level: business
pramaana labs announced a $27 million seed round led by khosla ventures, with backing from accel, boldcap, nexus venture partners, premji invest, and unbound. the startup aims to make ai reliable enough for sensitive areas such as law, drug discovery, and tax preparation. these fields have strict rules, and errors can lead to serious consequences. the company believes formal verification can prevent mistakes that large language models often make.
the system uses a conventional large language model for flexibility in handling natural language and complex problems. on top of that, a deterministic layer checks the model's work using formal verification tools. pramaana relies on the open-source lean programming language, which is used to verify mathematical proofs. this approach is inspired by projects like france's catala, which turns tax and benefit rules into executable code. domain experts help build the verification systems for each use case.
co-founder and ceo ranjan rajagopalan says many hard problems are not unsolvable but unformalized. the company works with former irs commissioner danny werfel for tax law, and professors from iit delhi, iit madras, and uc berkeley for cybersecurity and drug discovery. by codifying rules, pramaana hopes to make ai reasoning deterministic and trustworthy in high-stakes environments.
why it matters: formal verification could reduce ai errors in critical applications, making models safer for legal, medical, and financial tasks.