DeepJudge

Market position

DeepJudge positions itself in the legal knowledge management and enterprise search category, emphasizing retrieval quality as the foundation for downstream generative AI applications rather than competing primarily on document drafting. Forbes notes strong competition in US legal AI, citing Legora (reported $150M raise at a $1.8B valuation) and Harvey (reported $150M round at an $8B valuation); DeepJudge also partners with Harvey and Thomson Reuters on its agent handoff protocol.

Stated differentiators include founding by former Google search and AI researchers, an index that respects existing permissions and ethical walls while leaving data in source systems, deployment flexibility including on-premises, model-agnostic workflow orchestration, and coverage of a firm's full archive rather than curated document collections such as brief banks and clause libraries.

Technology

The platform builds a per-client index analogous to a web search index, mapping words and phrases to documents, tagging data and connecting related items such as similar deal types or clients. It automatically classifies documents into an extensive taxonomy, detects near-duplicates and redlines, and performs holistic, multi-level search combining client, matter, document and people data across sources at a scale described as billions of data points. Retrieval-augmented generation is used to ground generative outputs and reduce hallucination risk, and the LLM-agnostic agentic reasoning layer executes multi-step tasks across sources. MCP is used to give AI systems governed access to firm knowledge without custom integration code.

Go-to-market

The company sells direct to law firms and in-house legal departments, with a website demo-booking motion and published customer testimonials. Following the Series A it stated plans to expand the engineering team, hire a larger sales team and increase market adoption in the US and UK. Leadership includes a CRO responsible for global revenue and go-to-market and an SVP of Sales previously at Casetext.

Law firms across size ranges (20-99, 100-499 and 500+ lawyers) and corporate legal departments; user types cited include partners, associates, lawyers, general counsel, knowledge management staff, firm leadership, paralegals and law clerks, professional development and business development teams.

Geography

Headquartered in Zurich, Switzerland, with teams in the United States and Canada and clients served globally; the Series A proceeds are directed at scaling in the US and UK. Product languages: English, French, German and Italian.

History

DeepJudge originated as an ETH Zurich spin-off, with coverage of its document-processing work by ETH Zurich in October 2022 and inclusion in Legaltech Hub's "New AI Start-Ups to Watch in Legaltech" in October 2023. In June 2024 it announced a $10.7 million seed round led by Coatue alongside the Knowledge Assistant product. In November 2025 it announced a $41.2 million Series A led by Felicis with participation from Coatue at a $300 million valuation. Subsequent announcements include Agent Skills (February 2026), a partnership with Harvey around institutional knowledge (May 2026), and an Agent Handoff Protocol with Harvey and Thomson Reuters among first adopters (August 2026).

Risks & controversies

Forbes describes an intensely competitive US legal AI market with large, better-capitalized rivals raising at multi-billion-dollar valuations. The underlying problem the product addresses — AI hallucinations in legal work, with more than 120 court-filing hallucination cases identified since 2023 per Baker Donelson — is also a category risk. Many performance figures on the company website are vendor-reported and not independently verified.