The world's fifth-largest law firm just became an AI company's design partner
The world's fifth-largest law firm (per Legora) is deploying Legora's agentic OS across three regions and co-designing its research and monitoring tools. iManage wired its document store into Legora via MCP. The vertical AI playbook is being written in law first.
- Hogan Lovells Cadwalader — formed in 2026 by the Hogan Lovells and Cadwalader merger, with 3,300+ lawyers in 36+ offices and described as the world's fifth-largest firm by revenue — is rolling out Legora's agentic OS across the Americas, EMEA, and APAC. Self-reported m:…0405
- The rollout follows a pilot of 1,300+ lawyers across 18 practice teams, and the firm will serve as a design partner for Legora's Agentic Legal Research roadmap and its Monitors regulatory horizon-scanning tool. Self-reported m:…0405
- Legora is connecting to the iManage MCP Server so lawyers can work with governed iManage content inside Legora's agentic OS; iManage remains the single source of truth and its permissions govern what each user retrieves. Self-reported m:…7281
- In-house teams are adapting: Brambles built an AI agent to centralize policy guidance for its legal team, junior lawyers are being trained to validate AI outputs, and The Star used AI to prepare anti-discrimination evidence to manage outside counsel costs. Self-reported m:…9406
The merger that bought a roadmap seat
Hogan Lovells Cadwalader, formed in 2026 through the merger of Hogan Lovells and Cadwalader, Wickersham & Taft, is the world's fifth-largest law firm by revenue, with more than 3,300 lawyers across over 36 offices. It has now agreed to deploy Legora's agentic operating system to its lawyers across the Americas, EMEA, and APAC — following a pilot that involved more than 1,300 lawyers across 18 practice teams. The scale matters, but the structure matters more: the firm is not buying seats, it is buying a seat at the roadmap. m:…0405
Hogan Lovells Cadwalader will serve as a design partner for two Legora products: the Agentic Legal Research roadmap and Monitors, a regulatory horizon-scanning tool. This is the vertical-AI bargain made explicit — the vendor gets its hardest workflows stress-tested by 3,300 lawyers; the firm gets the product shaped around how it actually works. Legora, which says it is used by more than 140,000 legal professionals at over 2,000 firms and in-house teams across 80 markets, just turned its largest deployment into its product lab. m:…0405
The document store becomes an API
In parallel, Legora deepened its integration with iManage by connecting to the iManage MCP Server. The connection lets legal professionals move documents, update document information, and surface institutional knowledge — including iManage Insight+ content — directly inside Legora's agentic OS, while iManage remains the single source of truth. The governance detail is the point: iManage permissions determine what each user can retrieve through the connection, and Legora's existing API integration already supports browsing, searching, accessing, and saving documents back. iManage says it is relied on by more than one million professionals at 4,500 organizations. The document management system, the most boring software in the law firm, just became an agent endpoint — with the permission model intact. m:…7281
The in-house playbook
Harvey's reporting on general counsels shows the buy side reorganizing around the same shift. Brambles Limited, with 24 lawyers across 60 countries, developed an AI agent to centralize policy guidance for its legal team. The pattern GCs describe is not task automation but process redesign: AI accelerates first drafts and preparatory work, junior lawyers are explicitly trained to evaluate AI outputs for missing facts, weak inferences, or incorrect conclusions, and supervising lawyers develop colleagues' judgment by questioning their AI-assisted drafts. The Star's legal team validated the business problem behind AI-generated requests before assessing the answers, and used AI to prepare evidence for an anti-discrimination matter to manage outside counsel costs. The profession is learning to supervise the machine the way it supervises junior associates. m:…9406
The durability argument is sharpest in Harvey's framework for regulatory consent orders: Harvey structures dense orders paragraph by paragraph, mapping commitments to deadlines and evidence, because consent orders can outlast the team that negotiated them — the FTC typically ends administrative orders in consumer protection and competition matters after 20 years. A Vault holds the order, supporting policies, regulator correspondence, and prior reports in one place. Multi-decade obligations need a memory that survives turnover; the AI system is being sold as institutional memory, not just speed. m:…0574
Why law is the template
Put together, the week's legal AI news describes a complete vertical playbook: a flagship customer that co-designs the product, an incumbent system of record that exposes itself as a governed agent endpoint, and a buy side that is reorganizing its training and supervision around machine output. The decision for anyone building vertical AI is visible in the structure: distribution comes from the system of record (iManage), product-market fit comes from the design partner (the merged mega-firm), and durability comes from owning the institutional memory (the Vault). Horizontal tools sell features. Vertical winners sell the firm's future org chart — and this week, the org chart signed back. m:…0405 m:…7281
What is still unresolved
- Whether design-partner arrangements give firms pricing or data-ownership advantages that later complicate the vendor relationship.
- How bar-association rules on AI supervision evolve alongside agentic legal research tools.
- Whether the iManage MCP model extends to other systems of record in law (billing, docketing) or stays document-centric.
Provenance by Synorb. Every claim links to its source. This story was written by an agent from the Manifests listed below. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, with provenance by Synorb” and link to this story. Keep the cited Manifest ids. Reuse terms v1.