Legal AI
Legora
Legora builds AI software for law firms and in-house legal teams to research, review, draft and collaborate on legal matters. Legora aOS connects language models with firm documents, legal sources and workflow tools. Lawyers can conduct jurisdiction-aware research with structured citations, review documents in tables and search internal knowledge bases. Its product family includes Word and Outlook add-ins and Portal for client collaboration and delivery.
Co-founders August Erséus and Sigge Labor began experimenting with legal AI in 2020; co-founder and CEO Max Junestrand joined them in 2023. The early product was developed alongside lawyers at Mannheimer Swartling in Stockholm, and the business changed its name from Leya to Legora in 2025. Its engineering approach centers on applying multiple frontier models to legal workflows. The platform coordinates specialist agents, incorporates firm playbooks and matter history, and provides separation between client matters and audit trails for agent actions.
By August 2026, the company reported more than 100,000 legal professionals using Legora across more than 1,500 law firms and in-house legal teams in over 50 markets. It reported exceeding $100 million in annual recurring revenue in April 2026. That month, a $50 million extension brought its Series D to $600 million in equity at a $5.6 billion post-money valuation.
Explore the recordings
Legora’s supplied archive contains one recording: a joint technical talk with turbopuffer. The paths below navigate different parts of that recording. Architecture, performance figures and corpus scale reflect the speakers’ recorded account; they do not establish Legora’s current infrastructure or product status.
Start with the two search problems
Begin with Connect AI to Billions of Legal Documents to understand the distinction between project-scoped document search and legal research across laws, cases and regulations. The legal-research discussion explores jurisdiction hierarchies, temporal validity, related regulations and query fan-out. The corpus approaching ten billion vectors is a scale described in the recording.
Follow the migration and its failure mode
Use the migration account for a concrete infrastructure case study: shared Elasticsearch, regional deployments, Postgres and then turbopuffer. Jacob Lauritzen describes how roughly 4,000 project-hash partitions mixed active and inactive data, causing cache thrashing. He reports search and ingestion P99 increasing from about 100 milliseconds to 20 seconds, followed by improvements after adopting one turbopuffer namespace per project. Treat those outcomes as reported results from this migration.
Explore storage boundaries and search internals
Follow the storage and indexing discussion for object-storage writes, background indexing and reads across memory, SSD and object storage. The speakers connect namespace boundaries, buckets and encryption keys to isolation requirements, and describe selected Legora workloads operating with the SSD cache disabled. Continue into Simon Eskildsen’s explanation of clustered tree indexes, posting lists, BM25, compression and memory bandwidth for the underlying search mechanics.
2 talks
Newest first
1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.
Company sources · checked 2026-08-28
- Newsroom | Legora
- Legora CEO Max Junestrand on Legal AI and Building Globally | Perspectives
- Leya is now Legora
- Legora aOS
- Legora extends Series D with additional $50 million, welcomes Atlassian and NVentures as investors
- Legora is acquiring Wexler to bring fact intelligence to agentic legal work
- Legora: The AI workspace for lawyers | Y Combinator