The Neuron: AI Explained How Thomson Reuters Built AI Agents That Think Like Lawyers
34 snips
Sep 3, 2025 Joel Hron, CTO at Thomson Reuters, shares groundbreaking insights into AI's role in legal research. He introduces Deep Research, a tool that strategizes like a seasoned lawyer and handles high-stakes legal issues. The discussion covers challenges like hallucinations in legal AI and the importance of user interaction for refining these systems. Hron emphasizes the need for human oversight and adaptability in workflows, showcasing how AI can drastically improve efficiency in the legal field.
AI Snips
Chapters
Transcript
Episode notes
Domain-Tuned Agents Beat Generic Search
- Thomson Reuters pairs agent reasoning tuned for law with proprietary tools and content to improve legal research quality.
- The agent follows legal breadcrumb trails like KeyCite to surface comprehensive, jurisdiction-aware results.
Use Experts To Define Correctness
- Keep experts in the loop to score outputs and create rubrics for correctness, not just token matches to sources.
- Evaluate answers for legal reasoning and diverse valid paths, not only specific cited cases.
Tools Amplify Expert Judgment
- AI will augment skilled lawyers rather than replace them, amplifying judgment-heavy tasks while automating boilerplate.
- The best attorneys who learn these tools will outperform peers significantly.