The 12-month-before-it's-obvious window.

There's a window I've started thinking about a lot. It's the gap between "an early signal exists that this matters" and "the market generally understands that this matters." For most things, that gap is roughly twelve months.

Operating inside that window — building, publishing, teaching — is, in my experience, the highest-leverage thing an independent technical person can do right now. It's not that the work itself is uniquely hard. It's that the market has not yet allocated attention to it, so the work travels much further per unit of effort than it will twelve months later.

A few things I've noticed about this window:

You can't ride more than two at once

Each "12-month window" you choose to operate inside demands real depth — you have to actually build, ship, and learn the unsexy parts. If you try to operate in more than two of these at once you'll be a tourist in all of them, and the market punishes tourists.

Pick two. One adjacent to your existing depth (so you can move fast). One a stretch (so your career still has compounding optionality). Refuse the rest, even when they're interesting.

The signal you're looking for is "weird builders are spending time on this"

Not "VCs are funding this." VCs are 6-12 months early on hype and 6-12 months late on durable shifts. The signal that matters is when the kind of person who has built three useful things in the last two years quietly starts spending their weekends on something. That person is usually right about where the puck is going, and they're early in a way that's actually useful to be early about.

Track ten of those people. Notice what they're tinkering with. Notice what they're frustrated by. That's your window.

You'll feel dumb for about four months

The first four months of operating inside a real 12-month window feels exactly like wasting your time. The build is hard. Nobody is talking about it. The audience that will eventually care doesn't exist yet. The audience that does exist is asking you why you're not doing something more obvious.

This is, ironically, the strongest signal that you're in the right window. The discomfort is the entire reason the opportunity exists. If it felt comfortable, everyone would already be there, and there would be no window.

Publish from inside the window

The single highest-leverage thing you can do during this period is publish what you're learning, on whatever surface fits your voice, before you have it figured out. The "I built this thing and it works" post is fine. The "I'm building this thing and here is what is currently broken" post is what compounds.

You want to be the person the market remembers as the one who was already there when the rest of them arrived. Publishing from inside the window — not after — is what creates that memory.

There is, right now, a 12-month window open in voice agent infrastructure. There is another in agent-native operating systems. There is a smaller, weirder one in synthetic-media pipelines for one-person teams. I'm operating in two of those three. I am ignoring the rest.

If you're a builder, the only question that matters this year is which window are you in.

Pick one. Stop reading thinkpieces. Go build.