The trust layer for AI-built code.
AI made it possible to build software by describing what you want. Subak exists to answer the question that comes next: does the code do what you meant?
Building got easy. Trusting the result didn't.
A whole generation of people now ships software without reading the code, on Replit, Lovable, Claude Code, and whatever comes next. The demo works, so it goes out. What's missing is a way to know whether it holds up before real users find out first.
Subak closes that gap. It reads generated code against what you set out to build and shows you, in plain language, what's solid and what's about to break.
Subak is a Balinese tradition of shared trust.
For over a thousand years, Balinese rice farmers have shared water through subak, a cooperative system that keeps every terrace fairly fed through transparency and mutual trust. We took the name because that's the job: bringing shared, verifiable trust to the code you didn't write yourself.
Principles we build on
Proof over promises
Anyone can say code looks fine. We show what was checked and how, so trust rests on evidence, not a green light.
Plain language, not jargon
You shouldn't need to read the code to understand the risk. Every finding is written so you can act on it.
Built for the people building
Freelancers, small teams, solo founders. Subak is for the people shipping with AI, not the ones auditing it after.
Know what you're shipping.
Join the early access list and be first to check your AI-built code with Subak.