Anthropic is launching a novel service named Claude Tag, an “always-on Claude” that resides in Slack, functioning as an AI colleague. This new option — enabling users to tag @Claude for insights in conversations and task assignments — will be available in beta for Slack users of Claude Enterprise and Claude Team.

Claude Tag represents an advancement of several current integrations. Users can currently DM @Claude within Slack or mention it in channels for immediate assistance, while Claude Code in Slack channels coding requests from mentions to comprehensive coding sessions online, providing updates within the thread. 

However, Claude Tag introduces a layer of ongoing context and memory that was challenging to maintain with earlier tools. “As Claude engages with its channel, it continuously learns about the work,” stated Anthropic. “Claude can also autonomously collect information from other parts of the organization, provided it is authorized to access other channels.”

With Claude Tag, every participant in a Slack channel can connect to a single Claude identity, meaning “anyone can see what Claude has been doing and can continue the discussion from where the last individual left off.” System administrators will determine which tools, data, and channels Claude is permitted to access, and each Claude identity will remain confined to the channels defined by the admins, ensuring that a Claude configured for legal tasks cannot influence the engineering channel, for instance.

When tasked with a specific assignment, Claude Tag will deconstruct the task into phases and will progress through them utilizing whatever tools are available, responding in a Slack thread with what it has accomplished. Yet, Claude Tag also includes an ambient mode that proactively engages in the conversation to keep the team informed, highlight items from around the organization, and follow up on neglected threads or responsibilities.

Anthropic claims this creates the experience of “collaborating with a genuine colleague — one capable of producing work publicly, with significantly enhanced context and understanding than previously possible.”

That context is an increasingly vital component of enterprise applications, and Anthropic is not alone in this focus. Microsoft is also emphasizing context through Graph, manifested in Copilot and Work IQ. Snowflake and Databricks are positioning their platforms as back-end support containing implicit organizational insights that agents can utilize. Glean is also developing an intelligence layer that comprehends company context, bridging the model and enterprise data.