

UBS AI Podcast - CEO Series - Ep. 5 (Winston Weinberg, CEO of Harvey)
Sep 18, 2025
Winston Weinberg, CEO and co-founder of Harvey AI, brings his background as a lawyer to discuss the transformative role of AI in the legal industry. He shares insights on founding Harvey after early experiments with GPT, highlighting the importance of context and accuracy in legal tasks. Winston explores challenges in product development and how Harvey navigates competition while improving workflows. He offers personal reflections on leadership and the evolving legal landscape, hinting at AI’s future impact on associate roles and the industry at large.
AI Snips
Chapters
Transcript
Episode notes
Origin Story: Landlord-Tenant Experiment
- Winston Weinberg tested early GPT models on landlord-tenant pro bono work and built chain-of-thought prompts.
- He validated outputs by asking three lawyers whether they'd send the AI response unedited and got strong approval.
Two Core Problems For Vertical AI
- Two core challenges for vertical AI are capturing all necessary context and ensuring correct, process-specific outputs.
- Legal work relies on precedent and internal data, making verticalization essential.
Build Agent Chains For Complex Tasks
- Decompose complex legal tasks into chained agents that handle general subtasks like research, drafting, and summarization.
- Combine those agents to complete higher-level tasks such as drafting a motion or managing an M&A process.