What Claude Tag Is and How It Differs From the Old Claude App
Claude Tag in Slack is a persistent AI coworker that joins a workspace as a named user, keeps shared memory at the channel level, and automates tasks, so teams can delegate work without repeating context every time they interact with it. Unlike the earlier Claude in Slack app, which behaved like a one-off chatbot, Claude Tag is a full Claude workspace integration that lives inside selected channels and reads ongoing conversations. Members can mention the AI Slack agent with @Claude Tag to summarize discussions, draft updates, or chase down details from tools and codebases it can access. Anthropic describes this as “more like interacting collaboratively with a teammate” than consulting a transient bot, since everyone in a channel talks to the same instance. The older connector is being deprecated, with the new Claude Tag becoming the default model for Slack automation.
A Shared Agent With Memory: From Chatbot to Persistent AI Coworker
The core shift in Claude Tag Slack behavior is memory. Instead of forgetting every exchange, Claude Tag builds channel-level history and remembers decisions, deadlines, and open questions. Gadget Review notes that Claude Tag “doesn’t wait for commands. It follows along, accumulating context from every conversation, decision, and stalled thread.” That persistent awareness means no re-briefing when someone asks for a sprint recap or a status report across multiple channels. The same Claude agent serves everyone in that space, so follow-up questions build on what colleagues already discussed. When granted permission through Anthropic’s Model Context Protocol (MCP), Claude Tag can also pull facts from other channels and tools, while scoped identities let admins keep sensitive domains, such as legal or engineering, separate. Together, these features turn Slack into a kind of AI-powered company brain instead of a series of isolated bot chats.
Proactive Assistance: Ambient Mode, Stalled Threads, and Self-Scheduled Tasks
Beyond answering prompts, Claude Tag acts like a proactive AI Slack agent when ambient behavior is enabled. It can quietly monitor accessible channels for stalled threads, unresolved questions, and forgotten to-dos, then surface them back to the team without any @mention. CNET reports that Claude Tag can proactively message a thread that has not yet been resolved and remind users about tasks across Slack channels. Anthropic says the agent can even schedule tasks for itself and work on them “over hours or days,” posting intermediate outputs as it completes multi-step work. Internally, Anthropic claims that around 65 percent of its product team’s code is written by Tag, illustrating how far the agent can go once embedded in daily workflows. This ambient awareness changes Slack automation from reactive commands into continuous, background support for ongoing projects.
Deployment, Access Controls, and What This Means for Team Workflows
Claude Tag is available in beta for organizations on Claude’s Team and Enterprise plans, where it appears as a dedicated workspace user. Admins choose which channels it can join and what tools, data sources, or codebases it may read, then configure role-based access controls on Enterprise so different teams see only the Claude Tag agents meant for them. The legacy Claude in Slack connector will be removed on August 3 as Anthropic migrates customers to the new model. In day-to-day work, teams can delegate coordination and administrative tasks directly in Slack: chasing metrics, compiling status updates across channels, drafting responses to support tickets, or investigating root causes of bugs based on existing threads. With shared context and persistent memory, Claude Tag turns Slack into a multiplayer interface to a single, ongoing AI coworker rather than a collection of short-lived bot sessions.
