Slack is embedding AI coding agents directly into group chat channels, fundamentally shifting how software teams collaborate on code generation and review.
The Salesforce-owned messaging platform launched Slack Code today, a feature that integrates leading AI coding tools into dedicated Slack channels. The platform now supports Anthropic's Claude Code, Cognition's Devin, GitHub Copilot, and Vercel's agent. Teams can watch AI agents work in real time, provide steering instructions, review code changes, and deploy directly from chat rather than toggling between terminals and messaging apps.
The move targets a growing gap in how developers actually work. AI coding agents like Devin and Claude Code excel at autonomous task completion but live in isolated environments. Developers must switch contexts to see results, share findings with teammates, or incorporate feedback. Slack Code collapses that workflow. A developer can spin up an AI agent in a channel, watch it write code, have teammates comment on specific lines, request modifications, and merge changes without leaving Slack.
This is Slack's most direct play into the AI developer tools category, where competition has intensified sharply. GitHub Copilot dominates IDE-based code completion. Devin, funded at a $500 million valuation, positions itself as a junior developer that can handle end-to-end tasks. Claude Code, launched in beta by Anthropic, lets developers grant Claude direct file access. Vercel's agent focuses on full-stack deployment automation. Each operates in its own sandbox.
Slack's advantage is distribution. Slack hosts over 750,000 active workspaces and reaches millions of daily users. By making Code available across any Slack plan at launch, the company removes friction for adoption. Teams already using Slack for standups, pull request reviews, and deployment notifications now have native access to multiple AI agents without installing new tools.
The feature also addresses a real collaboration gap. Code review today happens asynchronously in pull request comments or synchronously in video calls. Watching an AI agent generate code in real time, then immediately discussing it with the full team, changes the review dynamic. Non-engineers in product or design can observe what the AI is building. Senior engineers can catch errors before deployment. Junior developers learn by watching agents work.
Slack faces pressure to defend its core territory against Microsoft Teams and to demonstrate stickiness as enterprises evaluate AI tooling spend. By integrating with multiple AI coding platforms rather than building proprietary models, Slack positions itself as infrastructure. This play echoes Slack's integration ecosystem strategy. The company doesn't need to beat Claude or Devin at code generation. It needs to become the hub where those agents operate.
The timing matters. Devin raised significant capital positioning itself as a replacement for junior developers. Investors and enterprises are testing whether AI agents can actually ship production code. Slack Code lets those agents operate transparently within existing team workflows, reducing friction and increasing adoption velocity.
Salesforce spent $27.7 billion acquiring Slack in 2021 and has since navigated criticism about integration roadmaps and AI strategy. This launch shows willingness to embed third-party AI tools rather than force Salesforce's own Einstein models. That pragmatism increases the feature's credibility with developers who already have AI agent preferences.