product update

Slack Launches 'Code Channels' to Put AI Coding Agents Into Group Chats

TL;DR

Slack has launched Slack Code, a feature that gives AI coding agents dedicated project channels where developers and non-technical staff can watch, review, and redirect their work in real time. Anthropic, GitHub, OpenAI, Cognition, and Vercel are building integrations, bringing Claude, Copilot, ChatGPT, and Devin into Slack's group chat environment.

3 min read
0

Slack has introduced Slack Code, a feature that moves AI coding agents out of individual developer terminals and into shared project channels where entire teams can watch, review, and redirect their work.

The Salesforce-owned company describes the feature as "multiplayer AI." Tagging an agent from within a Slack conversation prompts it to create a dedicated code channel for the task. From there, team members can follow the agent's progress, view live HTML previews, review proposed code changes, and redirect the agent's approach mid-task. When work concludes, the channel can archive itself, leaving a searchable log of what happened.

Slack says each channel corresponds to a specific project, intended to keep detailed agent activity — such as explanations for rewriting an authentication layer — out of main team channels.

The company is pitching Slack Code beyond engineering teams. According to Slack, product managers, designers, and other non-technical staff can participate without opening a terminal or learning a specialized tool. In one example Slack provided, a product manager flags a bug, asks an agent to draft a fix, and loops in an engineer to review the diff before the agent opens a pull request and merges the change. Slack summarizes the pitch as: "No ticket, no meeting, no waiting – just a fix, shipped."

Slack states that humans retain control throughout the process. Anyone in a code channel can pause, redirect, or stop an agent, and the company says higher-stakes actions — such as pushing code to production — are meant to be routed to a human expert for approval. Slack also says agents operating in code channels inherit the platform's existing permissions, security policies, and administrative controls, rather than requiring a separate management layer.

Slack is not building this alone. Anthropic, Cognition, GitHub, OpenAI, and Vercel are developing integrations for Slack Code, with Claude, Devin, Copilot, and ChatGPT among the agents expected to operate in code channels. Slack says APIs will eventually open to the broader developer community, with code channels potentially expanding beyond software development into areas like marketing and legal document review.

Alongside Slack Code, the company introduced two related features: Agent DMs, which allow direct conversations with AI agents, and a new Agents tab for finding agents, reviewing past sessions, checking status, or halting an agent that has gone off track. An "Add to Slack" option is designed to simplify deploying agents built on third-party platforms.

Slack says users can "get started with Slack Code today," but the company's announcement does not specify which subscription plans include access, how broadly the feature is rolling out, or which named agent integrations are live at launch. The Register has asked Slack for clarification on these details.

What this means

Slack Code addresses a real gap: AI-assisted coding today mostly happens invisibly, with a developer working alone in a terminal before surfacing results for review. Moving that process into a shared channel could genuinely improve visibility for non-technical stakeholders and create an audit trail of agent decisions.

But the launch raises unanswered questions. Slack hasn't disclosed pricing, plan tiers, or launch availability for the named integrations (Claude, Copilot, ChatGPT, Devin), meaning the practical reach of this feature is unclear. There's also an unresolved tension in the pitch: putting agent activity in a group chat that "everyone can watch" doesn't guarantee anyone actually reviews it carefully — it just changes where the unread notifications pile up. Whether code channels meaningfully improve oversight of AI-generated code, or just add another channel to ignore, will depend heavily on how teams actually adopt the workflow rather than how it's marketed.

Related Articles

product update

Anthropic Expands Claude Cowork to Mobile for All Paid Plans

Anthropic announced that Claude Cowork, its workspace-focused feature, is now available on mobile and web for all paid plans. The rollout began last month exclusively on Anthropic's most expensive tier before expanding today.

product update

OpenAI Launches 'Private Safety Processing' to Detect Misuse Without Storing Enterprise Data

OpenAI has built a system called Private Safety Processing that detects misuse patterns across multiple interactions without storing customer inputs or outputs. The company says it only receives narrow safety signals—type and severity of activity—while data stays encrypted on customer infrastructure.

product update

OpenAI Previews 'Private Safety Processing' to Detect Abuse Without Retaining Customer Data

OpenAI is previewing Private Safety Processing to select customers, an automated system that monitors for misuse across multiple sessions without retaining any customer data. The move directly contrasts with Anthropic's July policy allowing 30-day data retention for 'covered models' like Fable.

product update

AWS Adds Runtime Domain and Publish-Date Filters to AgentCore Web Search

Amazon Bedrock AgentCore's Web Search connector now supports per-request domain filtering and published-date windows, enforced server-side. The update, shipped as connector version 1.2.0, also expands availability to AWS regions in Dublin and Tokyo.

Comments

Loading...