AI enablement lead

AI adoption by role

The enablement lead runs the AI (artificial intelligence) adoption system: the enablement team or CoE (center of excellence), plus the change, HR (human resources), L&D (learning and development), communications and champions work around it. This track is the role's own view. The enablement program chapter describes the program itself, and this page links to it instead of repeating it. Counts, shares and durations on this page are illustrative.

The enablement lead is judged by what teams do without them in the room. Attendance, seat counts and demo applause are not evidence of adoption.

1. At a glance

Field	For the enablement lead
Mandate	Turn approved tools and guardrails into daily practice, role by role and tier by tier, inside governance and never around it
Primary outcome	Teams that climb the ladder on evidence, reuse shared assets, and keep going after coaching ends
First safe win	One pilot team moved one phase on one medium-tier task, with an honest baseline and the second line walked through the evidence
Biggest failure mode	Measuring attendance and enthusiasm, while assets have no owners and nobody checks that learning reached the work

Line: supports both (1st-line delivery teams and 2nd-line control partners) · Family: Adopt · Variants: enablement or CoE lead, change, HR, L&D, communications, champions

Variant	What differs
Enablement or CoE lead	Accountable for the program, the stage-move check and the shared asset portfolio; reports the scorecard
Change manager	Plans the people side: stakeholders, resistance causes, readiness, reinforcement
HR partner	Role redesign, performance expectations, consultation duties; owns workforce records
L&D partner	Learning paths, assessment of transfer to the work, learning records and attestation
Communications	One-page "what is approved, what is prohibited, and why"; what changed and when; after-incident messages
Champions	One volunteer per team with protected time; local support, early warning of friction

2. Your journey: first steps to fully adopted

The enablement lead's journey starts with listening and one pilot team, and ends with an adoption system that runs without the enablement lead in the room.

Start here: your first steps

Walk the current process with teams, the second line, HR and L&D, and map the gaps.
Publish the acceptable-use scenarios and check that every pilot team passes them.
Coach one pilot team on one real medium-tier task, from an honest baseline.
Publish the first catalog entries only with a named owner, a version and stated limits.

Fully adopted: how the work runs

Moment in the work	Before AI	Fully adopted (Stage 4–5)	Decision that stays human
A team asks for help	Demos for whoever asks loudest	Intake routes the request to a role journey and a lab that fits the task's risk tier	Which teams get coaching first
Learning	Attendance is counted; nobody checks the work	Labs run on approved tools and synthetic data; transfer is checked in real artifacts in the repository	Whether learning actually reached the work
Shared assets	Prompts are copied between chats, with no owner	An agent drafts catalog entries; each owner confirms owner, version, evaluation results and review date	Publishing or retiring an asset
Stage moves	Teams declare themselves advanced	A stage-move check against exit evidence; the second line also checks medium and high tiers	The stage-move check itself
Office hours and scorecard	The same question gets a different answer each week	An agent drafts cited answers from the catalog and policy and abstains when unsure; the scorecard updates continuously	Every answer sent, and anything said about roles

What changes for the organization

Before	After
Adoption measured by attendance and seats	Adoption measured by transfer, sustained use and reuse
Every team builds its own prompts and evaluations	Teams reuse owned catalog assets
Coaching never ends	Teams keep going after coaching ends

In any organization

Small or unregulated: a senior engineer or the engineering manager runs enablement part-time, with one champion per team.
What scales: the program team, formal learning records and second-line stage checks shrink. No customer data in labs or unapproved tools, human approval of merges, an owner on every shared asset and a kill switch for agents never do.

3. Your decisions

The column "your relationship" uses the decision relationship vocabulary: owner or approver · independent validator or challenger · consulted control partner · evidence recipient · notified. The enablement lead owns the program, not the risk decisions inside it.

Decision	Your relationship	Inputs	Escalation	Cannot be delegated
Learning paths per role and tier	Owner, with L&D	Role tracks, risk tiers, lab results	The sponsor	No; L&D can own the content
Check controls before a stage move	Owner of the check for low tier; for medium and high the second line checks too	Exit evidence, "must exist first" items	AI and model risk; security	Yes
Record the stage in the inventory	Owner	The approved stage move	AI governance function	No; a named delegate can record
Publish a shared asset	Owner of the publishing rules; the asset owner owns the asset	Evaluation results, owner, limitations, version	Architecture for technical assets	Yes, the rules
Retire a shared asset	Consulted; the asset owner and architecture decide	Usage, defects, replacement	Architecture	Not yours alone
Workforce communications	Owner with HR and communications	Sponsor decisions on role change	The sponsor and HR	Yes
Risk tier of a use case	Consulted; triage and AI and model risk decide	Intake card	AI governance function	Not yours to hold
Leader decision brief	Owner of the brief; the leader owns the decision	Pilot evidence, cost, residual risk	The sponsor	Yes, the brief

4. Two kinds of adoption

Kind	What it means for you	Example
How you work with AI	Drafting lab plans, workshop material, office-hours answers and catalog entries on approved tools	The assistant drafts a lab plan from the template; a control partner checks the governance note
What you enable that contains AI	Labs, demos, reusable prompts, reference agents, evaluation sets, control guidance and the learning journey itself	A reviewed prompt card for story drafting published with owner, version and limits

5. Your ladder

The adoption ladder describes controls and evidence. The enablement lead uses AI personally and also checks other teams' stage moves, so the table shows both.

