Anthropic launches Claude Tag in beta, putting a persistent multiplayer AI teammate inside Slack
Anthropic shipped Claude Tag on June 23, 2026, and the critical word in its announcement is "persistent." Earlier Claude integrations in Slack treated each conversation as a fresh session. Claude Tag maintains a single channel-scoped identity that every team member can address, picks up mid-task hand-offs without re-briefing, and builds institutional memory across conversations over time.
What shipped
Claude Tag is available today in beta for Claude Enterprise and Claude Team customers. It runs on Opus 4.8. Administrators pair it with a Slack workspace in four steps: connect to the workspace, grant tool access, set a monthly token-spend limit for the organization, and test in a private channel before opening it to the team.
The channel-scoping is the structural difference from the old Claude in Slack integration. Within a given channel, there is one Claude identity that all members share. Anyone can see what it has been working on. Anyone can pick up a conversation from where the last person left off. A Claude configured for the legal team cannot access the engineering channel's memory, per Anthropic's announcement, and a Claude set up for engineering cannot read sales data or tool access. Administrators define the scope; Claude Tag stays inside it.
Persistent memory accumulates from the channels Claude Tag is assigned to, plus any additional Slack channels or external data sources administrators grant read access to. The system does not pull from private channels unless explicitly permitted. Anthropic describes the result as tacit institutional knowledge: context about a team's active projects, terminology, and priorities that normally lives only in people's heads.
Claude Tag also operates in an ambient mode. With ambient behavior enabled, Claude monitors assigned channels and proactively surfaces information, flags forgotten threads, and sends updates without waiting to be tagged. Per The Next Web, this is the feature that most separates it from a conventional chatbot: rather than responding only to explicit requests, it intervenes when it judges that a team would benefit from a reminder or a summary pulled from elsewhere in the organization.
For specific tasks, Claude Tag breaks the work into stages, executes them using whatever tools it has access to, and posts results in a Slack thread. Supported tasks include code generation, data queries, and support ticket resolution, per the announcement.
Migration from the legacy Claude in Slack app is automatic. Administrators have 30 days to opt in. Anthropic said it is issuing a launch credit to eligible Enterprise and Team organizations to support initial rollout. The migration deadline is August 3, 2026.
The 65% figure
Anthropic cited one internal statistic in its announcement: 65% of its own product team's code is now generated by an internal version of Claude Tag. Per Anthropic, the same pattern had spread beyond engineering to product metrics tracking, support ticket triage, and bug root-cause analysis.
That figure is a vendor self-disclosure, not an audited measurement. Read it as Anthropic's characterization of its own internal workflow, not as a benchmark claim about productivity gains an outside team should expect to replicate.
Why it matters
The enterprise AI assistant market has been fragmented across session-based chatbots, one-off automations, and point integrations that require constant re-briefing. Claude Tag attacks that fragmentation directly. The persistent identity solves the most common complaint about AI tools in collaborative settings: every new person who picks up a project has to re-explain context from scratch. Channel-scoped memory addresses that by design.
TechCrunch noted that the organizational-context race is not unique to Anthropic. Microsoft's Copilot with Microsoft Graph, Glean's intelligence layer, and Snowflake and Databricks' data-platform positioning all converge on the same thesis: whichever system accumulates the most usable knowledge about how a company actually operates wins the enterprise deployment. Claude Tag is Anthropic's bid to make Slack the accumulation point.
For teams already running automation pipelines through tools like n8n or Make, Claude Tag reduces the need for manual trigger layers. A persistent agent that monitors channels and takes action on agentic tasks without being explicitly invoked each time covers some of the same ground those automations handle, though in a less structured way.
The enterprise positioning matters in one other respect. Anthropic's confidential S-1 filing process has been reported as in progress. Claude Tag is the first product the company launched that turns Claude into a persistent institutional presence rather than a query-response tool. That framing is relevant to enterprise contract renewals, to seat-license expansion, and to the usage metrics that appear in revenue reporting.
Context
Claude Tag follows a line of Anthropic Slack integrations. The original Claude in Slack launched in early 2026 for direct messages and on-demand channel tags. Claude Code in Slack, which routes coding requests from channel mentions to full coding sessions and posts results back into threads, shipped in December 2025, per TechCrunch. Claude Tag supersedes the original integration and extends Claude Code's Slack footprint into a persistent, memory-building identity. Anthropic described it as "the beginning of an evolution of Claude Code" in its announcement.
The September 2026 competitive calendar is relevant context. OpenAI's enterprise offerings and Google Workspace's Gemini integration both compete for the same Slack-adjacent enterprise coordination market. Claude Tag's ambient mode in particular has no direct published equivalent in either competitor's current Slack or Teams integration.
What to watch next
Two signals are worth tracking. First, whether Anthropic announces expansion to Microsoft Teams or Notion in the "coming weeks" the brief noted. The announcement page described Slack as the launch platform, with expansion to "the many other places" teams work as an explicit goal. Second, any measurable productivity data from an Enterprise customer outside Anthropic itself. The 65% code-generation figure is from internal Anthropic usage. A named customer reference with external adoption numbers is the next verification step. Anthropic has not disclosed contract terms for Claude Tag or per-seat pricing separate from existing Enterprise and Team subscription costs.
Sources
- Introducing Claude Tag (Anthropic, June 23 2026): primary source, vendor announcement
- Anthropic launches Claude Tag, an always-on AI teammate that lives in your Slack channels (The Next Web, June 23 2026): secondary coverage
- Anthropic's Claude Tag is learning your company, one Slack message at a time (TechCrunch, June 23 2026): secondary coverage