Tell me your problems

Everyone tells you to solve a problem you actually have. It's the most repeated piece of startup advice there is, and it's repeated because it's true. The trouble is that it quietly assumes the problems will present themselves, that you'll be going about your life and the gap in the market will tap you on the shoulder. Six weeks into Singapore, sitting between building something of my own and the very real need to find a job, I've been finding that the real ones don't tap you on the shoulder if you're at a desk with the door shut.

There are two ways to come at an idea. You can go looking for it, which usually means sitting down with a notebook and trying to think your way to a problem worth solving. Or you can put yourself in places where ideas find you, which means being around other people who are building things and letting the problems surface on their own. I used to think these were roughly the same activity done in different rooms. They're not.

You can absolutely find problems at your desk. The desk is good at producing them. What it can't do is tell you which ones are real. Thinking your way to a problem mostly generates plausible-sounding ideas you have no evidence for, and you talk yourself into them because it would be convenient if they were true. The gap isn't between having a problem and not having one, it's between a real problem and one you've invented because it suits you. Sitting alone, you can't tell the difference.

The room does the work

What's actually moved me forward these last few weeks has been Hackapura, a builder community here in Singapore. I've been going regularly, and the thing that struck me first was how much you pick up just by seeing what other people are working on. Not in an abstract, inspirational sense, though it is genuinely inspiring, but in the specific sense that everyone building something is carrying around a list of problems they've hit, and when you get a room of them together those problems start spilling out.

Bouncing ideas off people who think the same way, giving feedback and getting it back, is a completely different mechanism to thinking alone. It's the difference between guessing what people want and watching what they're actually struggling with. The community here has been excellent for that, and it's reminded me that working out whether a problem is real is a social activity, not a solitary one.

This is worth separating from building in public, which I gave up on a while back. Posting progress online and sitting in a room with builders look similar on paper, but one is performing for an audience and the other is an actual conversation. Online I was optimising for engagement from peers who'd never be customers. In person you get the unfiltered version: the half-finished idea, the thing that didn't work, the problem someone can't stop complaining about. That's the raw material.

The bottleneck moved

The other thing I've been learning, almost in parallel, is just how much you can now get an AI agent to do. I've written before about wiring Claude into my analytics, but the last couple of weeks have gone further. The part that's genuinely changed how I think is scheduling: setting work up to run overnight, going to sleep and waking up to find a job has run and the results are waiting. There's something slightly absurd about getting things done while you're unconscious, and it takes a while to stop feeling like cheating.

These two threads connect more than they look. For most of my career the bottleneck on acting on an idea was building it. You'd find a problem, then spend three months and a lot of goodwill turning it into something real, which meant the cost of being wrong was high, so you were careful, so you moved slowly. That bottleneck is dissolving. With agents doing the grunt work and running while I sleep, the building is no longer the hard part.

Which throws the weight back onto the other side of the equation. If building is cheap, the scarce thing is knowing what to build, and knowing what to build comes from exposure, not introspection. The constraint has moved from "can I make this" to "is this real", and you can't answer the second question at your desk. You answer it by being in the room.

None of this makes the job hunt go away, and there's a particular flavour of uncertainty in trying to find a problem worth building while also needing the stability of employment. But the two aren't as opposed as they feel. Both are solved the same way: by getting out, meeting people, and letting the useful things find you rather than trying to summon them alone. Six weeks in, that's the bit I'm most sure of. The ideas are sitting in other people's frustrations and half-built projects. You just have to be in the room when they come up.