Building an internal AI community: why some take off and others fade

Communities run on different fuel

Most of what happens at work runs on structure: roles, goals, deadlines, someone who decides. Communities don't. You can't order people to trust each other, share what they know or care about a shared goal. You can only build the conditions where those things happen on their own.

That's what makes communities so powerful, and so easy to get wrong. We've all seen the network that was launched with a nice kickoff, met three times and quietly faded. Usually nobody did anything wrong. The group just skipped a step.

Right now, organizations everywhere are learning this the hard way with AI. The tools have been rolled out, the trainings held. Some people have taken off, while most are still wondering where to start. What helps them isn't another course. It's a colleague who shows them how they did it, in their own work, in their own words. That's a community job, and it lands on the desks of HR and people teams more often than on IT's.

The Community Pyramid

Jussi Tapio and I have used one model for years to understand why some groups grow into real communities and others don't. We call it the Community Pyramid. It has six building blocks, and they're built in order:

1. Benefit. People come because they get something out of it. That's not cynical, it's how people spend their limited time.

2. Trust. Personal bonds: knowing each other a little, and believing the others mean well.

3. Sharing. People start giving what they have, including the things that didn't work. This is where a group starts creating value on its own.

4. Purpose. A shared story of where we're going and why it matters.

5. Ownership. Members take responsibility for what the community does. They stop waiting for instructions.

6. Identity. Being a member becomes part of who people are at work.

   
   

The first block makes an audience. Trust and sharing turn it into a network. Only purpose, ownership and identity make it a community.

That explains a common disappointment: a group with benefit, trust and sharing is a good, active network. But it won't take ownership or survive a busy season, because those blocks haven't been built yet.

Three lessons for building an internal AI community

1. When something stalls, look one level down. If people won't take ownership, check whether they trust each other and share a purpose. If sharing has dried up, check whether people still get something out of it. The problem is rarely where the symptom shows.

   
   

2. Start with what's in it for them. In the beginning, people are a little selfish. They join because they want something for themselves, and that's fine. But once trust and sharing are in place, something shifts: the members start creating the benefit for each other. That's when the whole thing gets off the ground.

   
   

3. Giving ownership away is the hardest step. Whoever starts a community has to let its members decide where it goes, and it will probably go somewhere they didn't plan. Take the reins back at that moment, and the community never becomes self-sustaining.

You can try this tomorrow: take a group you know well, whether a network, a team or an ambassador program, and go through the six blocks. Which is the lowest one that's weak? That's where to start, however far up you were hoping to be.

The book: The AI Community

I've packaged everything I know about building an internal AI community into a short, practical guide: The AI Community: A Practical Guide to Spreading AI Adoption from Colleague to Colleague.

You don't need to be an AI expert to read it. It's a book about people: how to choose the right kind of community, who to invite (hint: not your most advanced users), how to climb the pyramid one block at a time, what leadership gets from it and what it owes in return, and how to keep the community alive when the novelty wears off.

It's for anyone responsible for helping people adopt new ways of working, including HR and people teams, community builders, and the ambassadors who carry the change.

The book is available as a paperback and on Kindle: get The AI Community on Amazon.

Let's talk

I'd love to hear how this looks from where you sit. Are you building a community around AI, or around something else entirely? Which block of the pyramid gives you the most trouble?

If you'd like to discuss any of this, send me an email or connect with me on LinkedIn if we're not already connected.