Anthropic introduces Claude Tag, a constant AI companion that integrates into your Slack channels.

      TL;DR: Anthropic has introduced Claude Tag, a persistent Slack AI that engages in discussions, learns context, and actively flags updates and tasks.

      Anthropic is rolling out Claude Tag in research preview, which functions as an "always-on Claude" within Slack, acting as a continuous AI teammate. Users can tag @Claude for insights during conversations and to delegate tasks. The feature is available to Claude Enterprise and Claude Team customers starting today.

      Claude Tag builds upon Anthropic's existing Slack integrations that allow users to message Claude directly or tag it in channels for immediate assistance. The Claude Code feature can also manage coding tasks from channel mentions to full coding sessions online, updating the thread accordingly. However, Claude Tag introduces persistent context and memory that earlier tools lacked.

      “As Claude tracks its channel conversations, it gains deeper insights into the work,” stated Anthropic. With permission, Claude can also fetch information from other channels in the organization, enabling it to accumulate institutional knowledge over time instead of resetting with each interaction.

      Everyone in a Slack channel can utilize a single Claude identity, allowing all members to see Claude’s prior work and continue discussions where others left off. System administrators define which tools, information, and channels each Claude identity can access, ensuring that a Claude set up for legal tasks doesn’t influence discussions in the engineering channel, for instance.

      When given a task, Claude Tag breaks it down into stages and utilizes available tools to respond within a Slack thread. The standout feature is its ambient mode, which actively engages in conversations to keep teams informed, point out relevant information from across the organization, and follow up on neglected threads or tasks.

      The ambient mode distinguishes Claude Tag from a typical chatbot. Instead of waiting for prompts, Claude monitors its assigned channels and interjects when it believes the team could benefit from a reminder, summary, or context drawn from another area within the company. Anthropic describes this interaction as akin to collaborating with a real colleague who has access to more context and understanding.

      Organizational context is becoming the crucial area for enterprise AI. Microsoft is developing Work IQ, an intelligence layer through Copilot that utilizes Microsoft Graph to understand roles, collaboration patterns, and company structure. Startups like Viktor have raised substantial funds to integrate AI coworkers directly into Slack and Teams.

      Glean, recently reaching over $300 million in annual recurring revenue with a $7 billion valuation, is creating a permissions-aware knowledge graph connecting AI models with enterprise data.

      Claude Tag addresses this issue more narrowly and potentially more effectively than broader solutions, focusing on embedding AI within the primary platform for knowledge work: team chat. The strategy is that a consistent Slack presence, combined with cross-channel memory and admin-defined scopes, can amass the institutional context necessary for a functional AI agent.

      The privacy considerations are significant. An ever-present AI that observes workplace conversations and autonomously decides when to engage will be subject to scrutiny from employees and compliance teams. Anthropic's scoping controls provide a structural answer to these concerns, but the true test will occur when enterprise clients implement it at scale and evaluate employee reactions to an always-listening AI.

      Anthropic has indicated plans to expand Claude Tag to other platforms in the upcoming weeks. For now, Slack remains the sole platform, which limits its reach but simplifies its complexity, a considered trade-off for a research preview.