On August 20, 2026, Slack launched Slack Code, allowing teams to use AI coding agents in project channels, review plans, code diffs and live previews, and discuss approvals in the same thread.
AI coding agents are no longer confined to one engineer
Previously, AI coding often happened inside one engineer’s editor or chat. Others might not see the result until a pull request existed. Slack Code changes that flow: a product manager can raise a request, and engineers, designers and collaborators can follow the progress in the channel.
When a user tags an AI agent in a Slack project channel, Slack creates a corresponding coding channel. The channel shows the agent’s plan, code changes and live preview, while members can reply with feedback and request revisions.
Requests, code changes and review stay together
A product manager could report a broken button in the channel. After the agent proposes a fix, an engineer can inspect the diff and preview before deciding whether it fits the architecture and product requirements. AI becomes a visible team role rather than a private tool.
Slack Code works with agents from partners including Anthropic, Cognition, GitHub Copilot, OpenAI and Vercel. However, availability on Slack plans does not mean every user automatically has access to every agent; access depends on the partner service, region and customer agreement.
Human approval remains the production gate
Slack says teams can pause, redirect or stop an agent. High-stakes or production actions still require expert approval. A work channel can be archived after completion, while its record remains available for later review.
Shared visibility is not a substitute for permission design. Teams still need to define which repositories an agent may access, which actions require approval and who reviews generated code. A visible audit trail does not replace code review or production controls.
What it changes for development teams
The main benefit is reducing information gaps in AI-assisted development. Requirements, discussion, code changes and review happen in one place, so product staff need less second-hand explanation and engineers can spot mismatches earlier.
Slack Code is not only about having AI write more code. It turns a coding agent into part of the team workflow: raise a request, let the agent work, review the result together and let people decide whether to adopt it.
