At the Legal AI and Innovation Summit hosted by Inside Practice on November 21, 2024, three leaders in the intersection of law and technology took the stage to discuss “The Future (and current) State of Search and AI’s Impacts on Knowledge Work More Broadly”. This 4-part blog series recaps their insights and predictions for how search will continue to shape the legal profession.
The panel featured Oz Benamram, Chief Knowledge and Innovation Officer at Simpson Thacher & Bartlett LLP; Yannic Kilcher, CTO and Cofounder of DeepJudge; and moderator Ilona Logvinova, Director of Practice Innovation at Cleary Gottlieb Steen & Hamilton LLP. Together, they explored the rise of contextual and interactive search, how AI-powered tools are redefining document relevance and retrieval, and how these advancements are reshaping the practice of law.
The middle part of the panel discussed how AI-powered search tools, such as DeepJudge, are redefining document relevance and retrieval.
Oz Benamram praised DeepJudge for its focus on contextual search and its ability to abstract from the documents up to the matter level, noting that it aligns with his long-held belief that documents are merely “evidence” supporting larger matters. “We sell matters, not documents. By shifting the focus to matters, we get a clearer picture of what’s important to clients and how we can better serve them,” Benamram said.
Yannic Kilcher elaborated on how AI enables this shift. Capable of understanding and extracting nuanced legal concepts, modern techniques can automate or even glean on-the-fly a lot of useful information about documents that previously required human effort to create. Kilcher emphasized the importance of extracting insights from unstructured data often dismissed as “garbage.” “The so-called garbage—emails, drafts, all of it—contains valuable signals. AI can help us turn that into diamonds.”
Kilcher also pointed out that clients turn to top law firms not for routine tasks, but for complex, nuanced issues. The key to resolving these challenges might lie in a precedent buried in a partner’s email or a document from years ago—exactly the kind of hidden insight that can only be surfaced by searching comprehensively across all documents, including unstructured and uncategorized data.
Once you have good search, you can also use AI to complete various workflows. Explaining the concept of retrieval-augmented generation, Kilcher explained, “with RAG, we use search to retrieve the right documents and then let AI summarize, compare, or analyze them. It’s about enhancing the lawyer’s ability to interact directly with data.”
Kilcher explained that profiling, categorization and metadata tagging, while often useful, alone cannot keep up with the complexity of legal data. “The world is too messy to fully organize. Every attempt at rigid categorization or tagging the entire universe—whether it’s the Internet or all your internal data—falls short because you’ll always be two years behind what you actually need.” He noted that a combination of both is likely the best approach. Dynamic search systems like DeepJudge mitigate the need for exhaustive upfront tagging, enabling users to find relevant data in real time, but also benefit from incorporating metadata and tags.
Benamram agreed, using the contrasting archetypes of “filers” (those people who meticulously organize) and “pilers” (those who save everything in one place), pointing out that the latter group is growing as search technology improves.
“We design systems for both filers and pilers. But with stronger search tools, more people are becoming pilers” because they trust the system to find what they need, Benamram explained. He also stressed the importance of simplifying processes, focusing on contextual insights rather than overwhelming users with metadata requirements.
Watch the second part of the panel here:
Video will play from the 14 minute mark
Other posts in the series "The Critical Role of Search in Legal AI":
Part One: The History and Evolution of Search in Law Firms
Stay tuned for Part 3, coming next week!
