Our Investment in DeepJudge: AI-Powered Search Built for the Complexity of Law Firms

Law firms are not like other enterprise businesses. The context that drives decisions at most corporations is stored and shared in relatively straightforward ways: in their messaging app, an email client, a couple of different document-sharing services, and maybe a productivity tool. Law firms, by contrast, are a labyrinth of complexity. Their knowledge, precedent libraries, permission models, and workflows are uniquely intricate, regulated, and often date back decades. 

It’s this difference that leads to misunderstandings about the problems legal tech must solve. Making legal AI actually useful isn’t a problem of intelligence that can be solved with better models or more GPUs; it’s a problem of context. ​​

Rather than selling a generic GPT model with a serif font and calling it “legal AI,” DeepJudge approaches the problem head-on, building a powerful enterprise search engine purpose-built for law firms. What Glean did for enterprise search, DeepJudge is doing for law firms. 

DeepJudge indexes the full scale of a firm's internal document management, email, intranets, and client portals to provide instant pinpoint access to institutional knowledge, without moving sensitive data out of its secure boundaries. Unlike generic AI tools, DeepJudge uses custom neural networks and real-time data connectivity to deliver precise, context-aware search across all internal systems, without requiring data uploads or disrupting existing workflows. Its proprietary retrieval engine overcomes the messy, fragmented, permissioned structure of legal data, turning millions of precedents into actionable insight in real time.​​

DeepJudge is led by Paulina Grnarova, Kevin Roth, and Yannic Kilcher, a team of former Google researchers and AI PhDs. They’re deeply focused, meticulous founders who, for the most part, didn't want to leave their work and team to court investors.

So we came to them. I skipped a Felicis offsite (sorry, team), hopped on a plane to Zurich, and told the DeepJudge team  I “happened to be in town” and wanted to chat through a few GTM ideas if they were open to it. That visit laid the foundation for trust between us: that we would respect their commitment to the work, and that when we show up for them, we do so by providing value.

Felicis is proud to lead DeepJudge’s Series A. DeepJudge has seen explosive year-over-year revenue growth of over 500% and has attracted elite global law firms as customers, including Freshfields, Holland Knight, Cozen O’Connor, ArentFox Schiff, and Schoenherr. 

For law firms to see real gains from legal AI, the technology has to deliver more than just siloed chatbots. It must deeply wire AI into the legal knowledge network, meeting the highest bar for confidentiality while delivering speed and competitive advantage. DeepJudge’s technical rigor and product traction prove it’s the platform law firms want to use to harness their collective knowledge.​​ Search - not models - is the optimal wedge to unlock downstream AI capabilities, enabling intelligent workflows, automation, and insights built on a foundation of connected knowledge.

This is just the beginning. Felicis is thrilled to support DeepJudge as they lead a new era for legal professionals, unlocking the true value of AI in law.