Last month it was creative professionals. This week, Anthropic turned its attention to lawyers.
Anthropic launched Claude For Legal on May 12, adding six practice-area plugins covering commercial law, employment, privacy, product liability, corporate, and AI governance. The release also ships MCP connectors - integrations that let Claude pull data from tools lawyers already have open - to nine platforms: DocuSign, Ironclad, iManage, NetDocuments, LexisNexis, Thomson Reuters, Box, Everlaw, and LSuite.
MCP (Model Context Protocol) is Anthropic's open standard that lets AI models connect to external data sources and tools directly, without manual copy-pasting. For a lawyer, that means Claude can pull a contract from iManage, cross-reference it against LexisNexis case law, and help draft a redline - all without leaving the chat window.
This follows the same structure as Claude for Creative Work, which launched in April with plugins tailored to writers, designers, and video producers. Anthropic is building a clear pattern: pick a vertical, ship domain-specific plugins, add connectors to that category's dominant software platforms, and repeat.
What the Nine Connectors Cover
The integration list reads like a who's who of legal software:
- Contract management: DocuSign, Ironclad
- Document management: iManage, NetDocuments, Box
- Legal research: LexisNexis, Thomson Reuters
- Litigation: Everlaw
- Practice management: LSuite
For any firm already paying for LexisNexis or Thomson Reuters, the MCP connector is the most immediately useful piece - Claude can search case law and statutes directly rather than requiring you to paste excerpts in manually.
The practice-area plugins are harder to evaluate without seeing the actual retrieval logic behind them. The "AI governance law" plugin category is worth watching given how fast that regulatory landscape is moving in 2026.
A Natural Fit That Generic AI Has Struggled to Crack
Law firms have been slower than most industries to adopt general-purpose AI tools, and for understandable reasons: client confidentiality concerns, the risk of hallucinated citations, and the fact that a model trained on the general web doesn't carry the specialized legal knowledge that billable work actually requires. A model with LexisNexis integration and employment law context built in is a different proposition than a blank chat window.
The work is also structurally well-suited to AI assistance - document-heavy, time-billed, and full of repetitive drafting tasks where errors carry real liability. If the practice-area plugins hold up to scrutiny, this vertical push is more credible than most legal AI announcements of the past two years.