Slack Launches Slack Code to Let AI Agents Build Software Inside Team Channels

Slack launched a coding environment on Thursday called Slack Code, allowing engineering teams and autonomous artificial intelligence agents to write, test, and release software inside group chat channels.

Credit: Salesforce/Slack

The software update changes how developers interact with AI coding assistants. Instead of writing prompts inside private browser windows or local code editors, developers can summon AI tools directly into public team conversations.

The Operational Mechanism

When a user tags an AI coding assistant in a Slack channel, the platform automatically generates a temporary, project-specific channel designed specifically for software construction.

Slack outlined the core workflow in its official product statement:

"With Slack Code, when you have an idea or need to build a new feature, update a web page, or fix a bug, you simply tag in a coding agent like Anthropic's Claude or Cognition's Devin, and that agent then spins up a code channel to tackle the task."

Slack detailed the collaborative mechanics within the announcement:

"There, everyone has full visibility to the conversation, can audit code diffs, get live previews of the agent's output, give feedback, and approve the work before it ships."

Channel Architecture

Every dedicated code channel splits its interface into four dedicated workspace tabs:

• Conversation: The central discussion feed displaying prompts, developer comments, and agent updates in chronological order.
• Plan: An auto-generated task breakdown outlining the logical steps the agent intends to take.
• Code Diffs: A comparison screen displaying modified lines of code, striking through deleted text and highlighting newly written lines.
• Live Preview: A dynamic rendering window showing real-time HTML and user interface outputs generated by the code.

Security Boundaries and Lifecycle Controls

To maintain code security, Slack built safety guardrails into the channel structure.

When an AI agent finishes generating code, it cannot deploy changes to production systems automatically. High-risk actions require explicit human sign-off from designated team members inside the channel.

Once human reviewers approve the code and complete the task, the channel archives itself automatically. The platform retains recorded conversations, code diffs, and review decisions as an audit log for compliance and record-keeping.

Market Context and Partner Ecosystem

Salesforce acquired Slack in 2020 for $27.7 billion and has converted the chat application into the central user interface for its AI agent platform.

Salesforce CEO Marc Benioff previewed the system late Wednesday by posting a demonstration video on social media platform X before Thursday's formal launch.

Salesforce committed $300 million toward Anthropic AI tokens earlier this year, building toward broader integration between the chat tool and external AI systems.

Slack Code launches with immediate support for five third-party AI agents:

• Claude Code (Anthropic)
• Devin (Cognition)
• Vercel Agent (Vercel)
• GitHub Copilot (Microsoft)
• ChatGPT (OpenAI)

Distribution and Pricing Structure

Slack released Slack Code to all customer plans on Thursday, making the feature available to free workspaces alongside paid enterprise accounts.

Creating and operating code channels carries no additional fee from Slack. Workspaces must purchase individual AI agent access separately through partner vendor subscriptions.

To simplify customer access to these third-party agents, Salesforce announced plans to merge the AppExchange, Slack Marketplace, and AgentExchange into a unified corporate storefront.