How to run an AI ambassador program

An AI ambassador program is a network of trusted, visible peers whose job is to make AI use feel normal and safe across the company: they share wins in the open, tell the story of what changed in their own work, and lower the stigma for the people still hanging back. It is the advocacy layer, distinct from the technical support layer, and it exists because capability alone doesn't move a team. Harvard Business Review's 2026 research on AI rollouts found that when employees see trusted colleagues using AI on real tasks, sharing tips and discussing both wins and misses, AI starts to feel safe and relevant, and that peer influence can be more powerful than a top-down mandate. That is the whole thesis of an AI ambassador program: people don't adopt AI because leadership told them to, they adopt it because someone they trust made it look normal. Where this fits in the wider rollout is alongside the rest of your AI adoption effort, not bolted on after it.

The catch is that an ambassador is not the same role as the person who fixes your prompts, and conflating them is where these programs go soft.

The clean way to think about it: a champion is the help desk, an ambassador is the megaphone. Champions are picked for results, they answer 'how do I make this work,' fix the broken prompt, build the workflow, and you measure them by problems solved. Ambassadors are picked for reach, they make AI use visible and normal, recruit the hesitant, and you measure them by spread and sentiment. Champions create capability; ambassadors create permission. The reason the split matters is that most stalled rollouts have a perfectly good champion and a team that still won't touch AI, because using it openly still feels risky or like showing off. A champion shows you how; an ambassador makes you feel okay about trying. If you've already built out an AI champions program, this is the missing half, not a do-over, and on a small team one person may genuinely wear both hats.

The fair objection is that this is org-chart theater: you already told me to deputize my AI person, now you want a second title, second mandate, second set of meetings, for the same outcome. The honest answer is that the function is what's distinct, not the title. Most failed rollouts have a capable champion and a team that still won't touch AI, because capability without permission goes nowhere; people don't copy the power user, they wait to see if it's safe and sanctioned. You may not need two separate people or two programs, on a small team one person can carry both, but you can't skip the advocacy function. McKinsey found 48% of employees name training and support as the most important factor for successful AI adoption, and the ambassador is the visible, trusted face of that support across roles. Real networks bear this out: HSBC's group-wide AI ambassador network has signed up thousands of employees to drive education and adoption, which is the model working at scale. If your adoption is stuck despite having someone who knows the tools, you have a champion and no ambassador.