AI Agents Podcast AI Agents in Legal Tech - David Wong Thomson Reuters on Responsible AI | EP 127
Mar 25, 2026
David Wong, Chief Product Officer at Thomson Reuters, drives product and AI strategy for legal, tax, audit, and accounting software. He talks about how AI has evolved for legal research and document work. They explore AI automating tax prep and compliance. He outlines building professional AI agents using tools, authoritative data, and practitioner training.
AI Snips
Chapters
Transcript
Episode notes
Thomson Reuters Is Primarily A Software And Research Company
- Thomson Reuters earns most revenue from software and research used by lawyers, accountants, and risk professionals rather than news.
- David Wong explains over 90% of revenue is from B2B software that underpins justice, tax, and corporate systems, not the well-known Reuters news brand.
Early GPT Tests Failed Then Improved Rapidly
- David Wong's team tested GPT-3 on legal research in August 2020 and graded it with internal evals.
- GPT-3 performed very poorly, but the same evals later showed GPT-3.5 and GPT-4 improved from failing to B+ performance, prompting a $150M AI investment.
AI Targets Research And Drafting In Legal Workflows
- Thomson Reuters focuses AI on two core legal workflows: nuanced information retrieval and producing written work products.
- David Wong says these map to research (case law, statutes) and drafting briefs, contracts, and returns where AI can remove hours of grunt work.