Building internal champions and an AI council

Top-down AI mandates produce compliance. Peer-driven adoption produces belief. Internal champions are how the second happens.

Who makes a good champion

Not the most senior person. Not the most enthusiastic person (enthusiasm alone breeds cynicism). The right profile:

• Respected by peers (not necessarily liked).
• Technically capable enough to use AI tools credibly.
• Honest about limitations — doesn't overhype.
• Willing to invest time beyond their day job.
• Good at translating — can explain AI value to non-technical colleagues.

Often mid-level individual contributors or senior ICs, not managers.

How to find them

• Observe usage data. Who's using AI tools voluntarily and effectively?
• Ask around. "Who on your team is playing with AI?" usually produces names.
• Open calls with real incentives (time, recognition) attract self-identified candidates.
• Pilot volunteers — people who raised their hand for early trials.

Avoid anointing champions top-down. "You're now the AI champion" without buy-in fails.

The ask

Clearly scope what the role involves:

• Time commitment: typically 2-5 hours/week.
• Activities: lead office hours, write internal docs, train colleagues, report back on friction.
• Duration: 6-12 months (rotational makes sense).
• Recognition and support: explicit; this isn't "extra" work piled on.

Without clarity, champion roles become volunteer burnout.

The AI council

For larger orgs, champions should organize:

• Monthly cross-team meetings.
• Shared Slack channel for ongoing questions.
• Quarterly reporting to leadership.
• Rotating presentation — each champion shares wins and challenges.

This creates a peer network; champions support each other; best practices spread organically.

What champions actually do

Week to week:

• Help colleagues get unstuck. Specific questions, specific answers.
• Share patterns that work. "Here's a prompt that nailed it for me."
• Flag friction. Report issues to the AI program team.
• Demo real usage. Show-and-tell in team meetings.
• Advocate for governance changes. "This approval process is blocking adoption."

They're the translation layer between the program and the users.

What they shouldn't do

• Be the support desk. Sustainability means users learn to self-serve.
• Be unpaid consultants. Their time needs recognition.
• Defend every vendor decision. Independence from the procurement side matters.
• Be the only voice. Multiple champions per area; no single point of failure.

The power of visible example

Nothing beats "a peer I respect uses this and says it works" for adoption. Champions provide this at scale.

Practical patterns:

• Live demos in team meetings. Real work, real tools.
• Internal write-ups of how champions use AI in their daily work.
• Pair sessions where champions sit with colleagues learning.

These are worth more than any top-down training program.

Compensating champions

Options:

• Recognition. Internal awards, visibility to leadership.
• Training budget. Send them to AI conferences.
• Protected time. Explicit 10% of their week for the role.
• Stipends. Real money for real extra work.

At minimum, recognition and protected time. Paying for genuine extra work is appropriate when asks are substantial.

The anti-pattern: AI evangelist

Beware the person who:

• Promotes every new tool.
• Overstates AI capabilities.
• Dismisses concerns.
• Dominates conversation.

They create skepticism, not belief. You want credible enthusiasts, not zealots.

Sustaining the network

Champion programs decay without intentional sustenance:

• Refresh the roster annually — some cycle out, new ones come in.
• Keep leadership engaged — quarterly exec meetings with champions.
• Celebrate success stories publicly.
• Document institutional knowledge — when a champion rotates out, their knowledge stays.

The honest limits

Champion networks aren't sufficient alone. They complement:

• Central program management (strategy, vendor relationships, governance).
• Training infrastructure (formal learning, not just peer).
• Support systems (real helpdesk, not champions alone).

Champions accelerate adoption. They don't replace the operational backbone.