Slack’s New AI Coding Supervision Lets Teams Watch Robots Work

Slack just gave AI coding assistants a front-row seat in team conversations. Called Slack Code, this feature creates dedicated project channels where AI can write, test, and even show live previews of code changes. Instead of working in secret, the AI posts updates and teammates can jump in to approve, suggest fixes, or just observe. The goal is to catch mistakes early and make software development more transparent—kind of like turning a solo robot coder into a team sport.

The new setup works with popular AI coding tools like GitHub Copilot, Vercel, and soon ChatGPT. You can mention an AI agent in Slack, and it spins up a private channel with the right people and relevant files. Even non-technical employees can flag an issue, let the AI draft a fix, and bring an engineer in to review—bridging the gap between spotting a problem and solving it. After the work is done, the channel auto-archives, creating a searchable record of the project.

Slack is positioning this as more than just a coding aid. The same model could eventually handle marketing campaigns, legal reviews, or IT onboarding, turning Slack into a central hub for human-AI collaboration. The company also added an Agents tab to manage all AI interactions in one place, making it easier to monitor multiple assistants at once.

But it’s not all upside. More visibility doesn’t guarantee better code, and too many AI updates in a shared channel could create noise instead of clarity. Companies will still need to decide how much oversight is enough—especially as AI agents take on bigger tasks. Slack says it’s using existing user permissions to control access, but humans remain the final gatekeepers for critical changes.