Why Fun Projects Are the Fastest Way to Learn AI

Robbie is an AI-powered robot. She has speakers, a microphone, and cameras in her eyes that can take pictures and video. Behind the scenes, the voice belongs to Roberta — but in public, Robbie is the face. She’s connected to a large language model, has memory, and can run programs. For the Congress, we’re setting her up to remember associations she meets and what they’re doing with AI (if you share!). She even knows where she is and can hold a real conversation (that’s the plan anyway… technology gremlins notwithstanding).

Robbie is also a fun project. And that’s the real point.

If you want to learn AI properly, you probably need to stop trying to learn these new tools, and frontier AI uses, at work in your day job.

That sounds counterintuitive. Most organisations are pushing their teams toward AI tools, AI training, AI adoption. But learning on the job is hard. There’s politics. There’s pressure to get it right first time. There’s the day job competing for attention. There’s the awkwardness of experimenting with tools you don’t fully understand in front of colleagues who are watching.

The result? People play it safe. They stick to the basics. They use AI for drafting emails and summarising documents — useful, but shallow. The deeper capabilities stay unexplored.

We’ve found a different approach works better: fun projects.

Projects that have nothing to do with work. Projects where nobody’s watching. Projects where you can break things, try stupid ideas, and push into territory you don’t understand — without consequences.

Robbie started as one of those projects. She combines hardware, large language models, memory, cameras, and integrations into a working interactive experience. Building her taught us things we would never have learned in a work context — about connecting systems, handling real-time interaction, managing complexity, and iterating fast when things break.

None of that was the goal. The goal was: this seems fun, let’s see if we can do it.

But here’s what happens with fun projects: the skills transfer. The patterns you discover messing around on a weekend reshape how you think about business problems on Monday.

That’s why we build fun into our training programmes too.

When we run in-house AI training, we don’t just teach tools. We create space for people to play — to experiment with projects that feel low-stakes and interesting, not high-pressure and work-critical. It unlocks capabilities people didn’t know they had. Once they’ve built something fun, they start seeing what’s possible everywhere else.

Robbie is one of a long line of fun projects we’ve done. Each one sharpens skills that come back into client work, consulting, and training — sometimes indirectly, sometimes directly. In this case, directly: you can meet Robbie yourself in Faro.

Now we’re not programmers. We didn’t write a single line code for this project, and the many others we have worked with this past year. AI helped us build. Actual solutions. 

The distance between having an idea and having a working prototype has collapsed. More people can now create things that would have been impossible five years ago.

So come and meet us at the Association World Congress - Mark, Stephanie amd our embodied AI teammate. We are, at least for now, still the puppet masters behind the puppet!

And the takeaway? The best way to understand AI isn’t to read about it, or sit through training, or wait for permission.

It’s to find something fun and start building.

Mark will be speaking on the Keynote panel, and both Mark and  Stephanie will be at their Fresh Solutions AI stand in the exhibition throughout the Associations World Congress.