Salesforce and Anthropic have introduced Claude Tag, a new feature that embeds Anthropic’s Claude artificial intelligence model directly into Slack channels, allowing teams to assign tasks to an AI assistant without leaving their workplace conversations.
Instead of interacting with AI in separate chat windows, users can now summon the assistant by typing @Claude in any authorized Slack channel.
Claude can analyze discussions, access approved business tools and code repositories, complete assigned tasks, and respond within the same conversation thread.
According to the companies, the feature is designed to make AI collaboration visible to entire teams rather than limiting interactions to individual users.
“Slack is the only layer in the AI stack where teams work together. Bringing Claude Tag into Slack is about making AI multiplayer,” said Rob Seaman, executive vice president and general manager of Slack.
“Instead of a private back-and-forth, Claude Tag shows up in the open — in your channels, alongside your team — where it can see the real context of how your organization works and make the whole company smarter in the process. This virtuous cycle of sharing context, building together, and generating more context is how enterprises will finally move faster, compete differently, and make AI work at real scale.”
Claude Tag appears in Slack as a dedicated team member with its own account and permissions. Administrators determine which channels, tools, documents, and code repositories it can access, while users can assign tasks such as writing code, analyzing data, handling support requests, or investigating software bugs.
The AI completes the work asynchronously and posts its results back into the relevant Slack thread.
Anthropic said the shared approach differs from conventional AI chat interfaces because everyone in a Slack channel can view Claude’s work, contribute additional instructions, and continue previous conversations without starting from scratch.
The system also builds context over time by remembering information from channels it has been authorized to access, while maintaining separate knowledge boundaries for different departments.
The feature also supports what Anthropic calls “ambient” behavior, allowing Claude to proactively surface relevant information, follow up on unresolved discussions, and continue working on assigned tasks over hours or days.
According to Anthropic, the company has already adopted an internal version of Claude Tag, with approximately 65% of its product team’s code now generated through the system.
The company said its use has expanded beyond software engineering to product analytics, customer support, and debugging workflows.
Claude Tag is available in beta for Claude Enterprise and Team customers using Slack. It replaces the previous Claude in Slack application, with existing customers given a 30-day migration period.
The feature currently runs on Anthropic’s Opus 4.8 model.
