Pramaana Labs, a startup focused on improving AI reliability, announced $27 million in seed funding led by Khosla Ventures. The round also included participation from Accel, BoldCap, Nexus Venture Partners, Premji Invest, and Unbound. The company aims to address the challenges of deploying AI in highly sensitive sectors where errors can have severe consequences. According to the startup, its approach combines conventional large language models with deterministic verification to ensure accuracy and reliability in critical applications.
Pramaana’s system uses formal verification tools, specifically the open-source LEAN programming language, to codify rules and ensure deterministic outcomes. The company plans to build domain-specific verification systems for areas such as tax law, cybersecurity, and drug discovery. For tax law, Pramaana is collaborating with former IRS commissioner Danny Werfel, while experts from IIT Delhi, IIT Madras, and UC Berkeley oversee cybersecurity and drug discovery systems. Rajagopalan emphasized that the world’s hardest problems are not unsolvable but require formalization to ensure correctness and safety.
The startup’s approach is inspired by projects like France’s CATALA, which formalizes tax and benefit systems into executable code. Pramaana’s goal is to apply similar principles to AI systems, ensuring that the rules governing these domains are codified and verified. This method aims to reduce errors and increase trust in AI applications across industries where reliability is paramount.
Source: techcrunch