Stage	AI use from your seat	Guardrail	Artifact or evidence	Decision relationship	Step-down trigger
0 · Foundations	None; you build the foundations with others	Inventory, intake, acceptable-use scenarios, escalation path	Scenario check bank; intake live	Owner of the scenario check; consulted on policy	Scenarios out of date after a policy change
1 · Personal productivity	Drafts lab plans, emails, FAQs	No restricted data; you apply every change	Drafts you reviewed	Owner of your material	A lab example uses real customer data
2 · Team context	Program instructions file; reviewed prompts for lab plans and catalog entries	Prompts reviewed like code; templates from the templates page	Catalog entries with owners	Owner of the program prompts	Entries published without an owner
3 · Delegated tasks	An agent drafts office-hours answers from the catalog and policy, with citations; it abstains when unsure	Evaluation set of past questions; citations required; you send nothing unread	Answer drafts with citations	Owner of every answer sent	Abstention or citation failure on the evaluation set
4 · Agentic workflows	You check teams' Stage 4 entry, not run agents on the program	Least privilege, kill switch, independent approval exist and were tested	Stage-move check record	Owner of the check; the second line challenges	A Stage 4 team has an untested kill switch
5 · Scale and sustain	Continuous signal from the catalog and scorecard	Asset ownership, periodic review, material-change process	Scorecard and asset health report	Owner of the scorecard	Assets without owners rise

Human-operated at Stage 5. Coaching conversations, the stage-move check, and anything said to people about their roles stay human. An agent can draft an answer. It never certifies a team or speaks for HR.

6. Priority scenarios

Scenario	Value	Risk tier	Stage	First step	Do not use AI when…
Lab plan drafting	Faster, consistent labs	Low	2	Start from the lab plan template	The lab would use real data
Office-hours answers	Same answer for the same question	Medium	3	Ground in the catalog and policy, with citations and abstention; see Lab 6	The question is about a person's role or performance
Catalog entry drafts	More assets documented	Low	2	Draft from the owner's notes; the owner confirms	No owner has agreed to own it
Learning-transfer check	Evidence that learning reached the work	Medium	2	Compare lab outputs with real artifacts in the repository	It would grade a named person automatically
Emerging-tech translation	Short, neutral notes on what a new capability means here	Low	1	Draft, then check with security and architecture	The note would imply approval of an unapproved tool
Workforce messages	Clear, consistent wording	Medium	1	Draft with HR and communications	HR has not confirmed the decision being described

7. The due-date ask from your seat

The organization must show the contractual payment due date the same way on web, mobile, statements and reminder notifications, and the payment cutoff time in the customer's own time zone. The enablement lead does not build the change. The enablement lead coached the pilot team that did.

Card field	For the enablement lead
Decision	Run the coaching cycle with the web team on one medium-tier task: Stage 3 test generation for the temporal edge cases
Input received	Intake card (medium tier, proprietary source, no customer data); the team's baseline; the reuse story of ten date implementations (illustrative)
Output and evidence	Lab 3 reuse advisory pointing to `shared-datetime`; Lab 4 evaluation set with daylight-saving and missing-zone cases; catalog entry for the `formatPaymentCutoff(cutoffInstant, customerZone)` display extension, with architecture as owner; a leader decision brief for the sponsor
Recipient	The sponsor (decision brief); the second line (stage-move evidence); the three other display teams (reusable assets)
Done when	The other teams reuse the evaluation set and the catalog entry without the enablement lead in the room

8. Hand-offs and separation of duties

You give → to whom	You receive ← from whom	You challenge	You are notified of
Leader decision brief → the executive sponsor	Funding and priorities ← the sponsor	Benefit numbers without a baseline	Incidents and kill-switch use
Stage-move check → the AI inventory and the second line	Exit evidence ← team leads	Teams certifying themselves	Model or provider changes
Learning paths → L&D and every team	Role change decisions ← the sponsor and HR	Training offered for a problem training does not fix	Policy changes
Reusable assets → the asset catalog	Asset feedback ← consuming teams	Assets published without owner, limits or evaluation	Exceptions and their expiry

Separation of duties. The enablement lead coaches a team and then checks its stage move. For medium and high tiers the second line checks too, so the coach is never the only check. The enablement lead never sets a risk tier or approves a release.

