What actually counts as validation
Sign-ups aren't proof and interest isn't proof. Here's the test I use now, and what it looked like when I applied it to my own product.
Guides
Everything here comes from eleven years of building and selling software products, including one expensive failure. It's the part that doesn't change month to month: how to find out whether anyone will pay before you build, and how to plan a product clearly enough that AI can actually build it.
These used to be part of a course. They're not a course anymore. Every guide stands on its own, nothing is locked, and you don't have to start at the beginning.
Sign-ups aren't proof and interest isn't proof. Here's the test I use now, and what it looked like when I applied it to my own product.
How to get from one real problem to 10 paying customers without writing a product. Written for people building software for other businesses.
Your 20-page document will fit, and the AI will build something. The problem is everything it decides for you along the way, and what that costs you later.
The build notes
I'm building and selling my own products alongside client work. Every decision that costs real money or real time goes into an email: what I chose, what I passed on, what it cost, and what actually moved the numbers.
I don't send it on a fixed schedule. I send one when there's something worth sending.
Free · one click to leave · nothing else
Peter Bartfai designs and builds software products in Budapest — for clients during the week, and for himself after hours.