Admins on Claude Team and Enterprise plans can now connect a company app once and have every matching teammate signed in the first time they use it in Claude. The ClaudeDevs account announced auto-connect on X on 3 November 2026, and Claude for Business said on LinkedIn that it is in beta for eligible plans.
For a small team, this removes the slowest part of a rollout. It also puts every access decision on one admin screen, so check what Claude verifies before you switch it on.
What changes for a team
Until now, enabling a connector (Anthropic’s name for a link between Claude and another app) gave nobody access. Anthropic’s connectors help page says each person still “needs to authenticate individually” after an owner adds one.
With what Anthropic calls Enterprise-managed auth, the admin authorizes the connector for the organization and the team inherits it. The auto-connect help article says users see no sign-in page or consent screen. Anyone whose email has no matching account in the app gets the usual Connect prompt.
There are two routes. The identity provider route is generally available but needs single sign-on (SSO, one company login for every app), and Okta is the only supported provider so far. The beta route is Claude-issued: Claude vouches for the signed-in member itself, so no SSO is needed. That is the one most small teams can use.
Which apps work
The help article lists 21 connectors for managed auth, including Asana, Canva and Slack. Only 14 work with the Claude-issued beta: Amplitude, Apollo.io, Atlassian, Datadog, Dropbox, Figma, Granola, HubSpot, Linear, Mixpanel, Notion, PagerDuty, Semrush and Supabase. A team without Okta cannot auto-connect Slack yet.
Any maker of a Model Context Protocol server (MCP, the open standard behind Claude connectors) can add support. Our explainer on how AI agents use outside tools covers the basics.
Setup and its limits
For each app, an admin chooses Claude-issued in Organization settings and copies an issuer URL unique to the organization. An admin of the other app adds that URL as a trusted issuer. After a connection test, the admin picks who gets it, and all roles get it by default.
Anthropic gives no setup time, and we did not run it. You need an admin on both sides for each app. Piloting with one team means picking a custom role, but the Claude pricing page lists role-based access as Enterprise only. We could not confirm how the Team plan handles this.
Where it can bite
The help article is blunt: “Claude checks organization membership only, not SSO, MFA, or the connector’s own sign-in rules.” Claude sends each member’s email address and a unique identifier to the app, and matching is by email.
Offboarding is the weak spot. Removing someone from Claude stops new sign-ins, but access already granted in an app “lasts until it expires.” Anthropic tells admins to remove leavers from each app before their address is reused, or a new hire with that address could land in the old account.
One control helps. Owners can set each connector tool to always allow, needs approval or blocked, so Claude could read HubSpot records without writing back. It is the same kind of narrowing AWS applied to agents running Step Functions workflows.
Auto-connect has no separate price. The pricing page lists Enterprise at $20 per seat per month plus usage at API rates, and showed no Team seat price when we checked.
What to do this week
List the apps your team connects to Claude by hand and check them against the 14 in the beta. Pick one read-heavy app, set its write tools to needs approval, and have two colleagues test it. Before widening it, write down who removes leavers from each connected app.
Sources
Get the weekly email
The week's most useful AI news, a tutorial and an interview.



