Claude Tag, Anthropic's AI assistant for Slack channels, now supports personal connectors, meaning you can access your own private tools and data within team conversations without exposing sensitive information to the entire channel. Previously, Claude could only use connectors that administrators attached to the channel itself. This update, rolling out now on Team plans with Enterprise following, fundamentally changes how teams can collaborate while maintaining individual privacy and security.

What Are Personal Connectors and Why Do They Matter?

Personal connectors are integrations that link Claude to your individual accounts and tools, separate from shared channel resources. Think of them as private bridges between Claude and services you already use. When you ask Claude a question in a channel that requires accessing your Google Drive, your calendar, your CRM accounts, or your staging deployment environment, Claude can now reach those resources directly through your personal login.

This matters because most of the information teams actually need sits behind individual access controls. A project timeline might live only in your calendar. Your open deals exist only in your assigned CRM view. A critical deployment plan might be a document only you can open. Before this update, Claude would hit a wall when you asked it to reference these personal resources in a channel conversation. Now it can retrieve and analyze that information, then present it to the team in whatever way you decide.

How Do Personal Connectors Work in Practice?

The mechanics are straightforward. When you ask Claude to access a personal connector for the first time in a channel, Claude asks for permission. After you approve it once, Claude can use that connector in future requests within that thread. Everything Claude does through your connector appears in that tool's own audit log under your account, just as it would in a direct message. The channel's shared work stays under the service account your security team already monitors.

You maintain complete control over what Claude retrieves and what the channel sees. Anthropic offers two modes for managing this. In review mode, you see Claude's response before it posts to the channel. In auto mode, Claude posts automatically unless it detects sensitive content that requires your review. On Enterprise plans, administrators can require review for everyone, giving security teams additional governance options.

How to Set Up and Use Personal Connectors in Claude Tag

• Connect Your Tools: Link your personal accounts to Claude, such as Google Drive, your calendar, your CRM, or deployment systems. There is nothing to install; the integration happens within Claude itself.
• Choose Your Review Mode: Decide whether you want to review Claude's responses before they post to the channel, or allow auto-posting with sensitive-content screening.
• Ask Claude in the Channel: Mention @Claude and request information that only you can access, such as "Check my Google Drive doc against what we've shipped" or "What's still open on my calendar for this project?"
• Manage Permissions: Disconnect any connector at any time, and control which tools Claude can access on a per-request basis.

What Governance Options Do Administrators Get?

Personal connectors give administrators more nuanced control over how tools are accessed within channels. Organizations can choose to provide a shared set of tools under a service account identity, require channel members to use only personal connectors to enforce existing role-based access controls, or decide tool by tool which approach makes sense.

This flexibility addresses a common tension in enterprise AI: teams want Claude to be helpful and access relevant data, but security teams need to ensure that access follows the principle of least privilege. Personal connectors solve this by letting Claude respect each person's existing permissions rather than requiring administrators to grant broad channel-level access.

When Should Teams Use Personal vs. Shared Connectors?

Personal connectors work best for closely supervised, collaborative work that requires pulling sensitive data. For example, a team drafting a request for proposal (RFP) response might need Claude to access pricing information or other confidential sources that are not provisioned to the entire channel. Anything Claude posts is visible to everyone in the channel, so personal connectors give you control over what data Claude retrieves, not what the team ultimately sees.

Shared connectors, by contrast, are better suited for unattended work. If you want Claude to help with continuous integration (CI) triage and incident response in an on-call channel, even after work hours, Claude needs access to shared connectors for your runbook, monitoring tools, and deployment history. Scheduled routines and any actions Claude starts on its own rely on shared connectors, not personal ones.

The distinction reflects a practical reality: personal connectors require human oversight and decision-making, while shared connectors enable Claude to operate autonomously within guardrails your security team has already approved. Organizations can mix both approaches depending on the channel's purpose and risk profile.

What Does This Mean for Enterprise AI Adoption?

Personal connectors represent a shift in how enterprise AI tools handle data access and privacy. Rather than forcing a choice between security and utility, this feature lets teams have both. Employees can ask Claude for help with work that involves their personal data, while administrators maintain visibility and control over what gets accessed and when.

The rollout begins now on Team plans, with Enterprise plans to follow. For organizations already using Claude Tag in Slack, this update removes a significant friction point: the need to either share sensitive data with the entire channel or ask Claude to work with incomplete information. As more teams adopt AI assistants in their workflows, this kind of granular permission model may become table stakes for enterprise adoption.