9. Evidence and records

Evidence	Produced or consumed	System of record	Retention owner
Stage entries in the AI inventory	Produced	AI inventory	The AI governance function
Acceptable-use attestations and scenario results	Produced with L&D	Learning system	L&D
Lab outputs and transfer checks	Produced	Program workspace	The enablement lead
Catalog entries	Produced; owned by the asset owner	Asset catalog	The asset owner
Scorecard	Produced	Metrics system	The enablement lead
Workforce communications	Produced with HR	HR records	HR

10. Skills and learning path

Level	What the enablement lead learns	Labs	Done when
Foundation	The two axes, the control crosswalk, decision rights	All six labs as a participant	You can explain each lab's governance note to a control partner
Practitioner	Running the coaching cycle; reading traces and evaluation reports	Facilitate Lab 1 and Lab 4	One team moves a phase with evidence
Advanced	Change management under job fear; checking Stage 4 entry; translating new capabilities	Co-run Lab 6 with an AI risk partner	The second line accepts your stage-move checks without rework

11. Measures and incentives

The enablement lead reports the program, so the anti-gaming rules apply here most. See the measure chapter.

Metric	Definition	Anti-gaming note
Learning transfer	People whose real work shows the skill after a lab, divided by lab completers	Attendance is not transfer
Sustained use	Teams still using a practice a quarter after coaching ended	Coaching that never ends hides dependence
Reuse rate	Changes that reuse or extend a catalog asset, divided by changes where one existed	A justified "create" is fine
Asset health	Assets with a named owner and an in-date review	Count retired assets as healthy, not as losses
Stage-move rework	Stage-move checks the second line sent back	Low rework because nobody looks is worse
Near misses reported	Count per quarter	Never a target to reduce

Incentives you shape. Ask leaders to count review, evaluation and catalog work as real work in goals. Give demo-day credit for useful failures and reused assets.

12. Cost, capacity and workforce

Cost or capacity item	What it covers	Illustrative figure
Program team	Enablement lead plus coaches, for a few hundred engineers	1 coach per 5 to 8 teams in coaching
Champions' time	One volunteer per team	10% of their week, protected
Learning time	Per person, scheduled	4 hours per month
Lab environments	Sandbox, synthetic data, approved tools	Platform budget line
Asset upkeep	Owner time for each shared prompt, evaluation set and lab	1 to 2 days per asset per quarter
Control partner time	Second-line time for intake and stage-move checks	Agreed turnaround targets

Workforce topic	What the enablement lead does
Job fear	Asks the sponsor and HR for a plain statement of how roles change and how saved time is used, then repeats it; never improvises it
Role redesign	Helps HR map changing tasks per role: more review, evaluation, domain rules and asset ownership
Learning equity	Makes sure every eligible person, not only volunteers, gets protected time
Champions' burnout	Rotates champions; ends the role openly when protected time disappears
Consultation	Involves HR and, where they exist, employee representatives before roles change

13. Anti-patterns and stop conditions

Anti-pattern	Why it fails	Instead
Attendance as success	Shows presence, not change in the work	Measure transfer and sustained use
Showcase tasks	Chosen to impress, not to learn	Pick a real task that already hurts
Program around governance	Teams learn that controls are optional	Every lab uses approved tools and data
Assets with no owner	They decay and teams stop trusting the catalog	No owner, no catalog entry
Promises about jobs	Broken promises end trust in the program	Repeat only what the sponsor and HR decided

Stop condition	What the enablement lead does
Reviewer capacity gone in a coached team	Pauses its delegated tasks; raises it with the engineering manager
Lab uses real or restricted data	Stops the lab; reports it as a near miss
Policy or provider change	Updates scenarios and labs before the next session
Asset owner leaves	Marks the asset unsupported; warns consumers

14. Action kit and 30/60/90

Stage-move check card. Use it before recording a stage move in the inventory.

## Stage-move check
Team / role / use case: __________________  Tier: ______
From stage ___ to stage ___   Proposed by: ______________
Exit evidence attached, with denominators:  yes / no
"Must exist first" items for the target stage, each tested:  yes / no
Reviewer capacity confirmed by the engineering manager:  yes / no
Second-line check needed (medium or high tier):  yes / no   By: ________
Step-down triggers known to the team:  yes / no
Recorded in the AI inventory on: ________  Review date: ________

Questions to ask yourself as enablement lead (the questions other tracks ask you)

Which teams moved up this quarter without a second-line check they needed?
Which shared assets would break if their owner left tomorrow?
What did the last "not a fit" result teach, and is it in the catalog?

The detailed first 90 days are in the enablement program chapter. From the role's seat:

Days 1 to 30	Days 31 to 60	Days 61 to 90
Goal	Listen and map	Prove with one team	Make it repeatable
Actions	Process walks; meet the second line, HR, L&D and internal audit; inventory gaps	Coaching cycle on one medium-tier task; first catalog entries with owners	Learning paths per role; champions named; stage-move check agreed with the second line
Evidence	Current-state map; stakeholder list	Baseline and results; stage-move check	Scorecard; leader decision brief
Watch for	Promising outcomes before a baseline	A showcase task	Scaling before reviewer capacity exists