Slack Code vs. GitHub Copilot in Teams: AI Agents Enter Chat
Smart Market Insight Editorial
Editorial Team
This article may contain affiliate links. We only recommend tools we’ve personally tested. Read our full disclaimer.
Within 24 hours of each other, Slack and Microsoft Teams both turned their chat channels into a place where AI coding agents can actually write and ship code. Slack shipped "Slack Code" on August 20, 2026, and GitHub followed on August 21 with a Teams integration that lets anyone mention @GitHub to spin up a Copilot coding session — plus, on the same day, an overhauled GitHub-in-Slack experience of its own. Three separate launches, one clear signal: the terminal is no longer the only place AI coding agents live.
If your team already lives in Slack or Teams, this matters more than the average product update — it changes where bug reports and feature requests actually turn into pull requests. Here's what each product does, how they differ, and whether either is worth turning on yet.
Quick Take
- Slack Code (launched August 20) turns a tagged request into a dedicated channel where an AI coding agent works while your team watches, steers, and approves the result before anything ships.
- GitHub Copilot in Microsoft Teams (public preview, August 21) lets anyone mention
@GitHubin a channel, thread, or DM to start a Copilot cloud-agent session; only people with write access to the repo can actually trigger changes. - GitHub also shipped a separate, revamped Copilot experience directly inside Slack the same day — distinct from Slack Code, and available even if you never touch Slack Code's dedicated channels.
- Slack Code itself is free on every plan, including free workspaces; you still need your own paid access to whichever agent you tag (Claude Code, ChatGPT, Devin, Copilot, or Vercel's agent). Copilot in Teams runs on metered cloud-sandbox billing while in preview.
What Slack Code Actually Does
Slack Code isn't a single chatbot bolted onto Slack — it's a new type of channel. According to Slack's announcement, the unit of work is the channel itself: someone mentions an agent — Claude Code, OpenAI's ChatGPT, Cognition's Devin, GitHub Copilot, or Vercel's coding agent — on a bug report or feature request, and Slack spins up a dedicated channel scoped to just that task.
From there, the agent works in the open. Teammates can jump in, add context, ask questions, or redirect the approach mid-task, and nothing merges or deploys until a person in the channel signs off — making AI-written code a team activity with a visible paper trail, not something one engineer quietly runs solo and reports back on.
The rollout detail that matters most for smaller teams: Slack Code is available today on every Slack plan, free tier included. Slack isn't charging anything extra for the channels — the cost sits with whichever agent vendor you already pay (or don't) for Claude Code, Copilot, Devin, or Vercel's agent.
What Changed in Microsoft Teams
Microsoft's answer landed a day later and works differently. Per GitHub's changelog entry, mentioning @GitHub in any Teams channel, thread, or direct message starts a Copilot cloud-agent session. Anyone in that conversation can weigh in, ask questions, or help plan the work — but only participants with actual write access to the target repository can trigger Copilot to make changes.
The pitch is about closing the gap between "we discussed it" and "it's done": if a standup surfaces a bug, someone can hand it to Copilot mid-meeting and let it start investigating in a secure cloud sandbox before the call ends. It's explicitly a public preview, still subject to change, with cloud sandbox usage billed on metered compute and memory rates rather than folded into a flat subscription fee.
GitHub didn't stop at Teams. The same day, it shipped a separate, redesigned native Copilot experience inside Slack itself, built on the existing GitHub-Slack integration rather than Slack Code's new channel format. That gives Copilot users two distinct ways into Slack — the classic @GitHub-in-Slack integration, and a seat as one of Slack Code's five launch-partner agents.
Why This Landed the Same Week
This isn't coincidence so much as convergent strategy. Every major coding-agent vendor has spent 2026 racing to get out of the IDE and into wherever teams already coordinate work — Cursor, GitHub Copilot, and Claude Code have all shipped standalone or cloud-agent modes this year, and Cognition's Devin has kept pace with its own coding model and a security-scanning product shipped in July. Chat is the obvious next venue: it's where product managers and support staff already report bugs, and none of them want to learn a CLI to file one.
