The archive · AI & Models · Strategic decision · 2025–2026
Pramaana Labs bets provably correct AI beats plausible AI; $27M Khosla seed
Ex-Gemini and Glean engineers encode tax law into machine-checkable proofs so high-stakes AI refuses to answer unless it can prove itself.
Pramaana Labs
What the business is
An AI verification layer for high-stakes domains — tax, law, drug discovery and financial compliance — that runs LLM answers through formal proof engines.
Starting capital：$27M seed led by Khosla Ventures (June 2026), with Accel, BoldCap, Nexus Venture Partners, Premji Invest and Unbound.
How it started
Founded in 2025 in Palo Alto by three IIT Madras alumni: Ranjan Rajagopalan (ex-Google Maps), Krishnan Raghavan (built Glean's first assistant) and Sanjay Ganapathy (Gemini at Google DeepMind). Their shared conclusion was that in law, tax or medicine, an AI that is only 'probably right' is unusable — and that fighting hallucinations is a research problem, not a product problem.
What happened
On 2026-06-17 the company announced the $27M seed. Its tax formalisation effort is advised by former IRS Commissioner Danny Werfel and built with Yale law and Stanford researchers; professors from IIT Delhi, IIT Madras and UC Berkeley oversee cybersecurity and drug discovery, with sponsored research at Stanford's Centaur Lab. The funding trains formalisation and prover models and scales domain experts across regulated verticals.
How it ended up
Still building: the June 2026 seed funds model training and hiring; no commercial deployment announced as of September 2026.
Background
Pramaana Labs is building the verification layer for high-stakes AI: an LLM for natural-language questions wrapped in a deterministic, formal-verification layer that either returns a machine-checkable proof an answer is correct or states which rule breaks. The company says it refuses to answer before it proves. Its bet is that law, tax, drug discovery and financial compliance will not deploy AI until answers are accountable.
The company was founded in 2025 in Palo Alto by three IIT Madras alumni — Ranjan Rajagopalan, who led Google Maps Moderation; Krishnan Raghavan, who built Glean's first assistant and concluded that fighting hallucinations is a research problem; and Sanjay Ganapathy, a former Google DeepMind staff engineer who worked on Gemini. On 2026-06-17 it announced a $27M seed led by Khosla Ventures with Accel, BoldCap, Nexus Venture Partners, Premji Invest and Unbound.
Pramaana formalizes each domain with its own experts: former IRS Commissioner Danny Werfel advises the tax effort, professors from IIT Delhi, IIT Madras and UC Berkeley oversee cybersecurity and drug discovery, and research is sponsored at Stanford's Centaur Lab. Google DeepMind VP Pushmeet Kohli and Microsoft CoreAI VP Sriram Rajamani are early backers. As of September 2026 the company is pre-product, spending the round on training formalisation and prover models.
What has to be true
- Regulated domains already have written rules — tax code, clinical protocols — so they are the easiest to formalize and the natural first wedge.
- The team combines LLM engineering from Gemini and Glean with formal-methods credibility, plus an advisor like a former IRS commissioner.
- A deterministic proof layer turns an LLM's flexibility into accountability, the property enterprises in law, tax and health must have before deployment.
- Backing from Khosla and leading formal-verification researchers signals this is a research-grade problem worth solving, not a feature bolt-on.
What can be applied
When being wrong is expensive, 'plausible' doesn't sell — machine-checkable proof does; the cheapest wedge is a domain whose rules are already written down.
Aftermath
As of September 2026 Pramaana is pre-product, using the $27M seed to train formalisation and prover models and to scale domain experts across tax, diagnosis, cybersecurity and financial compliance. Its tax effort is advised by former IRS Commissioner Danny Werfel, sponsored research runs at Stanford's Centaur Lab, and professors from IIT Delhi, IIT Madras and UC Berkeley oversee other verticals. The company claims its system either returns a machine-checkable proof or says exactly which rule breaks, and that it refuses to answer before proving.
Sources
- Pramaana Labs raises $27M seed round from Khosla Ventures to bring formal verification to AI
- Pramaana Labs Raises $27M Led by Khosla Ventures to Build the AI Layer That Proves It Is Telling the Truth
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card