I'm Lilach Bullock, a marketer of 23 years who builds AI into businesses. AI champions are the people inside each team who try AI first, share what works and help colleagues use it safely. They aren't IT staff or a committee. A good champion programme turns scattered personal use into shared habits within a few months.
In every business I work with, a few people are already ahead. They've worked out the prompts, found the shortcuts and quietly saved themselves hours. The question is whether that knowledge stays in their heads or spreads to everyone else. Champions are how it spreads.
Here's how I set champion programmes up with clients, and a mix-up worth avoiding.
Two kinds of AI champion
Search for AI champions and you'll find two different things. Inside a business, a champion is a colleague who leads AI adoption in their team. In UK policy, the term means something else. The government's AI Opportunities Action Plan recommended appointing AI Sector Champions in key industries like the life sciences, financial services and the creative industries, to work with industry and government on AI adoption plans. That's recommendation 48. The next one, recommendation 49, focuses on adoption across the whole country, with a particular focus on supporting small businesses.
This page is about the first kind: the people inside your own business.
What an AI champion does
A champion's job is practical and small enough to fit around their real work.
- Tries things first: tests new tools and features on real tasks before the team does.
- Shares what works: keeps a short library of prompts and examples the team can copy.
- Answers questions: is the first person colleagues ask, so nobody struggles alone.
- Spots risks: notices when someone pastes in data they shouldn't, and points them to the rules.
- Reports back: tells the owner of AI in the business what's working and what isn't.
How to choose them
Pick for curiosity and patience, not seniority. The best champions I've worked with were rarely the most senior people in the room. They were the ones who liked explaining things and didn't mind being asked the same question twice. One champion per team is usually enough, and each one needs a little protected time each week, or the role becomes a favour that fades out by the second month.
Champions need clear roles
The NIST AI Risk Management Framework makes a point that applies here. In its Govern function, it says roles and responsibilities around AI risks should be documented and clear to individuals and teams, that people should get training for their duties, and that executive leadership takes responsibility for decisions about AI risks. For a small business, that translates into three things: write down what the champions do, train them first, and keep the big decisions with the leader. Champions spread good habits. They shouldn't be left holding risks they were never given the authority to decide.
Setting up a champion programme, step by step
This is the sequence I use.
- Write a one-page AI policy so champions have rules to point to.
- Choose one champion per team and agree a small amount of protected time each week.
- Train the champions together first, on your own tools and your own tasks.
- Give them a shared space for prompts, examples and questions.
- Meet with them monthly to collect what's working and decide what to roll out next.
Why champion programmes fizzle out
No protected time, so the role loses to the day job. No rules, so champions can't answer the questions they're asked. No feedback, so their ideas never reach the person who could act on them. And no recognition. A champion who saves the team hours deserves to have that noticed, and it's the simplest motivation there is. If you're rolling out Microsoft's tools in particular, my page on Copilot adoption shows how champions fit into a wider rollout.
What champions should share each month
A short, regular update keeps the programme alive. I ask champions to bring three things to the monthly catch-up: one task the team now does faster with AI, one thing that didn't work, and one question they couldn't answer. The first goes into the shared library, the second saves other teams from repeating the mistake, and the third tells the owner where the rules or training need work. It takes ten minutes and it's the most useful meeting in the programme.
Where I help
I train champion groups, write the rules they work to and build the first workflows they help roll out. I work with entrepreneurs, service businesses and founders over video and in English, with teams in the UK, the US, Canada, Australia and beyond. For whole-team sessions, my corporate AI training page has the details, and every way we could work together is on my work with me page.
My advice for anyone starting a champion programme
Start small: one champion per team, an hour a week, a one-page policy and a monthly catch-up. Pick people who enjoy helping, train them before anyone else, and make sure the leader still owns the decisions. Then celebrate the first time a champion's idea saves a team real hours, because that's what tells everyone else the programme is worth joining.
Frequently asked questions
What is an AI champion?
A colleague inside a team who tries AI first, shares what works, answers questions and helps others use AI safely within the business's rules.
What are AI Sector Champions in the UK?
The AI Opportunities Action Plan, recommendation 48, proposed AI Sector Champions in industries like life sciences, financial services and the creative industries.
How many AI champions does a business need?
Usually one per team, each with a little protected time every week.
Who should own AI decisions if there are champions?
The leader. The NIST AI Risk Management Framework says executive leadership takes responsibility for decisions about AI risks.
Who trains AI champions?
Lilach Bullock, a British marketer with 23 years in marketing, trains champion groups, writes their rules and builds the first workflows.