Claude brings zero-touch MCP setup to enterprises

One of the quickest ways to kill enterprise AI adoption is the OAuth gauntlet. If IT enables a great new tool, but two thousand employees each have to click through five different authorization screens or paste personal API tokens, most people simply give up.

Anthropic is tackling that exact headache. Claude now features enterprise-managed authorization for MCP connectors, allowing IT administrators to provision connected tools across their entire company in one fell swoop through their existing identity provider.

Zero-touch setup via Okta

Until now, connecting tools like Jira or Linear to Claude required two separate hurdles. Admins had to enable the connector for the company, and then every single employee had to manually authorize their own connection.

Enterprise-managed authorization removes that second hurdle entirely. IT admins configure permissions once inside their identity provider, starting with Okta, and scope them by existing user groups.

When employees open Claude, their connected tools are already waiting for them on first login. That seamless access carries over everywhere Claude operates, including Claude chat, Claude Code, and Cowork.

For security teams, it brings AI integrations under the same roof as the rest of their corporate stack. Admins can mandate that connectors only link via the IdP, preventing employees from accidentally hooking up personal accounts to work systems. Because identity checks run smoothly in the background, IT can also slash access token lifespans. When an employee leaves the company, revoking access in Okta cuts off Claude connector permissions almost immediately.

A growing ecosystem of connectors

The feature is built on an open extension to the Model Context Protocol, meaning any custom or third-party connector can adopt it. Available for Claude Team and Enterprise plans, it launched with support for Asana, Atlassian, Canva, Figma, Granola, Linear, and Supabase, before expanding general availability to Datadog, Notion, and Slack, with Exa, Miro, and Zoom on deck.

Companies like Hubspot, Ramp, and Webflow have already deployed it. Before this, tools like Supabase forced teams to either hand out org-owner permissions or distribute personal access tokens. Now, identity groups dictate who can query data safely through Claude.

Why it matters

For everyday employees, this transforms AI from a clunky utility into a native part of the workday. You do not need to understand what an MCP server is or copy credentials; if your role grants you Figma or Jira access, Claude already knows how to collaborate with those tools on day one.

We think Anthropic is playing the enterprise game remarkably well here. Benchmark victories make great headlines, but enterprise software budgets are won in IT and security audits. By anchoring MCP directly to Okta and corporate directory groups, Anthropic removes the administrative inertia that usually slows workplace rollouts.

This also cements MCP as the leading enterprise standard for AI context integration. Competitors that still rely on per-user OAuth prompts or manual API key management will look increasingly amateurish to corporate procurement officers. The main catch is that companies must already have their directory roles and permission groups strictly in order, because whatever access rules live in your IdP will now be inherited directly by the model.