According to Forbes, former Google AI researchers Paulina Granarova, Kevin Roth, and Yannic Kilcher have raised a $41 million Series A for their legal tech startup DeepJudge, valuing the company at $300 million. The funding was led by Felicis with participation from existing investor Coatue, bringing their total raise to significant capital since their 2020 founding. The Zurich-based company helps law firms search through their own confidential archives using AI, with clients including Gunderson Dettmer, Freshfields, and Holland & Knight. DeepJudge’s system specifically addresses the critical problem of AI hallucinations in legal work, where more than 120 cases of fabricated information in court filings have been identified since 2023. The company is now expanding its engineering and sales teams to push further into the competitive U.S. legal AI market.
So how does this actually work?
Here’s the thing about current AI tools like ChatGPT – they’re amazing at public information but completely useless for the confidential work product that makes law firms valuable. DeepJudge builds what’s essentially a Google Search index, but for a specific law firm’s entire document history. We’re talking decades of contracts, memos, case files – everything. This indexing process takes up to three weeks per client and creates a detailed map of where every term, clause, and concept appears across their entire archive.
The clever part? They don’t replace the AI models law firms are already using. Instead, they plug their custom index into whatever system the firm prefers – ChatGPT, Azure AI, open source models, whatever. So when a lawyer asks “Who in our London office has handled deals like this before?” the AI can reference the DeepJudge index rather than just making something up. It’s basically giving AI models the context they desperately need to stop hallucinating in high-stakes legal situations.
Why lawyers are absolutely terrified of hallucinations
Look, AI making up case citations or contract terms isn’t just inconvenient – it’s career-ending for lawyers. We’re talking about more than 120 documented cases of AI hallucinations in court filings since 2023 alone. That’s insane when you consider the consequences. As Coatue partner Caryn Marooney put it, “legal of all things really can’t afford the hallucinations.”
What makes DeepJudge’s approach interesting is they’re attacking the hallucination problem at the source. By building these rich contextual indexes, they’re giving AI models way more reference material to work with. It’s the difference between asking someone to recall a complex legal precedent from memory versus letting them look it up in a perfectly organized filing system. The latter is just way less likely to invent things.
Now about that crazy competitive landscape
This funding comes at a wild time for legal AI. Just last week, Harvey closed a $150 million round at an $8 billion valuation. Legora raised $150 million at $1.8 billion. We’re seeing investor money pouring into this space like there’s no tomorrow. And honestly, it makes sense – every law firm on earth is desperate for AI solutions that won’t get them disbarred.
But here’s what makes DeepJudge different: they’re not trying to build another AI model. They’re building the connective tissue between existing AI and the proprietary data that actually matters to law firms. It’s a fundamentally different approach than trying to train yet another legal-specific LLM from scratch. Whether that’s enough of a moat in this gold rush remains to be seen, but at $300 million valuation with serious clients already onboard, they’ve certainly got a fighting chance.
The ex-Google pedigree doesn’t hurt either. These founders literally helped build the search technology that now needs fixing for specialized use cases. There’s some beautiful irony in that – the people who helped create modern search are now building the next evolution specifically for the profession that needs accuracy most desperately.
