How to Build an AI Champions Program: A 90-Day Guide

After corporate training sessions, the question I hear most often is this: "The training was great, but how do we keep this energy going three months from now?" The answer is usually not another training session. It is a few people inside the teams who actually use AI in their own work. We call them AI champions. In this article I explain what an AI champions program is, how to choose champions, where their role starts and ends, and how you can set up the program in 90 days.

What is an AI champion?

An AI champion is a volunteer employee who stays in their own team, uses AI tools in their daily work and helps colleagues adapt these tools to their own tasks. A champion is not IT support and not a trainer. They are a colleague who knows the team's work and, when asked, can show how they prepare a given report with AI. An AI champions program is the structure through which these people are selected, supported and connected to each other.

Why central training alone is not enough

Good training builds awareness and basic skills. But when employees get back to their desks, their questions are very concrete: "How do I use the tool when replying to this customer objection?" or "Am I allowed to give it this spreadsheet?" A central team or an outside trainer cannot answer these questions every day for every department. Someone inside the team can, because they do the same work.

The second reason is trust. People adopt a new way of working fastest when they see someone who does the same job actually using it. In my article on why employees don't use AI, champions were step four of the 90-day plan. This article covers that step in detail. In that plan, champions come in during the second month. If you are setting up a dedicated champions program, I recommend selecting them earlier and training them before their teams, which is why the timeline below starts sooner.

How do you choose a champion? Five criteria

1. Volunteering: Champions apply or are nominated, not assigned. Someone forced into the role soon sees it as an extra chore.
2. Knowing the work: They should know how the team's work gets done and be someone people listen to. They don't need to be the most technical person. They need to be the one the team trusts.
3. Curiosity and willingness to share: Someone who is happy to show what they tried and to say openly what did not work.
4. Responsible use: Someone who respects the data rules and does not send work out without checking the output. The champion sets the example for the team.
5. Manager approval and time: The champion's manager has to agree to set aside time for the role. Without dedicated time, the role fades within a few weeks.

My recommendation is to start with one champion per team, or one for every group of 15 to 25 people. Having someone from every department matters more than the total number.

What a champion does, and what they don't

Champions programs most often break down because the role stays vague. If champions are held responsible for everything, they can't get anything done. Define the role in writing from the start.

What a champion does

• Finds and demonstrates useful AI examples for 2 or 3 of the team's recurring tasks.
• Adds prompts that work to a shared library.
• Serves as the first stop for colleagues' day-to-day questions.
• Reports back to the central team how the rules are understood in practice and where people get stuck.
• Makes the team's good examples visible in a short monthly sharing session.

What a champion does not do

• Technical support: Licenses, access and setup issues belong to IT.
• Monitoring: A champion does not report on who uses the tool and how. If the team sees them that way, people stop bringing them questions.
• Writing the rules: Management and the relevant departments set the usage rules. The champion explains them to the team and passes feedback back up. For what employees can and cannot share, see data rules for employees using AI.

Setting up the program in 90 days: six steps

1. Write down the role and the rules (week 1): Put the champion's responsibilities, the time they will set aside and the approved tools on a single page. Without rules, a champion gets stuck on the first question.
2. Select the champions (week 2): Pick one person per team through applications or manager nominations. When you announce the selection, say clearly that the role is voluntary and focused on support.
3. Train the champions first (weeks 3-4): Champions should get deeper training before their teams do: writing prompts, checking output, data rules and examples from their own department. When defining the baseline, the AI literacy framework is useful.
4. Pick 2 or 3 tasks per team (weeks 4-5): Each champion picks 2 or 3 recurring tasks with their team and records, in a short note, how long these tasks take today.
5. Start the rhythm (month 2): Champions meet briefly with each other every two weeks, and each team holds a monthly sharing session. Prompts that work go into the shared library. For a ready-made starting point, you can adapt the examples in my prompt library to your own work.
6. Measure and decide (end of month 3): Measure the selected tasks again, roll out what works to other teams and drop what doesn't. Replace champions who don't want to continue with new volunteers.

What should champions get?

If the champion role is just an honorary title, it fades within weeks. For the program to last, champions need:

• Dedicated time: A set amount of time each week, agreed with their manager and not simply added on top of their workload.
• Deeper training and early access: The chance to try new features and tools before the rest of the team.
• A direct channel: A person or team they can reach quickly when a rule is unclear.
• Ready-made materials: Short explanations, sample prompts and a one-page rules summary they can use in team meetings. Champions should not have to build everything from scratch.
• Visibility: Their team's best examples are credited to them by name in management meetings. Recognition motivates people as much as money does.

Champion, central team and manager: who does what?

Don't underestimate the manager's part. However motivated the champion is, if the manager doesn't use the tool and never asks about it, the team gets the message. As a starting point for managers, I recommend my AI guide for managers.

How do you know the program is working?

"How many champions do we have" is not a success metric. Look at these questions instead:

• How does weekly active use in teams with a champion compare with teams without one?
• How do time spent and revision rates on the selected tasks compare with the baseline?
• Are new prompts added to the shared library every month, or did it stop after the first month?
• Who do employees ask about the rules, and how quickly do they get an answer?
• How many champions want to continue in the role after three months?

To assess these indicators together with your training investment, you can use the before-and-after approach in how to measure the impact of corporate AI training.

Five common mistakes

• Choosing the most technical person: The champion's job is adapting AI to the team's work, not technical support. The person the team trusts is more effective than the most technical one.
• Not setting aside time: If the role becomes after-hours volunteering, it ends at the first busy period.
• Turning the champion into a monitor: Nobody brings questions to a champion who is asked to report on usage.
• Leaving the rules to the champion: The champion does not decide which data can't be entered. Without rules, the team either doesn't use AI at all or moves to personal accounts, which is known as shadow AI.
• Announcing the program and walking away: Without regular meetings between champions and a channel to the central team, each champion ends up on their own island.

Conclusion

What makes AI adoption last is not one big training session but a few volunteers inside the teams who know the work. Champions with a clearly written role, dedicated time, training of their own first and support from their manager are the ones who carry what was learned in training into daily work. To train your champions and teams on your own real work, take a look at corporate AI training, and to design the program for your company together, see AI consulting.