Slack Code Puts Claude, ChatGPT, and Devin in Your Group Chat — AI Coding Goes Multiplayer

Introduction: Coding Agents Get a Group Chat

On August 20, 2026, VentureBeat reported that Salesforce's Slack had launched Slack Code, a product that drags AI coding out of the terminal and into the group chat. Instead of wrestling with a coding agent alone in your IDE, the whole team can summon, watch, and steer agents like Claude Code, ChatGPT, Vercel's v0, and Cognition's Devin side by side, in the channel where the work is already being discussed.

It's a small-seeming move with a big implication: the place where your team already talks becomes the place where software gets built — because the next unlock isn't smarter agents, it's visible ones.

What Slack Code Actually Is

Slack Code is a new way to interact with coding agents inside chat, with the entire team in the loop. Per SiliconANGLE's launch coverage, it ships with connectors for agents from multiple vendors — Anthropic's Claude Code, Vercel's v0, Cognition's Devin, and OpenAI's ChatGPT among them — with more third-party connectors buildable via API. Three availability details matter:

• Any Slack plan works — available now, not gated to enterprise tiers — but you bring your own agent subscription; Slack is the collaboration layer, not the model vendor.
• Multi-vendor by design. Claude and ChatGPT can work in the same channel, positioning Slack as neutral ground above the model wars.
• Everything persists. Conversations, diffs, canvases, and repo links stay archived, so teams can retrace how a feature came to be.

How a Code Channel Works

The mechanics are Slack-native. A team is discussing a feature or a bug in an ordinary channel; instead of fragmenting to separate terminals:

• Tag the agent. A team member @-mentions a coding agent in the discussion.
• A Code Channel opens. Slack spins up a dedicated agent workspace.
• Output lands in a canvas. Changes appear as before/after diffs, alongside a working preview the team can poke at.
• Feedback loops in-channel. Teammates review, redirect, and approve before the work drifts off course.

Catching misdirected work early is the quiet killer feature: it cuts the expensive churn of building, reviewing, and tearing down code the team never wanted.

From Solo Player Sport to Team Sport

The pitch behind Slack Code is less about agents than organizational physics. “All this individual work is a solo player sport,” Slack CMO Ryan Gavin told SiliconANGLE. “The knowledge and the speed at which you're acting as an individual stays with you as an individual, and you're not getting an exponential return for the organization.”

Slack frames the payoff as accelerated learning: when agents work in the open, context compounds instead of evaporating in private sessions. Analyst Rebecca Wettemann of Valoir Research put it in ROI terms: “For companies to unlock the full value from AI, it has to move beyond individual productivity gains to operational transformation” — a shift from individual copilots to “embedded, contextual AI orchestration.”

To be clear, nobody claims terminals and IDEs are dead. Slack Code aims to own the scaffolding — ideation, prototyping, and early feedback — where alignment, not typing, is the bottleneck.

The Vibe-Coding Angle: When Non-Engineers Ship

The Verge described the new feature as “collaborative vibe-coding channels,” and that framing matters. Because agents run where anyone can talk to them, non-developers get a sanctioned path into the build process: marketing can weigh in while a UI change is still a draft, and a business unit can prototype an internal app — then tag an engineer to harden it once the idea proves out.

That's vibe coding's democratization promise with an audit trail: experts can see what's happening and step in before deployment, not after the incident review.

Why the Coding Tool Wars Just Moved Into Chat

Slack Code lands in a crowded, consolidating arena. Cursor just launched Origin, its agent-native GitHub rival; GitHub is defending its orbit; and Microsoft is pushing agents deeper into Teams — the most obvious head-to-head rival, since Teams plus Copilot can approximate the same chat-plus-agents recipe.

Slack's differentiation is neutrality. By hosting Claude, ChatGPT, v0, and Devin in one place, it becomes the Switzerland of coding agents — an orchestration layer that wins no matter which model leads this quarter. That reframes the buyer's question from “which coding agent should we standardize on?” to “which collaboration surface should orchestrate them all?”

The Watch-Outs: Injection, Sprawl, and Review Debt

Public channels are exactly where agent risks get amplified:

• Prompt injection: anything pasted into a channel — a bug report, a customer quote, a link — is untrusted input your agent may act on. Visibility helps detection; prevention still requires permissions and sandboxing.
• Agent sprawl: when anyone can summon an agent, you can end up with five vendors and no governance. Sanction specific agents up front.
• Review debt: visible diffs invite rubber-stamping. Seeing code isn't reviewing it — someone still owns the merge.

How to Pilot Slack Code Without Regretting It

• Start with one sanctioned agent (Claude Code and ChatGPT are the most battle-tested) before opening the floodgates.
• Pin a channel charter — what agents may build, which repos they may touch, who approves merges.
• Treat pasted content as hostile — no agent should act on instructions inside a bug report.
• Measure churn, not output. Track rework before declaring victory.

Frequently Asked Questions