For makers and artists, the era of trusting a black box to handle high-stakes logic is ending. While the creative community grapples with generative tools, a new wave of AI is being built for the exact opposite: environments where a single hallucination could cost millions, invalidate a legal brief, or compromise patient safety. Pramaana Labs is betting that the answer lies in marrying the flexibility of large language models with the rigour of mathematical formalisation.
The $27M bet on certainty
On Wednesday, Pramaana Labs announced it has secured $27 million in seed funding. The round was led by Khosla Ventures, with backing from Accel, Boldcap, Nexus Venture Partners, Premji Invest, and Unbound.
The startup is targeting highly sensitive verticals where errors are not just annoying but catastrophic: law, drug discovery, and tax preparation. In these sectors, reliability is non-negotiable. Deploying standard AI here requires a robust defence against the hallucinations that plague current systems. According to Pramaana co-founder and CEO Ranjan Rajagopalan, these domains are uniquely suited to formalisation.
“It’s like math in the sense that you have a lot of rules that you need to abide by,” Rajagopalan told TechCrunch, using the tax code as an example. “Once you have a codified version of it, the reasoning on top of it starts becoming deterministic.”
How the hybrid system works
Pramaana’s architecture retains a conventional LLM at its core. This engine provides the flexibility to answer natural language queries and solve complex problems that traditional computers cannot. However, a deterministic verification layer sits on top, ensuring the LLM’s output is logically sound before it is accepted.
This hybrid approach is gaining traction, but Pramaana distinguishes itself by utilising formal verification tools based on the open-source LEAN programming language, typically used to verify mathematical proofs. The concept is not entirely new; Rajagopalan cites France’s CATALA project, which has already formalised much of the nation’s tax and benefit system into executable code.
For every specific use case, Pramaana will construct a bespoke LEAN-style verification system, supervised by domain experts. For tax law, the company is collaborating with former IRS commissioner Danny Werfel. Meanwhile, academics from IIT Delhi, IIT Madras, and UC Berkeley are overseeing the verification layers for cybersecurity and drug discovery.
“The world’s hardest problems are not unsolvable. They are unformalized,” says Rajagopalan. “Every domain where being wrong can cost someone their health, money, or freedom has rules.”
The final step for this technology is simply to codify those rules.
Key takeaways
- Pramaana Labs has raised $27 million in seed funding to apply formal verification methods to high-risk AI applications in law, healthcare, and finance.
- The system combines the natural language capabilities of a standard LLM with a deterministic verification layer built on the LEAN programming language.
- Domain experts, including former IRS commissioner Danny Werfel and professors from top Indian and American universities, are curating the formal rules for specific verticals.
- The core philosophy is that complex problems in regulated industries are not unsolvable, but rather lack the formalised rules required to make AI reliable.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.
