Slack wants to drag AI coding out of the terminal and into the group chat
Slack announced Slack Code, embedding AI coding agents—including Anthropic’s Claude Code, Cognition’s Devin, GitHub Copilot, and Vercel’s agent—into dedicated Slack channels so teams can collaboratively watch, steer, review, and ship software. When tagged, an agent creates a project‑specific channel, posts code diffs, live previews, and a running plan, then archives the channel with a searchable audit trail. This shifts AI‑assisted development from isolated terminal sessions to a multiplayer, auditable workflow that can involve engineers, product managers, and designers.
⚡ Key Takeaways
- Slack Code integrates four named agents: Claude Code, Devin, GitHub Copilot, and Vercel’s agent.
- Agents spin up a project‑specific Slack channel and expose code diffs, live previews, and execution logs in dedicated tabs.
- The cloud‑based agent form factor enables auditable records and scaling across teams, reducing bottlenecks in code review and verification.
- Integration requires each customer to have access to the partner agents; Slack merely provides the channel orchestration.
- The workflow supports cross‑role collaboration (e.g., designers dropping Figma files) and automatically creates pull requests and verification demos.
Engineers can now embed AI‑generated code directly into their existing Slack collaboration layer, gaining real‑time visibility, auditability, and multi‑role input without leaving the chat platform, which accelerates delivery and reduces reliance on isolated developer terminals.
✅ Practical Steps
- Enable Slack Code on your Slack workspace (available on any plan at launch) and ensure your team has access to the desired partner agents (Claude Code, Devin, GitHub Copilot, Vercel’s agent).
- Tag a coding agent in any Slack conversation to automatically spin up a dedicated code channel where the agent will post diffs, previews, and execution logs.
- Use the generated channel to review the agent’s output, add assets (e.g., Figma files), and merge the resulting pull request once verification is complete.
Want the full story? Read the original article.
Read on VentureBeat AI ↗