I Didn't Build Products. I Built the Thing That Builds Products.

Give me a capable AI and a clear afternoon and I turn into a machine for starting things. Over these six months I started something new most fortnights: a landing page here, a working prototype there, a whole product sketched, costed, and half-built before the week was out.

Most of them are parked. If you’re keeping score the traditional way, that’s a lot of unfinished business. I want to argue it’s the opposite, because the thing that made all of it possible is the only part that really matters.

The valuable thing wasn’t a product

Early on I stopped building things and started building the means of building things.

First came a library of skills: repeatable recipes the AI could follow for jobs I do over and over, so I never had to explain from scratch again. Then, on top of those, agents that wield the skills without me in the loop for every step: something that picks up the next job, something that turns a rough task into a proper brief, batches that grind through work while I sleep.

That’s the move, and it took me a while to see it for what it was. The leverage isn’t in any single prototype. It’s in collapsing the cost of the next one. Once the means of building are built, “let’s try X” stops being a fortnight of effort and becomes an afternoon. When starting gets that cheap, you start more. Of course the chart looks manic.

So the real question stops being “what did you ship” and becomes “what can you now ship on a whim.” That second number is the one I actually grew.

The bar for “worth building” just dropped

Here’s the second-order effect, and it’s the one that will genuinely change how I work from here. Every backlog has a graveyard of things that were never worth building: too fiddly, too heavy, too niche, the little jobs that save ten minutes a week for a fortnight of effort. The sums never added up, so they sat there forever.

Collapse the cost of building and the sums change. A whole pile of “too low-value to bother” quietly flips to “why not, it’s an afternoon.” Work that was permanently parked, not because it was useless but because it was never worth the cost, is suddenly worth it. That isn’t the same work done faster. It’s a change in what even counts as a sensible thing to attempt, and it’s the shift I’ll be exploiting long after this sabbatical ends.

A few things reached daylight

Not everything stayed a sketch. A handful reached real, working function: one taken from nothing to taking live payments inside a week; one I genuinely use every single day; one that quietly solved a business problem I’d been paying for in time for years.

What separated those from the rest had nothing to do with how good the idea was. It was whether the thing scratched an itch I actually had that week. The ones I finished were the ones I needed. The rest were interesting, and interesting is a much weaker force than necessary.

Worth remembering next time you feel guilty about a stalled side project: it probably stalled because you didn’t need it, not because you’re lazy.

Most stayed ideas, and that’s fine

The long tail is a stack of one-line concepts, captured and left alone. A name, a sentence, a spark, filed and forgotten until it isn’t.

For years I’d have called that a graveyard. I’ve stopped. An experiment’s whole job is to answer a question cheaply, and a parked idea has already done its job the moment it’s out of my head and costing nothing to hold. Holding it costs nothing precisely because I built somewhere frictionless to put it.

When starting is cheap and holding is free, “did you finish it” is the wrong question. The pile isn’t a graveyard, it’s a library. A hundred sketches I can pull off the shelf the day one of them suddenly matters, at no cost for the wait. Most will never be needed. I keep them all anyway, for the handful that will.

So what did six months of this actually buy?

One thing that lasts, and it isn’t a product. It’s a capability.

I walked in able to advise people on building software and products. I’m walking out able to build them myself: quickly, on my own, with the means of building already sitting on my desk. The experiments were the tuition. The engine is what I keep, and it doesn’t reset when the sabbatical ends.

I set out to make things. What I actually made was a machine for making things, and a habit I have no intention of giving back.

Posts in this series

Working LifeIdeas and Adventures