Skip to content
Agents · Nov 3, 2026

Legora’s agentic platform for legal work keeps client data private by design

Legora’s CTO Jacob Lauritzen describes how the company builds and tests legal agents without accessing sensitive client data, relying on simulated environments and a team of 100 legal engineers.

Trust79
HypeLow hype

1 source · cross-referenced

ShareXLinkedInEmail
TL;DR
  • Legora’s agentic platform operates without visibility into client data, which remains under law firm control.
  • The company tests agents in simulated ‘worlds’ modeled on customer workflows, achieving citation accuracy improvements from 70% to 99% in one release without code changes.
  • Legora uses a model picker, runs entirely on Azure for privacy, and employs a tabular review tool handling up to 10,000 contracts.
  • The platform supports drafting, due diligence, and multi-agent coordination, with a team of 100 legal engineers and forward-deployed engineers.
  • Legora’s approach contrasts with model labs’ restrictions on coding tools, emphasizing privacy-first agentic workflows in legal tech.

Legora’s agentic platform is designed to operate without accessing or retaining client data, a constraint that shapes how the company builds, tests, and deploys its agents. According to Jacob Lauritzen, CTO of Legora, the company cannot read a single trace of what its agent does for a lawyer because the data belongs to the law firm’s clients. This privacy-first approach requires Legora to develop and evaluate its agents in simulated environments modeled on customer workflows, akin to the ‘worlds’ used in reinforcement learning.

The company’s evaluation strategy relies on a team of 100 legal engineers and forward-deployed engineers who build test environments to mirror real-world legal tasks. Lauritzen described a specific case where a citation problem was resolved from 70% accuracy to 99% in a single release without requiring changes to the underlying system, highlighting the effectiveness of their evaluation-driven approach. Legora’s platform supports a range of legal workflows, including drafting and editing Word documents, reviewing up to 10,000 contracts in a tabular format, and running due diligence. The agent also coordinates work between lawyers, clients, and other agents, and was used internally by Legora for acquisition diligence.

Legora’s technical stack includes a model picker, which the company resisted adding for years before introducing it to optimize performance across different tasks. The platform runs entirely on Azure, a choice Lauritzen attributed to data privacy requirements. The company does not use ElevenLabs for transcription, and its agentic tools include Tabular Review—a tool for handling large-scale contract review—and a Word add-in for document drafting. Legora’s intake process includes a Slack-based bot, and its agents can create lists that trigger additional agentic workflows.

The company’s hiring focus includes roles for legal engineers and forward-deployed engineers in New York, Stockholm, and London. Lauritzen emphasized the importance of building products that align with customer workflows, noting that drafting long contracts is fundamentally a search problem. He also discussed the challenges of balancing model capabilities with privacy constraints, such as avoiding default embeddings and managing hallucinations through a citation system. Despite the company’s growth from a Stockholm basement to a team of 100 engineers, Lauritzen described feeling understaffed, reflecting the high demand for specialized legal engineering talent.

Sources
  1. 01Latent Space — swyx — Building Legal Agents When You Can't See the Data - Jacob Lauritzen, CTO of Legora
Also on Agents

Stories may contain errors. Dispatch is assembled with AI assistance and curated by human editors; despite the trust-score filter, mistakes happen. We correct publicly — every article links to its revision history. Nothing here is financial, legal, or medical advice. Verify before relying on any claim.

© 2026 Dispatch. No ads. No sponsorships. No paid placement. Reader-supported via Ko-fi.

Built by a person who cares about honest AI news.