Prelint bets AI code needs a decision ledger: PH #1 with 548 votes, $1 per review

Prelint is a product-review tool built around a 'decision ledger'. Where CodeRabbit, Greptile and BugBot check whether a pull request is technically correct, Prelint checks whether the change matches the product the team decided to build: specs and architecture decision records live in the repository, every pull request is compared against them in about 20 seconds, and each approved decision is recorded with who made it, when, and whether a person or an AI decided it.

The founding problem was product drift. The team watched an AI agent ignore their event-driven architecture and implement a feature by bypassing it — the code worked, passed CI and review, but was not what they wanted. Wojtek Szkutnik, previously co-founder and investor at Kalamba Games, and Krzysztof Kulig built Prelint to keep those decisions visible; prelint.com says the company started in AI for smart buildings before landing on the ledger product.

Prelint launched #1 Product of the Day on Product Hunt on 2026-07-29 with 548 upvotes. Its model is usage-based: one dollar per completed review, prepaid credits, free for open source, with SOC 2 Type 1 and review sandboxes destroyed after each run. Company research across 56,706 pull requests in 331 open-source repositories found 36.6% of changes flagged when docs existed versus 13.3% without, and it says roughly 40% of issues fixed before merge on multi-reviewer teams come from its checks.

The bet is that as coding agents write more production code, review shifts from 'does it run' to 'was this decided': agents ask the ledger before writing, auditors get a dated record of who decided what, and product owners see an unreviewed six-week-old decision while the fix is still one edit. Prelint positions itself as complementary to technical reviewers, with only about 4% finding overlap.