As businesses grapple with the uncertainties of artificial intelligence (AI), Pramaana Labs is stepping in to bring mathematical rigour to the table. The startup has secured $27 million from tech giants like Khosla Ventures, aiming to make AI more reliable and error-proof.

Their approach? Combining cutting-edge formal verification techniques with natural language processing models (LLMs). This means that for industries such as law or drug discovery, where mistakes can be costly, AI decisions will undergo a deterministic check, ensuring they align with predefined rules.

For instance, Pramaana is working with former IRS commissioner Danny Werfel to formalise tax laws and collaborating with academic experts in cybersecurity and drug development. By translating complex regulations into executable code, the company hopes to make AI decisions more transparent and trustworthy.

Ranjan Rajagopalan, co-founder and CEO of Pramaana Labs, stresses that many of the world's hardest problems are simply unformalised. His vision is to codify these rules, making every domain where errors could lead to severe consequences a safer place for AI intervention.