How a wave of startups are betting on AI to make formal software verification practical and essential, including &Truth, which launches with a $70M seed

Formal software verification, once a niche and expensive process, is becoming significantly more practical and accessible for all developers.

A new wave of startups, including &Truth, Harmonic, and Antithesis, are leveraging generative AI to make formal software verification more accessible and practical. Formal verification uses mathematical proofs to ensure software is bug-free, a process historically too complex and time-consuming for widespread adoption. &Truth has launched with $70 million in seed funding, aiming to verify code in any programming language, a feat it demonstrated on Signal's encryption. Harmonic has secured $295 million and is testing its AI agent Aristotle with potential clients like Nvidia and Tensordyne. Anthropic and Theorem recently announced AI-assisted bug detection that surpassed trillions of traditional tests. These companies promise more reliable software and fewer security issues, though widespread adoption beyond critical systems remains uncertain as the technology is still in prototyping.

The integration of generative AI into formal verification is a watershed moment, lowering the barrier to entry for a process previously confined to high-stakes applications. Startups are automating the generation of specifications and translating code into verifiable formats, promising to democratize software reliability. This shift could fundamentally change how software is built, moving from extensive testing to rigorous mathematical proof as a standard practice.

The market is seeing substantial investment, with companies like &Truth and Harmonic raising significant capital, indicating strong investor confidence in AI's role in software assurance. The challenge now lies in customer adoption, pricing models, and ensuring AI-generated code does not introduce new, unforeseen vulnerabilities. The ultimate success will hinge on whether these tools can move beyond niche applications to become an integral part of the development lifecycle for a broad range of software.