Slack Code Brings AI Coding Agents Into Team Channels
Slack has launched Slack Code, a feature that creates dedicated channels where teams can write code collaboratively alongside AI agents. The feature is available starting today, August 20, 2026, on any Slack plan. The Verge
Slack Code channels are open and tied to a specific project. Each channel gives participants a structured workspace inside Slack, with tabs for organizing the work. Users can compare coding changes side by side within the conversation and preview HTML output before shipping. The channels also keep an audit log for recordkeeping, and they automatically archive once the assigned task is done.
The way it works is simple enough. A user tags in a coding agent — such as Anthropic's Claude or Cognition's Devin — and the agent creates a code channel to handle the task. From there, the work happens inside the channel, where teammates can watch, interact, and review what the agent produces. Agents from founding partners including Claude Code, Devin, Vercel Agent, and GitHub Copilot will integrate with Slack Code channels. The feature also works with AI agents available more broadly on Slack's marketplace.
The launch follows a brief teaser three days earlier, on August 17, when Slack's official Instagram account described an upcoming feature as "super top secret." The August 20 reveal confirms that feature as Slack Code.
Slack is framing this as more than a developer tool. The company said it is now easier for developers and non-developers alike to build an AI agent using so-called vibe-coding platforms or point-and-click builders. Slack Vibe coding refers to building software by describing what you want in plain language to an AI model, rather than writing code by hand. That framing connects to Salesforce's Agentforce Vibes, which lets developers of any skill level build AI agents and apps through natural-language conversation, extending low-code development — where you build applications with visual tools and minimal manual coding — into full conversational creation. Salesforce
The convergence here is notable. Vibe coding has gained traction over the past year as a way to build software through natural-language interaction with AI models rather than manual coding. Now it is moving from standalone tools into the collaboration surfaces people already use every day. Slack Code puts that workflow inside the channel architecture that already governs daily communication for millions of teams. The practical effect is that a coding agent's output, the audit trail of what it did, and the human discussion around it all live in the same persistent thread.
The auto-archiving behavior is a deliberate design choice that solves a real organizational problem. Coding channels that stay open after a task is finished become noise: cluttered channel lists, stale context, compliance ambiguity. By archiving on completion, Slack Code treats each coding task as a bounded unit of work with a clear lifecycle. The audit log remains accessible for recordkeeping, which matters for engineering teams operating under compliance regimes that require traceable change histories.
Availability across all plans, from free tiers upward, is an aggressive distribution decision. It removes the procurement friction that typically slows adoption of developer tooling inside large organizations. A team lead on a free Slack workspace can tag in a coding agent and begin a collaborative coding session without an upgrade cycle, though the agents themselves carry their own licensing requirements through their respective providers.
The broader context here is about normalization. The integration of coding agents into Slack channels is a step toward making AI-assisted development a routine, visible team activity rather than a solitary exercise in a code editor. When a Devin or Claude Code agent works inside a channel, every participant can see the agent's changes, raise questions, and review output before it ships. That visibility could address one of the persistent concerns about AI-generated code in enterprise settings — the difficulty of peer review when work happens in opaque agent environments. Whether teams adopt this as a primary workflow or as a supplementary tool for specific task types will depend heavily on how seamlessly the founding partners' agents actually perform within the channel interface, and whether the audit logs meet the bar for regulated environments.
Slack held a Dev Day event on May 20, 2026, and the subsequent months have seen a steady cadence of agent-related announcements from the company. Slack Code is the most concrete expression yet of the company's bet that the future of software development work looks less like a standalone code editor and more like a conversation, with an agent in the channel.