For Slack's parent Salesforce, embedding five outside agents as launch partners — instead of pushing only its own Agentforce — bets that openness keeps Slack central to how teams work, even as coding shifts toward agents its rivals build. Microsoft's move is more defensive: folding Copilot into Teams keeps developers inside Microsoft's ecosystem end-to-end, from the meeting where a bug gets discussed to the pull request that fixes it, without ever opening Slack.
How the Two Approaches Compare
| Slack Code | GitHub Copilot in Teams | |
|---|---|---|
| Launched | August 20, 2026 | August 21, 2026 (public preview) |
| Agents supported | Claude Code, ChatGPT, Devin, GitHub Copilot, Vercel agent | GitHub Copilot only |
| Cost of the feature itself | Free on every Slack plan | Metered cloud-sandbox billing (compute/memory) |
| Who can trigger changes | Anyone who can tag the agent; approval gate before shipping | Only users with write access to the repo |
| Work surface | New dedicated channel per task | Existing channel, thread, or DM |
The core difference isn't pricing — it's scope. Slack Code is agent-agnostic, letting a team standardize on whichever coding agent it already pays for. Copilot in Teams deepens Microsoft's own stack, closer to GitHub Copilot's other recent agent updates than a true Slack Code rival.
What This Means If You Use Slack or Teams for Dev Work
If your team is already on Slack and paying for Claude Code, Copilot, or another supported agent, Slack Code costs nothing extra to try — the friction is social (does your team want AI writing code in a shared channel everyone can see) rather than financial. Start it on a low-stakes bug or documentation fix before trusting it with anything customer-facing; the approval gate only protects you if someone actually reads the diff before clicking approve.
If you're a Microsoft-shop team, treat the Teams integration as one to watch rather than adopt outright — it's a public preview with usage-based billing that could still change, and GitHub's own materials aren't fully consistent yet on which Copilot plans qualify. For a broader read on Slack against its chat-platform rivals, see our Slack vs. Teams vs. Discord comparison.
Frequently Asked Questions
Is Slack Code a separate product I have to pay for? No. Slack Code's channels are free on every Slack plan, including free workspaces. You only pay for access to whichever AI coding agent you tag — through that vendor directly.
Can anyone in a Slack Code channel approve code to ship? A person in the channel must approve before anything merges or deploys, but Slack doesn't restrict that to a specific role — teams are expected to enforce their own review discipline, same as with any pull request.
Does GitHub Copilot in Teams work the same way as Slack Code?
No. It's limited to GitHub Copilot, mentioned via @GitHub in a Teams channel, thread, or DM, and only users with write access to the target repository can trigger code changes.
Is GitHub Copilot in Teams generally available? No, it's a public preview as of August 21, 2026, still subject to change. Cloud sandbox sessions used for the async work are billed on metered compute and memory rates.
Which AI coding agents work inside Slack Code? At launch: Claude Code (Anthropic), ChatGPT (OpenAI), Devin (Cognition), GitHub Copilot, and Vercel's coding agent.
Bottom Line
Slack Code and GitHub Copilot in Teams both point at the same shift — AI coding agents moving from solo terminal tools into the shared chat channels where teams already coordinate work — but they solve it differently. Slack Code bets on being the neutral ground where any agent can show up; Microsoft's Teams integration bets on depth within its own Copilot stack. Neither replaces the judgment of a developer reading a diff before it ships, and both are new enough that the sensible move is a small pilot, not a full rollout. If you're choosing which coding agent to standardize on before deciding which chat integration to trust it with, our Cursor vs. GitHub Copilot vs. Claude Code comparison is the place to start.
Related Articles
OpenAI Shuts Down the Assistants API: What Breaks Today
OpenAI's Assistants API shut down today, Aug 26 2026, with no automated migration tool. Here's what broke and how builders move to the Responses API.
Codex's 1M Context Window Is Real — but OpenAI Won't Turn It On for You
OpenAI's Codex team shared how to unlock GPT-5.6 Sol's full 1M context window — but Codex caps it by default, and even OpenAI is split on the fix.
GitHub Copilot Ships a New AI Agent in Visual Studio, Off by Default
GitHub Copilot's new Visual Studio agent and built-in skills ship off by default — here's what changed and why the billing model explains it.
MCP's Biggest Spec Update Yet: What Changes for AI Agents in 2026
MCP's 2026-07-28 spec update makes the AI agent protocol stateless — here's what changed, what breaks, and why it matters for builders.