As businesses face challenges in integrating AI pilot programs into their operations, ensuring reliability has become crucial. A new startup, Pramaana Labs, aims to address this issue by leveraging mathematical formalization, merging the dependability of established computer science systems with the unpredictability of AI.
Pramaana Labs has secured $27 million in seed funding, led by Khosla Ventures and supported by investors such as Accel, BoldCap, Nexus Venture Partners, Premji Invest, and Unbound. The company is targeting sectors where accuracy is critical, such as law, drug discovery, and tax preparation, where mistakes can be costly.
The company’s approach involves using a conventional Large Language Model (LLM) enhanced with a deterministic verification layer. This setup ensures the accuracy of the LLM’s outputs. Pramaana Labs distinguishes itself by employing formal verification tools, inspired by the open-source LEAN programming language, which is used for verifying mathematical proofs.
For each specific application, Pramaana Labs will develop a tailored formal verification system, guided by experts in the field. For instance, in the realm of tax law, they are collaborating with former IRS commissioner Danny Werfel, while academic experts from IIT Delhi, IIT Madras, and UC Berkeley are contributing to the cybersecurity and drug discovery initiatives.
According to co-founder and CEO Ranjan Rajagopalan, the key to solving complex issues lies in formalizing them. He emphasizes that domains where errors can impact health, finances, or freedom are governed by rules that need to be codified into executable systems.
