Anthropic launches Claude Tag, an always-on Slack teammate, amid IPO filing

Anthropic shipped Claude Tag in research preview, an in-Slack version of Claude that users tag into channels where it works alongside the team. Per TechCrunch and Reuters, it's positioned not as a passive bot but as proactive and 'multiplayer' — with its own identity, persistent memory, and the ability to operate across tools, integrations and compute environments, governed by admin-controlled data access for Enterprise and Team customers. Boris Cherny, Anthropic's Claude Code lead, said it has 'totally changed how we use Claude' internally over the past few months.

The pitch is that embedding Claude where work already happens — channel by channel, message by message — makes AI assistance feel inline with org-wide activity rather than a separate destination. Andrej Karpathy amplified this framing on X (21K+ likes), calling it 'a new paradigm' but flagging that the value depends on doing the hard under-the-hood engineering across tools, memory and security to make it 'just work.'

Strategically, Claude Tag is an enterprise land-grab on Slack, a surface Microsoft (Teams/Copilot) and Google (Workspace) both contest. It also dovetails with Anthropic's business momentum: the company reportedly hit a $965B valuation — surpassing OpenAI by some measures — and has filed confidentially for an IPO, with cross-source pulse noting ~75% growth in paying consumers since January.

Caveats: 'research preview' means limited availability and unproven reliability; an always-on agent reading channel history raises obvious data-governance and security questions, which is why admin controls feature prominently. Watch adoption metrics, whether Slack/Salesforce reacts, and how Claude Tag's memory and proactivity hold up against the inevitable noise of being mentioned everywhere.