they have to take them alone.
Step 3 · Before Gate 5
Leaving is measured, not promised.
Your teams now run the routines and chair the decisions while the office stands aside. These are the main transfer conditions, each evidenced rather than asserted:
- Internal teams complete two consecutive four-week review cycles without the office, including one without the lead coach.
- A named internal owner, not the Adoption Coach, produces and signs a real release pack without external drafting, and a control owner accepts it.
- The same owners chair the next wave’s mandate and selection decisions while the external team observes in silence.
- The value owner has stated the twelve-month run cost and the domain owner has accepted it into a named budget line.
- Licenses, contracts, audit rights and exit dependencies are explicit, with no lock-in on the office’s tooling.
- Transfer does not depend on one person: there are two trained coaches per domain, or a coach and a trained deputy.
A third party can check the exit test from the paperwork alone. The transfer scorecard is a self-assessment by the office being tested. When a wave is high-risk, or further work is under discussion, verification should pass to a party with no stake, such as internal audit.
Who decides: your named roles, at every gate. The weekly rhythm: your people run it and the office observes in silence. Making, challenging, substituting and exiting are now exercised independently.
Why it matters to you
A wave that cannot show this evidence waits, whatever the calendar says.
Step 4 · Gate 5 and day 90
The office leaves, and the capability stays with your people.
At Gate 5 the external team’s part ends: the dossier is signed, the assets are owned internally and the office leaves. Authority stayed with your named roles throughout, and the Adoption Coach stays. Ninety days later, a light review checks the five things the first months erode:
- Controls still operate and evidence is filed.
- Incidents and support went through the internal route.
- Adoption held or improved.
- One model or vendor change was absorbed without the office, or, where none occurred, a change drill was documented.
- The reuse register and its assets were maintained.
Who decides: your named roles. The weekly rhythm: your people own it and the coach stays. The office has left.
Why it matters to you
The capability lives inside your organization. The next wave starts on the assets this one left behind.
The Adoption Office, the four dependency questions, coaching and the exit test · Whitepaper, pp. 21–22, 27–29
05Value that can be evidenced
Measure what changed.
And what the next wave inherits.
05 / 06The thread
Everything is measurable: what a use case is worth to the organization, what you steer on, and whether the organization learns and grows in capability, wave after wave.
Every measure has an owner, an evidence location and a date. Time saved needs a realization decision: how does it become capacity, service, revenue, risk reduction or another business outcome?
0Mandated→1Selected→2Ready→3Live→4Proven→5Transferred
Health, controls & adoption
Check quality, cost, incidents, oversight, active task use, workarounds, trust and workload in production.
Workflow & value
Compare cycle time, quality, throughput or service with the baseline. Record the business owner’s value decision.
Reuse & transfer
Track assets consumed and produced, internal decision-making, support ownership and the ability to change suppliers.
Seven measurement layers · Whitepaper, pp. 17–19
06The working system
Five tools.
One connected way of working.
06 / 06The thread
We wrote the whitepaper to map this whole system of AI adoption capability into one model, one paper and five practical tools that walk you through the steps.
The whitepaper includes five practical canvases. They share identifiers so a use case can be followed from visible work through its foundation, obligations and architecture.
T1
Frontstage Canvas
Map the use case, people, workflow, human oversight and evidence of adoption.
What changes in the work?Whitepaper · p. 34
T2Backstage Canvas
Connect governance, data, architecture, operations and the assets the next wave inherits.
What makes it repeatable?Whitepaper · p. 35
T3Adoption Tree® Canvas
Bring visible delivery and the capability foundation into one management view.
Does the whole system connect?Whitepaper · p. 36
T4Regulation Navigator
Record intended purpose, applicability, obligations, accountable owners and evidence.
What applies to this use case?Whitepaper · p. 37
T5Operating Architecture
Map the layers, interfaces, decisions and ownership behind the working capability.
What do we own, connect and reuse?Whitepaper · p. 38
The five tools are included on pp. 34–38 of the full whitepaper.
07The whitepaper
Take the model
into your organization.
Forty-three pages of research and working tools: the model, six gates, decision rights, measurement, the transfer test and five canvases. Identify yourself with your work email and the access link arrives in your inbox.
- The model, six gates and decision rights
- Measurement, reuse and a concrete transfer test
- Responsible AI, workforce adoption and coaching
- Five canvases and an illustrative reference wave
Check your inbox
We’ve sent you access to The Adoption Tree® whitepaper.
The link in the email is personal and valid for 7 days. Not there after a few minutes? Check your spam folder, or .
08A few useful distinctions
Before you start.
Who is the model for?
Enterprise sponsors, domain leaders, product and delivery teams, architects, control owners and workforce leaders who need AI to become a dependable way of working. The whitepaper is oriented toward the European enterprise context.
Is this a technology platform or a consulting service?
It is an operating model: a way to connect business ownership, production delivery, responsible operation and internal capability. Its tools can be used alongside your existing platforms and governance.
Does every wave need to build a new foundation?
No. Each wave should inherit approved assets and leave useful additions behind. Readiness is assessed before Gate 0. If it is too weak, the first step is a readiness project or a smaller use case.
Does temporary mean every external specialist must leave?
No. A chosen, priced and replaceable specialist can remain. What should decline is dependence on a particular external party to make decisions, challenge them, substitute a supplier or exit an arrangement.
Who pays for it after the office leaves?
A named budget line does, agreed before the handover rather than at it. The cost side is built at the same gates as the benefit side: a candidate envelope at Gate 0, a first run-cost estimate at Gate 2, metered consumption during bounded production, and both halves measured at Gate 4. At Gate 5 the domain owner accepts the run cost into a budget line.
Has the model been validated in practice?
Version 1.1 is published as a design proposal. It combines established ideas; the complete configuration has not yet been evaluated in practice. The reference case is illustrative and the timings are planning assumptions.
Agree what will go live.
And who will run it after.