Anthropic says a user can ask Claude Tag in a Slack channel to use connectors attached to that user’s Claude account, such as a calendar, drive, CRM accounts, or staging deployments. This starts a distinct record because no prior New Runtime record was supplied for this specific change in Claude Tag’s channel access model.
The mechanism is an identity boundary rather than a blanket expansion of channel permissions: personal-connector activity is performed under the requester’s account and appears in that tool’s log, while channel-connector work remains under the channel’s service account. Anthropic also says the requester can review a response before posting it, or select auto mode; Enterprise administrators will be able to require review. Anthropic says scheduled routines and self-initiated Claude actions still use only administrator-attached shared connectors.
The immediate implication is that teams can bring private, role-scoped context into a shared workflow without provisioning that context to every channel member. The release matters now because it makes the boundary between supervised collaboration and unattended automation explicit: personal connectors fit request-driven work, while shared connectors remain the basis for recurring or autonomous operations. The evidence boundary is that these claims come from Anthropic’s product announcement and describe intended availability and controls, not independent evidence of reliability, adoption, or security outcomes. Watch whether Enterprise rollout preserves the stated separation of audit trails, review requirements, and connector-level governance in real deployments.