After AI training, almost every organisation sees the same thing: enthusiasm is high in the first week, and a few weeks later use returns to its old level. The main reason is that employees have no one nearby to ask when they get stuck.
AI champions fill exactly this gap. This article explains how to build a champions network, what champions should do, and how to run the first 90 days.
Who is an AI champion?
Microsoft's adoption guidance describes AI champions as trusted changemakers who experiment with AI, demonstrate its value in practice and make adoption approachable within their organisation. The role combines three things: the curiosity of an early adopter, the hands-on experience of a power user and the influence to inspire change.
The same guidance makes an important point: champions don't replace IT, compliance or the training team. They complement them, applying organisational guidance to the realities of daily work.
Why it works
- Proximity: Employees would rather ask a colleague doing the same job "how did you do that?" than write to a central support team.
- Context: A champion knows the unit's work. The finance champion talks about report commentary, the HR champion about job ads.
- Continuity: Training lasts a day; a champion is there every day.
- Feedback: Champions bring back what works and what doesn't on the ground.
In its own internal adoption programme, Microsoft uses a hub-and-spoke structure: the central change team offers a single front door to all resources, and the champions community directs employees to the content most relevant to their work.
Step 1: Choose your champions
A good champion is not always the most technical person. Look for:
- Curiosity and willingness to experiment: Not afraid to try new tools.
- Respect within the unit: Colleagues already ask them for advice.
- Willingness to share: Enjoys explaining what they know.
- Knowing the work well: Can connect AI to the unit's real work.
- Respect for rules: Sets an example of safe use.
Tip: Keep it voluntary, but get the unit manager's approval too. A champion without manager support won't find the time.
Step 2: Put the role and time in writing
If championing stays "a favour you do in your spare time", it fades within weeks. Write a short role description:
| Topic | Content |
|---|---|
| Responsibilities | Q&A support in the unit, sharing good examples, proposing new use cases, escalating issues to the centre |
| Time | Weekly time agreed with the manager |
| Boundaries | Applies the rules, doesn't change them; routes data and security questions to the right unit |
| Support | Advanced training, early access, monthly champions meeting |
| Visibility | Contribution featured in internal communications and performance reviews |
Step 3: Train champions first
Champions should receive training before other employees, and in more depth:
- Advanced use of approved tools and prompting techniques
- The usage policy: what data goes into which tool, who to ask in which case
- Spotting and verifying faulty outputs
- How to help colleagues: short demos, trying things together, sharing examples
Step 4: Build a community
Champions left on their own burn out. Set up a regular structure that brings them together:
- A shared channel: Teams, Slack or an email group for questions and examples.
- A monthly champions meeting: New features, questions from the field, success stories.
- A shared example library: Use cases and example prompts organised by unit.
- A direct line to the centre: One person through whom champions can reach IT, security and compliance quickly.
The first 90 days
| Period | What happens |
|---|---|
| Weeks 1–2 | Identify champion candidates, get manager approval, write the role description |
| Weeks 3–4 | Champion training, open the shared channel and example library |
| Weeks 5–8 | Champions take part in unit training; weekly Q&A sessions |
| Weeks 9–12 | Share examples champions collected, first measurement, review the network |
How do you measure the network's success?
- Compare approved tool use in teams with and without a champion
- Number and topics of questions champions receive (where do people struggle most?)
- Use cases proposed by champions that went live
- Agreement with "I know who to ask when I get stuck" (in a short survey)
For the full measurement approach, see our article on how to measure AI adoption.
Common mistakes
- Choosing champions only from IT: Champions who speak the business units' language are more effective.
- Not giving them time: Without a role description, champions go quiet within weeks.
- Not recognising them: Invisible contribution doesn't last.
- Making champions rule-makers: Champions apply rules and bring feedback from the field; they don't make policy decisions.
Conclusion
A champions network is one of the most effective ways to turn AI training from a one-off event into a lasting habit. The right people, a written role, dedicated time and a regular community; the champions do the rest.
We build the champions network together in the implementation phase of our Corporate AI Adoption Programme, and after the programme our Ongoing Support module provides coaching and regular Q&A sessions for champions.
Sources