Pre-launch · $299 one-time
VowAdvisor
One shared wedding plan for two: timeline, budget, vendors, and decisions, with an advisor grounded in the plan that proposes changes for confirmation.
See the product →Month 01 · Shipped
I run a tiny AI factory.
One new product a month, priced like I am not trying to squeeze anybody, with the real numbers published either way. This is where the work gets written down — including the parts that go badly.
The streak
The board starts empty.
Twelve cells, one per month. Each one ends up holding a product or an honest note about why it does not. You are looking at month one, so there is nothing to hide behind yet — which is exactly why it goes up now instead of once it looks impressive.
Jul 2026
Shipped
FeatureJet
Aug
In flight
VowAdvisor
Sep
Oct
Nov
Dec
Jan 2027
Feb
Mar
Apr
May
Jun
- In flight
- Shipped
- Open
Updated when something actually ships, not on a schedule.
The products
Two products so far.
Everything gets built on the same rails, so a new bet costs a month rather than a rewrite. Everything gets priced the same way too.
Live · $15 flat
FeatureJet
Feedback boards for small teams. Your users post what they want, vote on it, and hear back when it ships. One flat price — no per-seat meter that punishes you for growing, and no basics held hostage in a higher tier.
See the product →How a month goes
The factory, in four moves.
The agents do a lot of the typing. The judgment is mine, and I would rather show you exactly where that line falls than let you assume it sits further out than it does.
Pick the bet
Every candidate gets a one-page intake: who it is for, what it costs to run, and the conditions under which I would kill it. If I cannot write those three things down, it does not get built.
Write the spec
Agents do not read minds. The spec is where the judgment lives — what the product refuses to do, which edge cases matter, what done means. This is the part I never delegate.
Let the factory build
Agents write the implementation, the tests and most of the review. I read the diffs, break the ties, and own whatever ships. When it is interesting, the transcripts get published too.
Ship it, then post the numbers
Launch, then a recap with the real figures. Ninety days and one paying customer, or the product gets retired — and the retirement gets written up the same way a launch would.
The writing
Hype without numbers, or numbers without reasoning.
That is most build-in-public writing, and it is hard to learn anything from either half. I am trying to do the third thing: the decision, the number it produced, and what I would do differently. Three threads run through all of it.
The streak
What shipped this month, what slipped, and what it cost to find out.
The numbers
Monthly recaps with real figures. A bad month gets the same table as a good one.
The reasoning
Why flat pricing, why this bet, why that agent setup fell over. Enough detail to steal.
Posts land when the work does. No content calendar, and no newsletter you have to escape from.
Not greedy. Not corporate. Built for joy.
I am not trying to squeeze a small team for every possible cent. Flat prices. No seat metering, no basics locked in a higher tier, no “contact sales” where a number belongs. I build because making something people enjoy using is the point, and pricing is just another place to say that out loud.
The rule every product here gets held to