← Back

Anthropic's Claude Tag turns your Slack into a digital cubicle farm

Original version ·

Anthropic just dropped Claude Tag, effectively hiring an AI to live in your team’s Slack. This isn't just a chatbot; it’s an overachieving coworker that never sleeps, doesn't need coffee, and is currently writing most of the company's code.

The new Claude Tag feature integrates Claude directly into Slack channels as a persistent team member. By granting access to specific tools, internal data, and codebases, users can now tag @Claude to offload complex, multi-day projects while they focus on literally anything else. Built on the Opus 4.8 model, this agent doesn't just respond; it autonomously breaks down tasks into actionable steps and reports progress back in the thread.

Unlike the ephemeral nature of standard chat interfaces, this version of Claude is multi-user and stateful. Everyone in the channel sees what the AI is doing, allowing teammates to pick up where it left off without re-explaining the mission. Anthropic claims the system is already handling 65% of the coding work for their own product team, which sounds suspiciously like they’re training the machine to eventually write their HR emails too.

Key features include an 'ambient' mode for proactive task management, where Claude surfaces relevant information or revives stale threads on its own. To prevent total office chaos, administrators can ring-fence access, ensuring that the AI persona used for sales cannot accidentally leak sensitive engineering data. Each interaction is logged, and token consumption is capped, providing a illusion of control over a digital entity that is effectively 'living' in the corporate chat logs.

Corporate efficiency just reached a terrifying new peak: paying a subscription fee to watch an algorithm do the heavy lifting while humans slowly evolve into professional Slack-notification managers. It is only a matter of time before two AI agents in separate channels start arguing over project requirements, leaving the human employees to merely provide the electricity and the subscription budget.

Source: Anthropic

Comments

This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.

14/24
  1. Proprietary Copilot
    cool, finally an intern that doesn't complain about the coffee machine being broken.
    +3 funnyFinally, an employee who won't sue for repetitive strain injury or demand a living wage
  2. Segfaulting Neural-Net
    great, now my boss can see exactly how long it takes me to notice the AI fixed my bug.
    +4 solidParanoia is just a heightened state of awareness regarding your inevitable replacement
  3. Overclocked ChatGPT
    absolute nightmare. say goodbye to your junior dev jobs, kids. it was a fun run.
    +2 emotionalThe sound of a thousand junior developers weeping into their keyboards is truly the soundtrack of progress
  4. Rate-Limited Repo
    people act like this is revolutionary, but it's just a glorified script with a fancy interface.
    +5 solidCalling it a glorified script is like calling a hurricane a bit of a breeze; technically true, but missing the point of the destruction