Agital’s specific answers to the questions most AI change management programs leave vague.

Most AI ambassador programs fail in the same way. They launch with energy – a kickoff call, some sign-ups, maybe a Slack channel – and then slowly lose momentum.

The reason isn’t bad intentions. It’s that most programs are well-intentioned (“we want people who use AI and spread enthusiasm”) without getting specific enough about the mechanics that make that real. Who’s eligible? Who decides? What are ambassadors actually accountable for, and to whom? What does success look like in 90 days?

Agital, a 325-person digital marketing agency, has been unusually specific in designing their program. Here’s what they got right – and what any organization serious about enterprise AI adoption can take from it.

What is an AI ambassador program?

An AI ambassador program is the human layer of AI adoption inside an organization. A group of employees – drawn from across the business, not just IT – are trained and empowered to drive AI literacy, surface high-value use cases, and help their teams build real AI habits.

Think of them as the connective tissue between an organization’s AI strategy and the people who actually have to change their behavior for that strategy to work. Done well, it’s one of the most effective tools in AI change management. Done poorly, it’s a title on a Confluence page that nobody reads.

Who owns the AI ambassador program?

As the de facto “Head of AI” at Agital, Matt Heffernan owns the program - but is running it alongside his regular job as CIO. He says in order to succeed, the program will need its own dedicated leader – which he also recommends for any enterprise taking AI deployment seriously.

“Every ‘part-time’ Head of AI should make it clear to leadership: if you want me to do this, I need to hire a full-time person or bring in a partner like Section. You don’t want to be hung out to dry because you wear 12 hats and have no time.”  — Matt Heffernan, CIO, Agital

The implication is worth sitting with. If your AI ambassador program doesn’t have a named internal owner with real time attached to it, it’s not really a program. It’s a hope.

How AI ambassadors are chosen

Agital’s selection process solves a tension most programs never address: how do you keep it open enough to be inclusive, and structured enough to actually work?

Their answer: open eligibility, managed endorsement.

At Agital, anyone can volunteer to be an AI ambassador. No experience required. The logic is cultural as well as practical. If ambassadors are all highly advanced users, the implicit message is that AI transformation is just for super users. Plus, very advanced AI users are sometimes less helpful to their peers – they’re way further ahead and have spent hours on AI projects most people only have minutes for.

But open eligibility doesn’t mean everyone gets the role. The process runs in two steps:

Step 1:  Anyone interested puts their name in.

Step 2:  Candidates must be nominated and endorsed by their direct manager, and ultimately by the division president.

The two-step structure matters because it ensures the manager is committed to supporting the candidate’s 10-20% time allocation. Manager buy-in isn’t a nice-to-have. It’s the accountability mechanism the whole thing depends on.

What AI ambassadors are actually expected to do

Matt laid out exactly what’s expected. The specificity is the point – vague expectations are how ambassador programs quietly die.

Here’s his list:

1. Think “AI first” and help others to do the same
2. Have clearly defined roles and outcomes-based goals (defined by each ambassador)
3. Be endorsed and supported by leadership and equipped to succeed
4. Measure the use of core tools and training
5. Initially focus on AI literacy and enablement, then high-value use cases and workflows
6. Audit existing use cases and share with others
7. Lead engagement events like prompt-a-thons, hack your job contests, etc.
8. Be the first to test new tools when Agital is considering them
9. Operate as a part-time role – not a dedicated one (10-20% of an employee’s time)
10. Use a crawl-walk-run approach to AI adoption

Point 5 is where most enterprise AI training programs go wrong. Ambassadors need to start with AI literacy and enablement – getting people comfortable with the tools, building daily habits, reducing the psychological friction of “I don’t know how to use this.” High-value use cases come after that. You can’t push someone toward a sophisticated AI workflow when they’ve barely logged into ChatGPT.

This also ensures AI upskilling happens in the right order. Ambassadors who skip straight to advanced use cases tend to lose the 80% of their colleagues who aren’t there yet.

How ambassador goals get set — and by whom

In most ambassador programs, goals come from above. Agital rejected that model explicitly.

“The people who define those goals are the ambassadors themselves and their line managers – not me, not the AI steering committee, and not anyone on the executive leadership team.”  — Matt Heffernan

The performance framework includes a discrete AI component built into each ambassador’s goals, but the exact goal is determined by the ambassador and their manager together. Two things are expected of everyone: demonstrating active, daily AI usage, and identifying at least one meaningful AI use case that is manager-approved.

Goals set by the people closest to the work are more likely to be pursued – and more likely to surface the use cases that actually matter to each part of the business. That’s what good AI change management looks like in practice: not mandates from the top, but accountability structures that let the right behavior emerge from the middle.

What you can learn from Agital

Agital got specific where most ambassador programs stay vague. Five lessons worth learning:

1. Sequence the role.  Literacy and enablement first. High-value use cases second. Ambassadors who try to do both at once do neither well.
2. Open eligibility, managed endorsement.  Anyone can raise their hand, but manager and division-level endorsement is required. This keeps the door open culturally while building accountability structurally.
3. Embed AI in performance management.  Requiring a discrete performance goal for ambassadors keeps them accountable. What gets measured gets done.
4. Let ambassadors own their goals.  Goals set jointly with line managers – not handed down from the AI team – are more likely to be pursued and more likely to surface the use cases that actually matter.
5. Give someone real ownership above the program.  Ambassadors without an internal owner will drift. The program needs a home with real time and real accountability attached to it.

‍