You know that feeling when a meeting ends and everyone's aligned, energised, and absolutely certain something needs to change, and then three months later nothing has? That's not a motivation problem. It's a method one.

Most organisations sitting on a product problem face the same bind. You know something needs to change. A customer need isn't being met, a revenue opportunity is going uncaptured. But you don't have the resources or buy-in to explore it properly.

AI tooling has changed this. One person can now take an idea from rough concept to working prototype in days, not sprints. The most talked-about version of this is vibe coding. The window between "what if we..." and "here, look at this" has collapsed.

But teams are still struggling to get useful outcomes from using AI in early experimentation because the tools have set up the wrong expectations. The newfound speed helps. But without clarity, speed isn't direction.

Over the last year, we've been running two-to-three day hackathons with client teams, using vibe coding to surface and stress-test problems and breakthrough ideas fast. Here's what we've learned about how to make the most of vibe coding in early product work.

A lot of teams are conflating the technology with the thinking, assuming the speed of these tools removes the need for rigour upfront. It doesn't. It amplifies whatever you bring to it.

Go in with a clear problem, a sharp brief, and a point of view on the opportunity, and you come out with something worth pursuing. Go in broad, with multiple vague directions across different industries, and you end up with a lot of impressive-looking prototypes that don't tell you anything.

We've seen both. Where the prep work hadn't been done, there was no clear brief, no defined opportunity areas, no examples of what a good idea might look like. Vibe coding just added more noise.

Running these hackathon sessions, we've come to see a more fundamental misunderstanding at play. The value hasn't been in the prototyping. It's been in the product thinking that precedes it: interrogating the problem, challenging assumptions, finding the idea that's worth making real.

We've found a two-part structure works best:

1. Get the right people in the room

That means your product and design people, but also the subject matter experts and, wherever possible, real users. Set aside existing assumptions. Generate distinct concepts across format, tone and mechanic. The ideas that survive this stage do so because of the quality of the thinking behind them, not because they were quick to build.

The sessions where real users or domain experts are present in the room produce something qualitatively different. When the people you're designing for are part of the process, hearing ideas, reacting in real time, pushing back, you stop designing for an imagined user and start designing for an actual one. The insights that come from that are not things you can synthesise from a brief.

2. Bring AI into the process

Once you've identified the ideas worth pursuing, you use the tools to give them real shape fast. Working prototypes, tangible concepts, something you can actually put in front of people. The strongest ideas get made real. The weaker ones get eliminated on evidence, not instinct.

What you walk away with isn't just a prototype. It's a clear point of view on which idea to back, early signal from real customers, and a concrete recommendation on where to start. That last part matters as much as the idea itself.

What we've heard back from the people who've been through these sessions is consistent. After two days, one client told us she was pleased they'd still done rigorous product thinking. She hadn't been sure what to expect from a session with AI in it, but was relieved the thinking hadn't been shortcut away.

That's the version we’ve seen lead to the best results: not AI instead of good process, AI that makes good process faster.

The organisations that will get the most from AI in product development aren't the ones moving fastest. They're the ones who are clearest about what the thinking is for, and disciplined enough not to skip it.

Have an idea worth pressure-testing? We're running three-day AI-powered hackathons that help product and innovation teams get clarity on whether they have an opportunity worth pursuing before committing to a full discovery. Email us at hey@planes.agency to find out more.