Pramaana Labs says enterprises can’t afford AI that only “works in demos.” The startup has raised a $27 million seed round led by Khosla Ventures to add formal verification on top of conventional LLMs, targeting high-stakes uses in law, drug discovery, and tax
For businesses trying to move from AI pilots to day-to-day operations, the problem often isn’t capability. It’s trust.
Pramaana Labs is betting that the missing piece is reliability — built with mathematical certainty rather than guesswork. On Wednesday. the startup announced a $27 million seed funding round led by Khosla Ventures. with participation from Accel. Boldcap. Nexus Venture Partners. Premji Invest. and Unbound.
The company plans to focus on highly sensitive verticals like law, drug discovery, and tax preparation. In those areas, an error isn’t just embarrassing; it can be costly in ways that linger for years. Pramaana’s approach is aimed at giving AI systems stronger protections against hallucinations and mistakes.
At the center of the pitch is how formalization fits these domains. Pramaana co-founder and CEO Ranjan Rajagopalan described rules — like those in the tax code — as something you can codify, turning reasoning into something closer to determinism.
“It’s like math in the sense that you have a lot of rules that you need to abide by,” Rajagopalan told TechCrunch, describing the rules of the tax code. “Once you have a codified version of it, the reasoning on top of it starts becoming deterministic.”
Pramaana’s system still uses a conventional LLM. keeping the flexibility to answer natural language questions and handle complex problem-solving tasks that normal computers struggle with. But it doesn’t rely on the model’s output alone. A deterministic layer sits on top to ensure the LLM’s work checks out.
That layer is where Pramaana says it differs from typical verification setups. Instead of treating correctness as an afterthought. it uses formal verification tools. drawing on LEAN — an open-source programming language used to verify mathematical proofs. The idea isn’t starting from scratch, either. Rajagopalan points to France’s CATALA project. which formalizes much of the country’s tax and benefit system into executable code.
For each use case, Pramaana will build its own LEAN-style formal verification system, overseen by domain experts. In tax law, the company is working with former IRS commissioner Danny Werfel. Professors from IIT Delhi, IIT Madras, and UC Berkeley are overseeing the cybersecurity and drug discovery system.
Rajagopalan frames the problem more bluntly: “The world’s hardest problems are not unsolvable. They are unformalized,” he said. “Every domain where being wrong can cost someone their health, money, or freedom has rules.”
Pramaana’s bet is that those rules can be written down — and once they are, AI deployments in high-stakes settings won’t have to operate on faith.
Pramaana Labs Khosla Ventures seed round formal verification LEAN artificial intelligence LLM hallucinations cybersecurity drug discovery tax law Danny Werfel IIT Delhi IIT Madras UC Berkeley Accel Boldcap Nexus Venture Partners Premji Invest Unbound