Slack Code: Where Teams and Agents Build Together

Slack has launched Slack Code to integrate AI agents into existing workflows, emphasizing the value of conversation and collaboration in the software development process. While AI agents previously operated in isolation within personal browser tabs or terminals, Slack Code brings them into the team's public channels, providing a transparent and collaborative development environment.

When you mention an agent for complex coding tasks, Slack Code automatically creates a code channel. This channel is organized around elements necessary for the work, such as code diffs, planning documents, and live HTML previews, and automatically invites relevant team members. Once the task is complete, the channel is automatically archived, remaining as a record in Slack like an audit log.

Anthropic, Cognition, GitHub, OpenAI, and Vercel have joined as partners for this launch. In particular, GitHub has integrated Copilot into code channels, based on the vision that "humans set the direction and agents close the loop," helping to translate shared intent into executable software.

Developed based on feedback from Slack's internal engineering team, this feature focuses on reducing noise by separating complex agent executions from standard channel threads and gathering the right people at the right time. As a result, over 70% of code channels are created and closed within a day, shortening the time from idea to merged PR and reducing the context required for reviews.