Slack is adding a new way for teams to work with AI coding agents without moving the work out of the conversation where decisions already happen. The feature, called Slack Code, creates dedicated channels for project-specific coding tasks, giving teams a shared place to request work, review results, and approve changes before they ship.
The launch reflects a broader shift in how software teams are being asked to use AI agents: not only as private assistants, but as visible participants in collaborative work. Slack’s pitch is that code generation, feedback, review, and record keeping can sit together in one open channel rather than being scattered across separate tools and messages.
What Slack Code Adds
Slack Code introduces open, project-specific code channels designed for collaboration with AI agents. A team member can bring an agent into a task, and the agent can then create a dedicated code channel for that assignment.
These channels are meant to keep the work visible. Instead of an AI agent producing output in isolation, the surrounding team can follow the conversation, inspect the changes, and decide whether the result is ready to move forward.
The feature includes dedicated user tabs, code comparison tools, and HTML previews. That matters because AI-assisted coding is not just about generating a patch. Teams also need to understand what changed, see what the output looks like, and decide whether it meets the goal.
In practical terms, Slack Code gives teams a single workspace for several steps that often sit apart:
- Starting a coding task with an AI agent
- Keeping the discussion visible to the team
- Reviewing code diffs before approval
- Previewing HTML output before a project ships
- Preserving an audit log for later reference
How AI Agents Fit Into The Workflow
Slack describes the experience as one where a user can tag in a coding agent for work such as building a new feature, updating a web page, or fixing a bug. Examples named in the source include Anthropic’s Claude and Cognition’s Devin.
From there, the agent creates a code channel focused on the task. That channel becomes the place where the team can see the work unfold, provide feedback, and approve the result before it goes out.
This is different from treating an AI coding agent as a private drafting tool. Slack Code is built around shared visibility. The agent’s output is not only something one person receives; it becomes part of a channel where others can check the conversation and the code changes.
Slack is framing the feature as a way to work with AI agents like teammates. The important point is not that the agent replaces the review process. The feature is designed around review, feedback, approval, and records, which are all signals that Slack expects people to stay involved before work ships.
Why Dedicated Code Channels Matter
AI-assisted coding can quickly create coordination problems if the work happens in private threads, separate development tools, or disconnected chats. A generated change may be useful, but a team still needs to know what was requested, what the agent produced, who reviewed it, and whether anyone approved it.
Slack Code attempts to solve that by giving each assignment its own open space. Project-specific channels make the task easier to follow because the discussion, diffs, previews, feedback, and approvals are organized around the same piece of work.
The automatic archiving feature also matters for workspace hygiene. Slack says these coding channels are designed to archive themselves automatically when their dedicated assignments are complete. That suggests the channels are meant to be temporary work areas, not permanent rooms that accumulate after every task.
The audit log is the other important piece. When AI agents are involved in code work, record keeping becomes part of the workflow. Teams need a way to look back at what happened, especially when a change was discussed, revised, previewed, and approved inside a collaborative space.
Availability And Agent Support
Slack Code is available starting today on any Slack plan. It works with AI agents available through Slack’s marketplace.
Slack also named several founding partners whose agents will integrate with code channels. That list includes Claude Code, Devin, Vercel Agent, and GitHub Copilot.
Those integrations are central to the launch because Slack Code is not presented as a standalone coding engine. It is a Slack workspace layer for collaborating with agents that teams may already use or discover through the marketplace.
For software teams, the immediate value is straightforward: fewer handoffs between chat, review, preview, and approval. For Slack, the move positions the platform as a place where AI coding work can be coordinated openly, instead of happening off to the side and returning only when there is output to inspect.
The Bigger Takeaway
Slack Code is a workflow product as much as an AI product. Its core idea is that coding agents become more useful to teams when their work is visible, reviewable, and tied to the same collaboration space where decisions happen.
The feature does not remove the need for human judgment. In fact, the launch emphasizes the opposite: teams can audit diffs, preview output, give feedback, and approve work before it ships. That places AI-generated work inside a team process rather than outside it.
As teams experiment with vibe coding and agent-assisted development, that structure may be the part that matters most. Slack Code is betting that the future of AI coding is not just faster code generation, but clearer coordination around the work agents produce.