ILTACon: AI Search in Legal Tech & Client Data Security Concerns
The annual ILTACon conference, a pivotal gathering for legal technology professionals, recently concluded its first day, offering a deep dive into the transformative power of artificial intelligence, particularly in the realm of legal search. Insights from key sessions and emerging startups painted a clear picture of how AI is reshaping legal practice, from precedent analysis to client intake. The Draftwise team provided valuable on-the-ground coverage of these developments.
A central theme emerged from the session titled ‘Charting Your Search Journey in the Age of AI,’ featuring industry leaders such as Douglas Freeman of Simpson Thacher, Yannic Kilcher from DeepJudge, Ilona Logvinova of Cleary Gottlieb, and Oz Benamram, a prominent Chief Knowledge & Innovation Advisor. A critical distinction was immediately drawn: Large Language Models (LLMs) are not, inherently, search engines. While LLMs excel at processing and generating human-like text, their effectiveness in legal contexts hinges on being paired with robust search capabilities and high-quality, relevant input. LLMs are adept at handling well-known facts but struggle with dynamic information or specifics not widely documented, making them ill-suited for directly retrieving precise legal documents from a firm’s internal systems.
The speakers emphasized that effective legal search tools are essential for narrowing down vast datasets, much like turning a “needle in a haystack” into a “haystack full of needles,” as Ilona Logvinova aptly put it. These tools can identify the most relevant documents—perhaps a crucial ten—which an LLM can then analyze for deeper insights. The ability to supply an LLM with the correct context at the precise moment, according to Oz Benamram, is what truly unlocks its value. This symbiotic relationship between advanced search and analytical AI allows legal professionals to not just store past precedents but to dynamically interrogate them, uncovering new intelligence and patterns that were previously inaccessible. Benamram also posed a strategic question to firms: with clients likely to solve a significant portion of their legal questions using AI independently, is it not preferable for them to use a solution built on the firm’s own institutional knowledge rather than a general-purpose tool like ChatGPT?
The Q&A segment of the session heavily focused on data security and client resistance to AI adoption. Concerns ranged from the potential for LLMs to “hallucinate” or generate incorrect information without proper context and safeguards, to the critical need for robust security measures within the tools themselves to prevent sophisticated threats like prompt injection attacks. Speakers highlighted that security must be integrated into the system’s architecture, not merely layered on the LLM. Interestingly, discussions also suggested that as search capabilities improve, the traditional, rigid folder structures for data organization become less vital; instead, connecting diverse data sources and enriching document profiles takes precedence. While large, general-purpose LLMs are powerful, firms can also consider lighter, in-house models hosted within their secure perimeters, as context, not just model size, is paramount. Educating clients on how LLMs function—specifically, that the models do not “learn” from sensitive data sent to them—was identified as a key strategy to alleviate privacy concerns. The overarching message was a call for firms to balance the immense potential of AI with stringent security, governance, and data management practices.
Beyond the main sessions, ILTACon’s Startup Alley showcased innovative solutions poised to disrupt the legal sector. Among them was Lexara, founded by former lawyer Amy Swanner. Her company addresses a common pain point: the unbillable hours spent on initial client interviews. Lexara Engage, their AI-powered tool, guides potential clients through a structured interview, gathering essential information and conducting real-time, high-quality conflict checks. The system can even customize screening questions to specific practice areas or client preferences, ensuring only suitable leads progress. Lawyers then receive a concise case summary and a full transcript, streamlining the intake process.
Another notable startup was Crimson, a YC-backed company currently participating in the A&O Sherman Fuse batch. Specializing exclusively in litigation and arbitration for the UK and US markets, Crimson was founded by former litigators from firms like Clifford Chance and Wilmer Hale. They built the AI platform they wished they had during their years in practice. Crimson serves a diverse clientele, from litigation boutiques to large regional firms, and sought to expand its footprint in the US market at ILTACon, gathering insights on the specific AI needs of American litigators.
ILTACon Day One underscored a critical juncture in legal tech: the integration of AI is no longer a distant prospect but a present reality, demanding careful navigation of its power, practical application, and inherent security challenges.