Pramaana Labs raises $27 million（約43億円） to add mathematical verification to AI systems used in law, drug discovery, and tax preparation, where errors are costly.

What happened: Pramaana Labs announced $27 million（約43億円） in seed funding led by Khosla Ventures, with participation from Accel, BoldCap, Nexus Venture Partners, Premji Invest, and Unbound. The startup will use formal verification tools—drawing on the LEAN programming language used to verify mathematical proofs—to add a deterministic verification layer on top of conventional LLMs, checking the AI's reasoning before it reaches users.

Why it matters: Enterprises are struggling to turn AI pilot programs into reliable business tools, especially in high-stakes fields where errors are costly. Pramaana's approach addresses this by formalizing the rules of domains like tax law and drug discovery into executable code, making the AI's reasoning deterministic and verifiable rather than opaque. This addresses a core pain point: current AI systems lack sufficient protection against hallucinations and errors in sensitive verticals.

What to watch: Pramaana is building domain-specific verification systems, starting with tax law (working with former IRS commissioner Danny Werfel) and cybersecurity and drug discovery (overseen by professors from IIT Delhi, IIT Madras, and UC Berkeley). The company's model requires codifying the rules of each industry before verification can work, making execution and domain expertise the critical path forward.