Most organisations have invested in generative AI tools and run an "AI champions" programme. Adoption and ROI remain inconsistent. The pattern is familiar.
A small group becomes power users. The majority dabble, then plateau. Risk behaviours vary by team. Leaders cannot confidently quantify impact or explain what "great" looks like.
The gap sits one level up from the tools. AI enablement needs to be treated as a company capability, comparable to leadership development, security awareness, or management fundamentals: three connected layers, not one.
Why a champions programme alone is not enough
Champions are useful. They translate generic guidance into examples that make sense for Finance, People, Sales, Engineering, Legal. They create social proof. They surface friction early. A good champions programme accelerates adoption.
What a champions programme cannot do, on its own, is produce sustained capability. Champions rotate. Their priorities shift. The teams around them go back to old habits the moment the champion is occupied with something else. Without an underlying system, the wins evaporate.
Champions are a distribution layer, not a strategy. Build the backbone underneath them and the same champions stop restarting from zero every time someone rotates off the programme.
The three layers
1. Org-wide enablement
This is the backbone. It includes:
- Standards and guardrails. What is approved, what is forbidden, what requires review. Plain English. One page.
- Tooling strategy. Which models are licensed, which workflows are blessed, where the data lives.
- Capability expectations. What every employee, every manager, every leader is expected to be able to do with AI. Tied to the levelling framework, not bolted on as a side initiative.
- Governance. Decision rights for who can deploy what, where, with what oversight.
- Measurement. Adoption, quality, outcomes. Reported openly, monthly.
- Incentives. Promotion criteria, performance reviews, recognition rituals all reflect AI-fluent behaviour.
Org-wide enablement is the layer most companies skip. It is unglamorous. It does not produce a launch moment. It is also the layer that decides whether the rest of the work compounds or evaporates.
2. Team-wide enablement
This is where the operating model meets the work. The unit is the team, not the individual.
- Workflow transformation. Each team picks its top three workflows and rebuilds them with AI in the loop. Take onboarding: redesign the whole flow so AI sits at the moments that actually slow people down, day one, week two, first review, rather than being bolted onto induction as an add-on. The before-and-after is visible enough that new hires notice it themselves.
- Shared assets. Prompts, Custom GPTs, automations, agents. Owned by the team, versioned, shared.
- Cross-functional process ownership. Workflows that cross teams (offer-to-onboard, hire-to-pay) get an explicit owner who is accountable for the AI-enabled version.
- Operating cadence. Weekly demos. Monthly retros. Quarterly outcome reviews. The rituals make adoption visible and create the feedback loop that drives improvement.
Team-wide enablement is where the value shows up. It is also where most champions programmes fall short, because they leave the team without the structures to sustain the work.
3. Individual enablement
The final layer is the human. Without individual capability, the other two layers are theoretical.
- Persistent workspaces. Every employee has an AI workspace with reusable context, memory, and shared prompts. This is foundational. Without it, every AI session starts from zero.
- Reusable context. People know how to load context efficiently, write good prompts, evaluate outputs. The basics covered in prompting patterns.
- Output standards. What "good" looks like by task type. Models are not the source of standards. The team is.
- Evaluation habits. Critique passes are normal. Nobody ships a first draft.
Individual enablement is the layer most companies do invest in, often the only one. It is necessary and not sufficient.
Sequencing
The temptation is to do all three layers at once. Do not. The operating cadence that works:
Months 0 to 3, prove value fast.
Pick three workflows with clear outcomes. Onboard the first cohort of champions. Publish a one-page AI policy. Stand up a weekly demo. Build a simple metrics pack. Ship something visible. Resist the urge to write the strategy deck before the first thing ships.
Months 3 to 6, scale the wins.
Extend the workflow set. Launch the team-wide enablement layer in two to three pilot teams. Add the operating cadence. Tie capability expectations to the levelling framework. Begin the governance work in earnest.
Months 6 to 12, embed.
Org-wide standards published and enforced. Team-wide enablement live across most of the function. Capability expectations baked into hiring, performance, and promotion. Measurement now routine. Champions programme becomes a feeder for builders, not the only mechanism.
The order matters. Org-wide standards drafted in month one will be wrong in month four. Team-wide rituals introduced in month one will get gamed before they produce value. Individual enablement that runs ahead of team rituals will create power users who feel unsupported.
Move in sequence. Let each layer earn the next.
Behaviours that scale
Underneath the three layers sit a small set of leadership behaviours that decide whether the operating model takes root.
- Context first. Translate every People decision into commercial terms.
- Speed with safety. Ship in slices, capture learning, add controls.
- Show the work. Prompts, evaluations, outcomes are visible by default.
- Coaching over policing. Teach managers to think with AI, not just to monitor compliance.
- Transparency. Employees know what is automated, why, how it is monitored.
The signal a leader is living it: they can explain any People decision in two sentences, in commercial terms, and they volunteer examples of safe automation their team shipped this month. The anti-signal: tool-first, problem-second, pilots that never end, dashboards without decisions.
Where this lands
The CHRO who owns this work is doing something specific: building a system that turns AI into a sustained capability of the function, then the company. Licenses and training programmes are inputs to that, not the deliverable.
That work has the same shape as any other operating system work, which is why it sits inside the broader frame of an AI operating system and a wider People Ops domain map. The diagnosis comes first, see diagnosing AI readiness, and the practice underneath is the move from prompts to systems.
Get the operating model right and the champions programme finally does what it was always meant to do: distribute capability the system already supports.
What this connects to
Auto-recommended next reads in the People Ops cluster, ranked by shared concepts and headings:
- Leading the AI transformation in People
- From prompts to systems
- The champion model
- Prompting patterns for People Ops
Common questions
- Why does a champions programme alone not produce sustained AI adoption?
- Because it has no way to survive a champion moving on. The tell is that adoption tracks the champion's calendar, not the team's workflow: move them onto a new project and usage drops within a quarter, because nothing structural was holding it up. Standards, guardrails, governance, and incentives are what keep capability in place after the individual has left the room.
- What are the three layers of an AI enablement operating model?
- Org-wide is the unglamorous backbone: standards, guardrails, tooling strategy, capability expectations, governance, measurement, incentives. Team-wide is where the work actually happens: workflow transformation, shared assets, an operating cadence. Individual is the human layer: workspace, context, output standards, evaluation habits. Skip one and the other two stall. Rituals with no org backbone give you inconsistent standards team by team. Individual skill with no team cadence gives you isolated power users the rest of the function can't learn from.
- What does "good" look like in the first 90 days?
- Three to five use cases with a number attached to each: hours saved, error rate, cycle time. One published policy page. A weekly show-the-thing cadence nobody needs reminding about by month two. The real test: could a sceptical manager on a completely different team name one thing that's changed. If the answer's no, you've built a project, not a capability.
- Whose job is this?
- The CHRO or CPO. AI enablement is fundamentally a people system: expectations, capability development, incentives, performance, progression, culture, trust, change management. IT can deploy tools. Only People leadership can standardise behaviours and embed AI capability into how work is done.
When reading turns into doing
The Grain Audit maps one People Ops process end to end, ranks the highest-return automations, and hands you a 90-day plan you keep whether or not we work together.
Two weeks. GBP 2,000, credited in full against a programme. Three slots a month.
Book a Grain AuditIf this resonated, there's more.
Subscribe to receive new Intelligence pieces as they're published. No noise, just the work.
By subscribing you agree to our Privacy Policy. Unsubscribe any time.
