Startups Race to Use AI for Formal Software Verification as &Truth Launches With $70M
AI-assisted formal verification cut one engineering project from five months of work by three experts to under half a day for less than $500. Stanford researcher Chuyue Sun’s team also used AI to verify part of Signal’s encryption code.
A wave of startups is betting that generative AI can make formal software verification — a mathematical process for proving code is bug-free — practical enough for widespread use. The most recent entrant, &Truth, publicly launched in October 2026 after raising $70 million from investors.
Formal verification has existed for decades but has remained largely confined to high-stakes environments at organizations like NASA and Apple. The process requires engineers to write precise statements about their software’s intended behavior and convert code into mathematical formulas that can be validated — work that is time-consuming and expensive. Companies typically rely on conventional testing instead.
Stanford researcher Chuyue Sun, who joined &Truth in August 2026, co-authored research showing what AI assistance could do for the field. Her team completed an engineering project in under half a day for less than $500 — work that had previously taken three skilled experts five months to do manually. Sun’s research also used AI to verify part of the encryption code in the messaging app Signal.
“Bridges, buildings, they don’t collapse because people did math before constructing them,” said Ilya Sergey, &Truth’s chief science officer. “Why don’t we do the same for code? The technology to mathematically reason was all there, just difficult to scale.”
&Truth aims to eventually verify any programming language. Rival startup Harmonic, co-founded by Robinhood CEO Vlad Tenev and backed by $295 million in funding, announced this week that its AI agent Aristotle can verify software written in C and hardware designs in Verilog. Harmonic is evaluating the technology with chipmakers Tensordyne and Nvidia, whose venture capital arm is an investor.
Also last week, Anthropic and verification startup Theorem announced they used AI to find bugs in open source software that trillions of conventional tests had missed, and introduced an add-on for Anthropic’s Claude model to help developers with verification work.
Whether AI can bring formal verification into mainstream software development beyond the most critical systems remains an open question.