Why I built this
I'm a principal engineer at a large e-commerce company. A meaningful part of my job is writing decision documents, and every one of them starts with reading material I can't take at face value.
I used to spend half of a day fact-checking a single vendor whitepaper before I'd cite it in my work. So I built a pipeline that does the tedious part: pull out every claim, research each one, show me the evidence. I've used it for my own decisions for over a year. Assay is that pipeline, cleaned up so you can use it too.
It's a small product built by one person. It charges from the first document because each analysis runs frontier models hard but you'll always see a full example analysis and the exact price before you commit a cent. If it saves you one bad architecture bet, it's paid for itself for years.
If it saves you one bad architecture bet, it's paid for itself for years.
— Dmitry
What I care about
Evidence over confidence
Verdicts are evidence-backed judgments, not a manufactured certainty score. Every one links its sources so you can disagree with it.
Honest about unverifiable
When no independent source exists, Assay says so instead of guessing. Knowing what rests on the author's word alone is half the value.
Cite everything
No verdict without a citation. The evidence trail is part of the deliverable, so you can check the checker in minutes.
Small by design
Built and run by one engineer. No growth targets to hit, no roadmap shaped by anyone but the people actually using it.
Who it's for
Staff and principal engineers reading and writing RFCs, ADRs, and build-vs-buy recommendations. Architects evaluating vendors. Anyone whose name goes on a decision document that someone else will read in two years and judge.