Slack just made its biggest push yet into the AI coding workflow, and it’s doing it by leaning on what the platform has always been good at: channels. The company announced Slack Code today, a feature that lets teams spin up dedicated project channels where AI agents like Anthropic’s Claude or Cognition’s Devin do the actual building while humans watch, review, and approve. No more tab-hopping between Slack, GitHub, and whatever agent tool you’re experimenting with this week.

The mechanics are straightforward. Tag a coding agent in Slack, and it creates a code channel for the task. From there, everyone on the team gets visibility into the agent’s conversation, can audit code diffs as they happen, preview HTML output before anything ships, and sign off on the work. Once the job’s done, the channel archives itself automatically and leaves behind an audit log. Slack’s pitch is that agents should work “like teammates” — a framing that’s clearly spreading across the industry, given that xAI is pushing a similar message with its Grok Bot agents.

Availability is unusually broad for a launch like this. Slack Code works on any Slack plan starting today, and it’s designed to integrate with agents from the Slack marketplace. Founding partners include Claude Code, Devin, Vercel Agent, and GitHub Copilot, which Slack says will integrate “seamlessly” with code channels. That’s a meaningful roster — it covers the two most visible autonomous coding agents right now plus the incumbent developer tooling players.

The auto-archiving detail is worth pausing on. Most AI coding tools generate a sprawling mess of conversations, diffs, and abandoned experiments that nobody ever cleans up. Building archival into the channel lifecycle means teams get a natural paper trail for compliance and review without having to think about it. Whether that’s enough to make Slack the default home for vibe-coding remains an open question — GitHub and the major IDEs aren’t ceding that ground quietly.

But Slack has one advantage those tools don’t: the conversation is already happening there. The people who need to approve a feature, question a design decision, or kill a project entirely are already in Slack all day. Putting the agent’s work in that same stream lowers the friction dramatically. The agents are coming to where the decisions get made, not the other way around.