You cannot mandate AI adoption. You can mandate training attendance, tool subscriptions, and process changes. But you cannot mandate enthusiasm, curiosity, or the daily habit of reaching for AI before doing something manually.
That is why top-down AI rollouts stall. The leadership team gets excited, buys the tools, runs the training, and three months later most of the team has gone back to their old workflows. We covered the reasons for this in our piece on why agency AI pilots fail. The solution is not more training. It is champions.
What an AI champion is (and is not)
An AI champion is someone in each team or department who leads AI adoption from within. They are not a full-time AI role. They are a designer, a copywriter, an account manager, or a project manager who happens to be enthusiastic about AI and willing to help others get there too.
They are not:
- A replacement for proper training
- An unpaid extra job dumped on a willing person
- The person who fixes every AI problem in the agency
- A gatekeeper who approves AI usage
They are:
- The first person to try new tools and report back
- The person who shares what is working (and what is not)
- A peer trainer who shows colleagues how to use AI in their specific role
- A bridge between leadership’s AI ambitions and the team’s daily reality
How to select your champions
Do not ask for volunteers and do not appoint the most senior person. Look for these qualities:
Curiosity over seniority. Your best champion might be a mid-level creative who has already been experimenting with AI on their own time. Seniority does not predict enthusiasm. Often the opposite is true, because senior staff have the most to unlearn.
Credibility with peers. The champion needs to be someone their team respects and listens to. If they recommend a workflow, people will try it. If they are seen as an outsider or a management proxy, adoption stalls.
Practicality over theory. You want someone who builds things, not someone who reads about AI trends. The best champions are the ones who show up with a working prompt template and say “try this on your next brief.”
Aim for one champion per team or department. In a 20-person agency, that is probably 3-4 champions. In a 50-person agency, 5-7.
What champions actually do
Give them a clear, manageable scope. This is not a vague “promote AI” mandate. It is a defined set of activities:
1. Test and evaluate tools. When a new AI tool appears (and they appear constantly), your champion tests it against real agency work. They spend an hour with it and report back: does it work, is it better than what we have, is it worth the cost? This saves the whole team from tool fatigue and shiny-object syndrome.
2. Share wins publicly. Every week, champions share one thing that worked. A prompt that saved time, a workflow that improved quality, a tool that solved a problem. This goes in Slack, in a team meeting, wherever people will see it. These small, specific wins do more for adoption than any training session.
3. Train peers informally. Not workshops. Not presentations. Sitting next to someone and showing them how to do something. “Here, let me show you how I use AI for that.” Five minutes of peer coaching is worth an hour of formal training. This is the mechanism that makes training stick beyond the initial session.
4. Flag issues and risks. Champions are also your early warning system. They spot when AI is being used inappropriately, when outputs are not being reviewed, or when a tool’s quality has dropped. This feeds directly into your AI governance framework.
5. Collect feedback. Champions hear what the team actually thinks about AI, the frustrations, the fears, the requests. They pass this upward so leadership can adjust the approach. Without this feedback loop, you are flying blind.
How to support your champions
This is where most programmes fail. Agencies appoint champions and then give them nothing. No time, no budget, no recognition. Here is what champions need:
Protected time. At least 2-3 hours per week dedicated to AI experimentation and champion activities. This is not on top of their existing workload. It replaces something. If they do not have time, they will not do it. Put it in their calendar and protect it.
A small budget. Give each champion £50-£100 per month for AI tool subscriptions and experimentation. This lets them test tools without going through a procurement process every time. The return on this investment is enormous.
Access to leadership. Champions need a direct line to whoever is driving AI strategy in the agency. A monthly 30-minute check-in where they share what is working, what is not, and what they need. This keeps champions connected to the bigger picture and gives leadership ground-level intelligence.
Recognition. Publicly acknowledge what your champions are doing. Mention their contributions in all-hands meetings. Include AI adoption metrics in their performance reviews. If being a champion is invisible work, people will stop doing it.
A peer network. Connect your champions with each other. A monthly champions meetup (even 30 minutes) where they share discoveries and solve problems together. This prevents isolation and accelerates learning across teams.
How this scales adoption
The maths is straightforward. If you train 25 people in a workshop, maybe 5 will change their behaviour. That is a 20% adoption rate. With champions, each one influences 4-6 people through daily proximity and peer demonstration. Four champions in a 25-person agency can reach everyone within a few weeks.
The difference is not just reach. It is sustained adoption. Workshop knowledge fades. A champion sitting three desks away who keeps showing you better ways to work does not fade. They are a permanent, embedded source of momentum.
This is exactly the approach that turns AI from a one-off experiment into an AI-first culture where using AI is simply how work gets done.
Common mistakes
Making it mandatory. If someone is forced to be a champion, they will do the minimum. Only appoint people who genuinely want the role.
Overloading them. Champions are not your AI department. They have a day job. If champion duties start taking more than 3-4 hours a week, you need more champions or a dedicated AI operations role.
Ignoring their feedback. If champions report that a tool is not working or the team is struggling, act on it. Ignoring feedback is the fastest way to lose your champions.
No programme structure. “Just go and promote AI” is not a programme. Define activities, cadence, and success metrics. What does a good month look like for a champion?
Measuring success
Track these metrics quarterly:
- Tool adoption rate: percentage of the team actively using AI tools (not just having accounts)
- Workflow integration: number of documented AI-assisted workflows in use
- Time saved: estimated hours saved per week across the team
- Champion activity: number of tips shared, peer training sessions run, tools tested
You are not looking for perfection. You are looking for a steady upward trend. If adoption is growing quarter over quarter, your programme is working.
Starting this week
Pick your first champions. Have a conversation with each one. Explain what you are asking, what you will give them in return, and what success looks like. Set up a Slack channel for champions to share discoveries. Schedule the first monthly check-in.
It is that simple. The hard part is not the programme. It is the consistency of supporting it month after month. But if you do, you will build something that scales AI adoption faster and more sustainably than any amount of top-down mandates.
This is part of Delivery Notes, a series on implementing AI inside your agency. Subscribe to the newsletter to get new articles weekly.
