Anthropic has unveiled Claude Tag, a persistent artificial intelligence agent designed to integrate seamlessly into Slack as an always-on teammate. Currently available in research preview for Claude Enterprise and Claude Team customers, Claude Tag represents a significant step beyond traditional on-demand chatbots by maintaining continuous context and proactively engaging in conversations.
The feature allows users to tag @Claude in Slack channels to receive insights, ask questions, or assign tasks. Unlike earlier Slack integrations that required explicit prompting every time, Claude Tag follows along with channel discussions, accumulating knowledge over time. This persistent awareness enables it to understand team workflows, ongoing projects, and even institutional history.
Evolution of Slack AI Integrations
Anthropic has been building toward this capability for months. Earlier versions of Claude's Slack integration allowed direct messaging or one-off tagging for help with analysis, summarization, or code review. Claude Code in Slack could also route coding tasks from channel mentions to full web-based coding sessions, posting results back into the thread. However, these tools lacked memory across interactions. Each request started fresh, without awareness of past conversations or the broader context of a project.
Claude Tag solves that by maintaining a single Claude identity per channel or workspace. Everyone in a Slack channel can interact with the same Claude instance, which remembers what it has seen and done. Administrators control which tools, information sources, and channels each Claude identity can access. This scoping ensures that a Claude set up for legal work, for example, cannot glean insights from engineering channels or vice versa.
Ambient Mode: Proactive Assistance
The most notable feature of Claude Tag is its ambient mode. Rather than waiting to be asked, Claude monitors assigned channels and proactively intervenes when it determines that a team would benefit from a reminder, a summary, or a piece of context pulled from another part of the organization. This mimics the behavior of a helpful human colleague who notices when a deadline is approaching or when a relevant document has been shared elsewhere.
When assigned a task, Claude Tag breaks it into stages and works through them using the tools it has access to, responding in a Slack thread with the output. But the ambient mode goes further: it can flag updates about ongoing projects, follow up on threads that have become dormant, and surface information from across channels that the team might have missed. Anthropic describes this as working with a real colleague that can produce work in public view with far greater context and understanding than before.
Organizational Context as Competitive Landscape
The push for persistent, context-aware AI in the workplace is intensifying. Microsoft has been building Work IQ, an intelligence layer expressed through Copilot that draws on Microsoft Graph to understand roles, collaboration patterns, and organizational structure. Startups like Viktor have raised significant funding to embed AI coworkers directly into Slack and Teams. Glean, which recently surpassed $300 million in annual recurring revenue at a valuation of $7 billion, is developing a permissions-aware knowledge graph that sits between AI models and enterprise data.
Claude Tag represents Anthropic's focused answer to the same problem. Rather than building a horizontal intelligence layer spanning all enterprise applications, Anthropic plants its AI inside the one surface where most knowledge work already occurs: team chat. The bet is that persistent presence in Slack, combined with cross-channel memory and admin-controlled scoping, is sufficient to accumulate the institutional context needed for an AI agent to be genuinely useful.
Privacy and Compliance Considerations
The privacy implications of an always-on AI that follows workplace conversations and autonomously decides when to intervene are substantial. Employees may feel uneasy about an AI that is always listening and recording interactions. Compliance teams will scrutinize data handling, especially in regulated industries. Anthropic's admin-scoping controls are the structural answer to these concerns, but the real test will come when enterprise customers deploy Claude Tag at scale and observe how workers respond to an AI that is constantly present.
Anthropic has emphasized that each Claude identity stays scoped to the channels defined by administrators, and the system does not leak information between different Claude instances. However, the broader question of data privacy in AI-enhanced Slack channels remains an active discussion point among enterprise security professionals.
Availability and Future Plans
Claude Tag is launching in research preview today for existing Claude Enterprise and Claude Team customers. Anthropic says it is working to bring Claude Tag to other platforms in the coming weeks, but for now Slack is the only supported surface. This limitation constrains the feature's reach but also allows the company to gather focused feedback during the preview phase. The narrow approach is deliberate, letting Anthropic refine the product before expanding to other collaboration tools.
As businesses continue to explore how AI can augment teamwork without disrupting existing workflows, Claude Tag offers a glimpse into a future where AI teammates are persistent, context-aware, and increasingly autonomous. The success of such tools will depend not only on their technical capabilities but also on how well they integrate into the social and cultural dynamics of workplace communication.
Anthropic's move reflects a broader industry trend: treating AI not as a separate application to be opened and closed, but as an always-present participant in the digital workspace. Whether that becomes a source of productivity or friction will be determined by the experiences of early adopters in the coming months.
Source: TNW | Anthropic News