What Senior Leaders Actually Need From an AI Hackathon

I recently ran an AI hackathon for a group of senior leaders at a large enterprise. These weren't developers. They were managing directors, heads of operations, compliance leads — people who make decisions about whether AI gets adopted, and how much budget it gets.

The most important moment wasn't when the tool produced something impressive. It was when someone said: "Wait, I just built that?"

The demo problem

Most executive AI sessions are demos. Someone from the technology team — or worse, a vendor — shows the tool doing impressive things. The executives watch, ask a few polite questions, and leave thinking "that's clever" without any visceral understanding of what it means for their teams.

Demos create spectators. Hackathons create believers.

The difference matters because AI adoption decisions aren't rational. They're emotional. A senior leader who has personally experienced the moment where AI produces something useful in two minutes that would have taken their team two hours — that person doesn't need a business case. They've felt it. They'll fund it.

Building beats watching

Here's what we did: instead of showing the executives what AI could do, we gave them a problem and the tools to solve it themselves. Not a toy problem — a real scenario relevant to their business. And not a simplified interface — the actual tools their teams would use.

The first twenty minutes were messy. People who hadn't typed a prompt before were fumbling. Some got frustrated. A few were quietly convinced it wouldn't work.

Then someone got a result. Not a perfect result — a rough, slightly wrong, but recognisably useful result. And suddenly the room shifted. Because what they'd just experienced wasn't "AI is clever." It was "I can do this." Those are fundamentally different realisations.

Chat vs Code: knowing your audience

There's a crucial design decision in any executive hackathon: what tools do you put in their hands?

For non-technical leaders, a conversational interface is the right starting point. The barrier to entry is low — you type in natural language, you get something back. The feedback loop is immediate. You can iterate by saying "make it shorter" or "add the compliance angle." It feels like a conversation, not like programming.

The temptation is to jump straight to more powerful tools — code-based AI, automation platforms, agent frameworks. Resist that temptation. The goal isn't to make executives into developers. It's to give them an accurate mental model of what AI can do, so they make better decisions about where to invest.

Once they've felt the conversational interface work, you can show them what the technical tools look like — not to train them, but to close the gap between "this is magic" and "this is engineering with specific capabilities and limitations." A senior leader who understands that gap makes better strategic decisions than one who thinks AI is either omnipotent or useless.

What they actually take away

The executives who build something themselves leave with three things that no slide deck can provide.

First, calibration. They now know — from experience, not from a vendor pitch — roughly what AI is good at and where it falls over. They've seen it produce something useful and they've seen it produce nonsense. That calibration is worth more than any analyst report.

Second, vocabulary. They can now have a real conversation with their technical teams. Not because they learned the jargon, but because they've experienced the workflow. They know what a prompt is because they've written one. They know what hallucination means because they've seen it happen.

Third, urgency. Not the manufactured urgency of a keynote speaker saying "AI will transform everything." The genuine urgency of someone who has just realised that a task their team spends forty hours a month on could be done in minutes. That urgency drives budget, hires, and strategy in a way that presentations never will.

Design for the "wait, I built that?" moment

Every executive hackathon should be engineered toward one moment: a senior leader looks at something useful on their screen and realises they made it themselves, in minutes, with no technical background.

Everything else — the agenda, the tool selection, the scenario design, the facilitation — exists to make that moment happen. Because once it does, you don't need to sell AI adoption. They'll sell it for you.