Anthropic unveiled Claude Tag on Tuesday, a new AI product that embeds its latest model directly into Slack as a shared teammate that can be summoned with @Claude.
It offers a persistent assistant that stays in the channel where work happens.
How the tool works inside Slack
Administrators link Claude Tag to a Slack workspace, assign it access to selected tools and data sources, set spending caps, and choose the channels it may operate in. Once configured, any team member can tag @Claude with a request—such as drafting a pull request, pulling sales figures, or running a data analysis. The AI then breaks the request into steps, executes the tasks using the granted tools, and posts the outcome in a thread.
Four features that set it apart
First, the AI is multiplayer. Within a channel there is one Claude that interacts with all participants, allowing anyone to see its progress and pick up a conversation where another left off. This differs from most Slack AI integrations that create a separate instance per user.
Second, it accumulates context over time. As Claude follows a channel’s discussions, it retains relevant information, so users don’t need to repeat project details. If granted permission, it can also draw context from other channels, though Anthropic says it will not pull data from private channels.
Third, ambient behavior lets Claude proactively surface information it deems useful and follow up on idle threads. The agent can monitor multiple channels and decide what humans need to know without being prompted.
Fourth, the system works asynchronously, handling projects that span hours or days. Anthropic notes its own product teams now spend much of their time delegating tasks to many Claudes in parallel.
Related: Digital Shield: Best Online Reputation Management Tools for Businesses
Enterprise controls and governance
Security and administration are built into the design. Separate Claude identities can be created for different functions, each scoped to specific channels and tools. Memory and data stay within those boundaries, meaning a Claude used for sales work won’t share information with one handling engineering tasks.
Token‑spend limits can be set at the organization or channel level, and a complete log records every action Claude takes and the user who requested it. Such logging is often a prerequisite for compliance, audit, or regulatory requirements, and its absence has turned away many potential enterprise buyers of AI collaboration tools.
To migrate from the older Claude in Slack app, administrators must opt in within 30 days.
Market context and competition
Claude Tag appears amid a crowded push to embed AI agents in the Slack channel, which many see as the new “agentic operating system.” Salesforce, after acquiring Slack, rolled out more than 30 enhancements to Slackbot earlier this year, turning it into a full‑spectrum enterprise agent. OpenAI introduced “Workspace Agents” that can act across Slack, Google Drive, Microsoft apps, and other platforms. Perplexity’s “Computer” agent and Cognition’s Devin also operate directly inside Slack, while Microsoft has brought GitHub Copilot to Teams.
The rationale is simple: enterprises juggle thousands of applications, and employees lose up to 40 % of productivity to context switching. AI that lives in the channel where work is coordinated gains both distribution and data advantages, absorbing institutional knowledge that makes it harder to replace.
Anthropic’s build‑up to Claude Tag
Anthropic first added Claude to Slack in October 2025, allowing two‑way connectivity for individual productivity. By January 2026, interactive Claude apps expanded the integration to include tools like Canva, Figma, and Box. Parallel developments—Claude Code for enterprise plans, Claude Managed Agents for composable APIs, and the Opus 4.8 model—provided the technical foundation for the new product.
Benchmark improvements in Opus 4.8 showed a rise in agentic coding scores from 64.3 % to 69.2 % and a knowledge‑work score increase from 1,753 to 1,890, highlighting the model’s enhanced ability to work independently for longer periods.
Related: The Boys’ last season struggles culminated in strong finale
Financial stakes and growth
Anthropic raised $65 billion in a Series H round in May, reaching a $965 billion valuation, and reported a run‑rate revenue of $47 billion. Revenue from Claude Code alone topped $2.5 billion, with enterprise customers now accounting for more than half of that figure. Deloitte’s deployment of Claude across 470,000 employees in 150 countries illustrates the scale of adoption Anthropic is targeting.
Industry forecasts predict the global agentic AI market will expand from $9.14 billion in 2026 to $139 billion by 2034, and Gartner expects a significant share of enterprise applications to feature task‑specific AI agents by the end of the year, up from less than 5 % a year earlier.
Risks and considerations for buyers
Enterprise buyers must weigh vendor lock‑in, as Claude Tag’s accumulated channel memory could make migration costly. Governance around ambient monitoring also raises questions: an AI that surveils information flows and decides what to surface may outpace existing policy frameworks, especially in regulated sectors.
Pricing remains opaque. Anthropic says Claude Tag uses token‑based spending with administrative controls, but continuous monitoring and long‑running tasks could shift consumption patterns compared with on‑demand tools. Reliability is another factor; recent reports of infrastructure strain suggest that an always‑on agent could be vulnerable to downtime that impacts productivity.
Future outlook
Anthropic plans to extend Claude Tag beyond Slack, aiming to let users tag @Claude in other collaboration spaces such as Microsoft Teams, email, and project‑management tools. If successful, the product could cement a new category of AI worker that never sleeps, never forgets past discussions, and never needs re‑on‑boarding.
For many technology leaders, the issue is no longer whether an AI agent will appear in the communication layer, but whether their organizations are prepared to manage the responsibilities that come with a permanent, autonomous teammate.
Leave a Reply