Writing
On building with judgment
Your Replit app is making money. Now what?
If your Replit app is turning into a business, it currently hangs by a thread: one platform, one login, one database you don’t fully control. Here’s how to fix that without rage‑quitting the platform or rebuilding from scratch.
Read →The $300 prototype that never shipped — and what we’ve learned from it
A founder spent $300 and two weeks on an AI builder and never shipped. The replies weren’t pushback — they were “same here.” Here’s why this pattern keeps happening, and what breaks it.
Read →When a vibe-coded prototype needs real architecture
There is a specific moment when "it works" stops being enough. Learning to recognize that moment before it becomes expensive is one of the most useful skills a non-technical founder can build.
Read →The most common security mistakes in AI-generated MVPs
Most security holes in vibe-coded apps are not exotic. They are the same five or six patterns, reproduced at speed by tools that have no stake in what happens after you ship.
Read →The difference between code that runs and code you can trust
A working application and a trustworthy one are not the same thing. Understanding the gap — and why it is invisible to most builders until something breaks — is where judgment begins.
Read →