DeepJudge Raises $10.7M Seed Round, Led by Coatue, and Announces Knowledge Assistant

Paulina Grnarova


Today I’m excited to share that DeepJudge has raised $10.7 million in an oversubscribed seed funding round led by Coatue, with participation from notable angel investors such as Gokul Rajaram, Michele Catasta and others.

“DeepJudge is one of the first products on a mission to empower legal teams to harness their internal data to make strategic decisions. From a technology standpoint, we are impressed by DeepJudge’s approach to solving data retrieval, which we believe is crucial to achieve high-quality generative results. We are excited to support DeepJudge as they seek to gain traction in the legal industry and grow their client base.”

- Caryn Marooney, General Partner at Coatue

Since founding the company in 2021 with my co-founders, Yannic Kilcher and Kevin Roth, we have been dedicated to addressing a fundamental challenge in the legal sector: the ability to efficiently find and use the troves of internal data that lawyers and other legal professionals generate daily. 

Today, DeepJudge Knowledge Search allows firms to instantly locate relevant content from across the enterprise, including document management systems, the Microsoft 365 ecosystem, and other tools. 

We are also proud to announce the availability of DeepJudge Knowledge Assistant, the only generative AI interface for law firms and legal departments powered by the institutional knowledge contained within all of a firm's documents.  DeepJudge Knowledge Assistant uses Retrieval-Augmented Generation (RAG) technology to provide the most relevant, up-to-date information, ensuring high-quality answers are grounded in each organization’s comprehensive internal knowledge base.

Bringing intuitive, full-scale search inside the law firm

Over time, lawyers generate millions of documents, forming a vast reservoir of intellectual output. Yet, when tackling new tasks, lawyers still too often have to start from scratch. They struggle to locate colleagues with relevant expertise, or spend hours crafting complex Boolean queries.

We live in a world where finding the right information is easy. In our personal lives, we can effortlessly find whatever we need. Want to check the latest score? Need directions to a friend's house? Looking for reviews of the newest restaurant in town? Just search for it!

Meanwhile, law firms have not experienced this breakthrough – at least not for their internal documents, despite this internal knowledge being a critical tool of their trade.

This is where DeepJudge steps in. Our mission is to connect legal professionals with their entire collective knowledge in an intuitive manner, turning it into a powerful competitive advantage.

"For some time we’ve been looking for an advanced search engine that can effectively navigate through millions of internal documents in various formats, so that our collective know-how and expertise are at our lawyers’ fingertips. Finding relevant knowledge at the right time is critical for providing tailored legal advice in high-stakes strategic matters and DeepJudge, customised to our specialised needs, now enables us to do precisely that. Thanks to its intuitive interface and capacity to handle precise and nuanced queries, our lawyers can uncover key insights with ease."
"The output from DeepJudge's search function can then seamlessly connect with a generative AI-based workflow to help our people to efficiently deliver high-quality, consistent value for our clients. DeepJudge has both impressive technical credentials and a vision for how fast, focused information retrieval can take law firms to the next level."

- Fedor Poskriakov, Deputy Managing Partner at Lenz & Staehelin

Unlocking new possibilities for the practice—and business—of law

Without a good search, many law firms and legal departments rely on a curated set of documents as the go-to storehouse for their institutional knowledge. These manifest as brief banks, agreement templates, clause libraries and more.

Curated archives absolutely are useful. They enable discovery of vetted and trusted precedent work that can be reused. They also help to identify internal experts, facilitating connections between colleagues. 

However, many challenges cannot be solved through curation alone. Most lawyers have stories about struggling to find variations on language tailored to each unique client-matter situation. This is where a comprehensive search becomes incredibly useful. With just a few keystrokes, you can not only identify who is an expert but also find someone who faced a similar specific scenario and see how they addressed it.

AI-powered search also unlocks entirely new possibilities for law firms, such as spotting trends and even business development opportunities from across their entire client base. 

For example, imagine the possibilities if a quick search of your documents could tell you which clients have a certain clause in their contracts that has just been made problematic due to a recent regulatory change. Imagine the ability to delight a client by proactively offering to remediate that issue. And for law departments, imagine if a search could similarly spot those clauses so you could handle the remediation yourself.

It is in this way that DeepJudge can make legal teams instantly smarter and more proactive. We imagine a future in which AI search could even alert lawyers in real time to emerging issues that are arising from across the practice. In short, we are only getting started.

Solving Search for Legal in the Age of Generative AI: Focusing on the “R” in RAG

Effective search for internal documents is particularly important to legal professionals as they embark on implementing Generative AI systems. The initial wave of legal AI products highlighted the surprising ability of generative AI to draft contracts, write memos, or summarize cases. While these are neat time-saving tools, the quality of the output relies entirely on the input data.

In short, firms are beginning to discover that Retrieval-Augmented Generation (RAG) systems are limited by the capabilities of the underlying retrieval (i.e., search) systems. With Generative AI, there will always be "hallucinations." This has been mathematically proven [1]. Therefore, it’s critical to provide your generative AI tool with the most comprehensive grounding in your firm's own data, which requires an excellent retrieval system. 

Another crucial point is the human-in-the-loop connection: machines excel at processing large volumes of information, while humans excel at reasoning and decision-making based on that information. Combining the strengths of both allows for a more effective approach, enabling users to quickly and easily verify the answers at each step rather than relying on a fully end-to-end automated process.

Implementing AI in the legal sector requires scalable and secure connections between LLMs and firms’ data, reflecting the comprehensive knowledge within a firm while adhering to various access rights and ethical walls. 

Our platform delivers trusted answers grounded in your organization’s knowledge base, scales responsibly to meet enterprise needs, and offers turnkey implementation through ready-to-integrate connectors.

Effective search for internal documents is notoriously difficult to perfect. Not only do natural language queries include any amount of legal jargon, there is a potentially massive scale of 100s of millions of documents, each with stringent and varying security requirements. Not to mention filtering through duplicates, multiple versions, redlines, emails and more.

DeepJudge is uniquely positioned to meet these challenges. Our co-founders are a team of ex-Googlers and PhDs in AI. We’ve paired up with legaltech experts including Tony Ensinger, previously Head of Sales at Casetext (acquired by Thomson Reuters), as our SVP of Sales and Product Strategy, Steve Obenski, former senior leader at Kira Systems (acquired by Litera), as an interim Chief Strategy Officer, and Jan Puzicha, one of the founders of Recommind (acquired by OpenText), as an advisor.

Unparalleled Usage and Adoption

We are proud not just of our technical chops, but also the ease of use and simplicity of the product. There’s no better testament to that than the words of one of our first customers:

“Knowledge Search allows me to write intuitive queries and to be as specific as I want to be; something that was not possible before with the keyword-based enterprise search engines. Knowledge Search is a fundamental part of our work and our lawyers were quick to adapt as it’s as intuitive as using Google. The adoption rate has been remarkable, with more than 80% of Homburger’s legal professionals incorporating it into their workflow and a level of engagement that is unparalleled compared with other legal tech tools at the firm.” 

- David Oser, partner in M&A at Homburger, one of the largest Swiss law firms

Interested in joining us on our journey? Request a demo or check out our open positions on our careers page.

1. Xu, Ziwei, Sanjay Jain, and Mohan Kankanhalli. "Hallucination is inevitable: An innate limitation of large language models." arXiv preprint arXiv:2401.11817 (2024). Copy available at