Claude for Legal adds artifacts. NetDocuments provides a governed foundation.

Anthropic’s new legal gallery turns everyday legal work into shareable tools. What those tools work from matters more than ever.

Claude for Legal’s latest release is out today, and NetDocuments is available as a connector. That means legal teams can point Claude at their work in NetDocuments, home of the industry’s first Legal Context Graph, with every permission and ethical wall still enforced.

This news builds on our work together in May, when we collaborated with Anthropic to connect Claude to NetDocuments through MCP. Then, Claude could search, retrieve and draft from an organization’s own documents. Now, with Anthropic’s new artifacts for legal work, Claude can turn that work into tools a whole team can use and share.

AI agents and tools need a trusted foundation

Legal teams increasingly work across more than one AI tool. In our analysis of ILTA data, law firms using generative AI are using or exploring four to eight AI tools each, depending on size. With artifacts, Claude for Legal is a compelling alternative to legal-specific AI tools.

Many AI capabilities are available within NetDocuments ndMAX: the Legal AI Assistant, tabular review, AI search, and 45+ ready-to-go AI apps in the ndMAX Studio library, all under the organization’s existing permissions.

Our strategy is to meet lawyers where they work: in NetDocuments, in in Claude, or in the AI tools they prefer. Wherever lawyers work, we deliver the same governed context, so AI tools don’t waste time and tokens reconstructing a matter the organization already understands.

The payoff is practical. Lawyers can find relevant work across the organization more easily, and the knowledge behind it stays available when a senior lawyer or partner leaves.

Governed at the source

Claude connects to NetDocuments through MCP, the open standard Anthropic introduced. NetDocuments delivers the content and context Claude needs, and the organization’s governance comes with every request.

Governance policies are set once, in NetDocuments, and enforced in real time:

• Every request runs under the user’s current permissions. When access changes, the next request reflects it, so Claude gets only what that user is allowed to see at that moment.
• Ethical walls and matter-level restrictions apply automatically, as soon as they’re updated.
• AI exclusion and data loss prevention policies keep designated content out of AI requests.
• Audit trails capture every request.
• IT stays in control, switching the connection on or off for the organization.

That’s what centralized governance means in practice. Legal teams don’t have to rebuild their rules in every AI tool they add or wonder whether each one kept up when a wall went up or someone left a matter team. The policy lives in one place, and every connected tool is held to it.

The two layers work together. Anthropic’s sharing settings govern who can see an artifact, with support for ethical walls between teams. NetDocuments governs what Claude can retrieve in the first place.

Context, not just files

MCP connectors are quickly becoming standard. As we wrote in June, MCP is standard; what an AI tool gets through it isn’t.

Some connections hand the AI a stack of files. But an organization’s real knowledge is in how its work connects: the matter, the history, the people and the permissions. That’s what our Legal Context Graph maps. We introduced it in May as the industry’s first legal context graph, and it continuously connects every matter, document and communication while respecting existing permissions and ethical walls.

Why does that matter for Claude? Because when AI starts with the right context, it doesn’t have to rebuild the matter from scratch on every question. In our own Legal Context Engineering Benchmark, holding the model constant and changing only the context, the Legal Context Graph cut the cost of a correct answer by 48% with essentially the same quality. For a 2,000-person organization asking four million AI questions a year, that could mean nearly $1 million in annual savings. Legal teams can bank that saving or reinvest it: 7% more accurate, still 18% cheaper.

Through MCP, Claude draws on that same Legal Context Graph, so it can start every task from the organization’s institutional knowledge, not a single session upload. The context is built once and shared across every user session and AI agent, so each new tool and each new artifact makes that investment worth more.

One foundation for the AI tools you choose

Claude is one of the AI tools that connect to NetDocuments through the same governed MCP server, alongside ChatGPT, Google Gemini Enterprise, Microsoft Copilot, Perplexity, Harvey, Legora and more. For a CIO, that means one security review. For a knowledge leader, it means the organization’s expertise shows up wherever its lawyers work. For a managing partner or general counsel, it means the investment in context doesn’t start over when the next tool arrives.

The tools and models will keep changing. An organization’s knowledge, and the governance around it, is the foundation that lasts.

Getting started. The NetDocuments connector for Claude is available to ndMAX Enterprise customers. Find it in Anthropic’s legal connector directory, browse the new legal artifacts gallery, and see every AI tool we support on our AI and MCP integrations page. To turn on MCP for your organization, contact your NetDocuments representative.