AI reliability startup Pramaana Labs has raised $27 million in seed funding led by Khosla Ventures, with participation from Accel, BoldCap, Nexus Venture Partners, Premji Invest, and Unbound. The company is applying formal verification techniques to large language models, targeting industries where AI errors carry serious legal, financial, or medical consequences.
Co-founder and CEO Ranjan Rajagopalan described the core insight as recognising that domains such as tax law, drug discovery, and legal work are governed by dense but codifiable rules — and that once those rules are formally encoded, AI reasoning built on top of them can become deterministic rather than probabilistic. Pramaana’s architecture runs a conventional LLM for natural language flexibility while layering a deterministic verification system above it to validate outputs, drawing on the open-source LEAN programming language used to verify mathematical proofs.
For each vertical, Pramaana builds a bespoke formal verification system overseen by domain experts. Former IRS Commissioner Danny Werfel is advising on the tax law system, while professors from IIT Delhi, IIT Madras, and UC Berkeley are guiding work on cybersecurity and drug discovery applications.
The approach draws precedent from France’s CATALA project, which formalises the country’s tax and benefit code into executable logic. Pramaana’s ambition is to extend that methodology across any domain where errors in AI output could cost people their health, finances, or legal standing.
