Nov 2, 2026 · 1h 7m
Legora builds legal agents without seeing most client data
Building Legal Agents When You Can't See the Data — Jacob Lauritzen, CTO of Legora
Legora’s approach shows how specialized AI companies can improve legal work while confidentiality limits access to the data needed for training and evaluation.
- 1Legora evolved from legal research chat into a platform for document review, agents, structured analysis, and collaboration.
- 2Model selection in legal work balances intelligence, cost, latency, governance, regional availability, and task-specific performance.
- 3Confidentiality pushes Legora toward anonymized feedback, synthetic data, customer traces, and product-level evaluation instead of raw documents.
Don't miss
Jacob Lauritzen explains how Legora can improve its systems through anonymized feedback, synthetic examples, and complete work traces without accessing confidential client documents.
The brief
Legora began as a ChatGPT-based legal research tool and expanded into document review, workflow orchestration, structured lists, and agentic work as models improved.
Jacob Lauritzen describes a model strategy built around routing: routine legal tasks need different tradeoffs in cost and latency than complex reasoning and drafting.
The company evaluates systems with engineers, lawyers, legal engineers, synthetic examples, and task-specific testing, while confidentiality prevents direct inspection of most client data.
Tabular Review shows the product’s distinctive shape: lawyers can ask structured questions across thousands of documents, inspect citations, and collaborate in a spreadsheet-like interface.
Legora’s forward-deployed engineers customize workflows and integrations, but the larger strategy is to build legal-specific primitives rather than a general-purpose frontier agent.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.