If you want ChatGPT inside Microsoft Teams, you no longer need custom code. As of October 2026 there are three practical routes: OpenAI's own @ChatGPT app for Business, Enterprise and Edu workspaces, your personal dot that you can message in Teams, or a shared agent that a whole team @mentions in its Teams channels. This guide explains what each one does, who it answers, whose permissions it uses, and how to set up the shared-agent route step by step.
Option 1: @ChatGPT, OpenAI's app for Microsoft Teams
Per OpenAI's Help Center (as of October 2026), @ChatGPT is available to ChatGPT Business, Enterprise and Edu workspaces. In Teams, people mention @ChatGPT in a channel message or thread where the app is available, or send it a message in a one-on-one chat. A workspace admin sets it up and decides which Teams channels and users can use it.
The part teams most often miss is identity. OpenAI's setup guide says @ChatGPT has its own service account and uses the company connections your organization configures for the assistant. It does not automatically inherit each requester's access to apps or private files, and some requests may need a person's separate authorization. When a member authorizes work that uses their personal connections, that work runs under their account. Usage for requests handled through the service account is attributed to that account, and the workspace needs credits.
- Good fit: your company already pays for ChatGPT Business or Enterprise, and you want one admin-managed assistant in Teams that uses OpenAI models and a set of approved company tools.
- Less good fit: you are not on a ChatGPT business plan, you want a different model for different jobs, or you want the admin to choose, per tool, whether the agent uses the asker's own account or one shared account.
Option 2: your dot, a personal agent you can message in Teams
Dots are OpenAI's always-on personal agents. According to OpenAI's dots documentation (as of October 2026), dots are rolling out gradually to eligible accounts, and you reach the same dot in ChatGPT, Slack, Teams, or a call. In Slack it responds to you by default, and you can instruct it to engage with others.
That makes a dot a strong personal assistant that happens to live in Teams, not a team assistant. By default a dot responds to its owner, and each person who wants a dot of their own needs an eligible account. A dot can be told to engage with others in Slack, but it is still one person's agent, running with that person's connected apps. We compare dots with team-oriented options in ChatGPT Dots alternatives for Slack and Microsoft Teams.
Option 3: a shared agent the whole team can @mention
A shared agent is one named assistant, such as @HR Assistant or @IT Help, that you add to a Teams team or group chat. Anyone in that conversation can @mention it, and everyone sees the answer in the conversation. This is the route Runbear is built for, and it is the closest match to what most people mean when they search for a Microsoft Teams ChatGPT integration for their company.
What a shared Runbear agent adds, based on Runbear's docs as of October 2026:
- Model choice per agent. On the Claude Agent SDK runtime, which is usually the default, you can pick Claude, GPT, or Gemini for each agent, so an IT agent and a sales agent do not have to use the same model.
- Per-user or shared tool access. Each tool connection is set to Per-User, where each member links their own account and the agent acts with that member's permissions, or Shared, where one account serves everyone (Runbear docs).
- Knowledge from your existing docs. Sources such as Notion, Confluence, SharePoint and Google Drive are synced through the account of the member who connected them and indexed so the agent can answer from them (Runbear docs). Sync is incremental, not a live read of every page.
- The same agent in Slack. If part of your company is on Slack, the same agent can run there too.
How to add a ChatGPT-powered agent to Microsoft Teams with Runbear
Create the agent first: give it instructions, pick a GPT model (or another model), and connect the knowledge sources and tools it needs. Then deploy it to Teams. The steps below follow Runbear's Microsoft Teams docs.
- In the Runbear dashboard, open your agent and go to the Chatbots tab.
- Click the Microsoft Teams tile and choose Use @Runbear. If the Runbear app is not in your Teams organization yet, install it from the dashboard or from Microsoft AppSource. To show the agent under your own name and icon, create a branded bot instead.
- In Teams, click Add to a Team and choose the team or group chat where you want the agent.
- The bot posts a card with a Complete Setup button. Click it to go back to the Runbear dashboard.
- Enter your Teams name and click Add. The agent is now active in that team or group chat.
If a team or group chat does not show up during setup, the app is probably not installed there yet. Install it in that conversation and refresh the connection page. Each team or chat you deploy to becomes its own connection with its own settings.
What it costs: seats versus usage
@ChatGPT sits inside your ChatGPT Business, Enterprise or Edu workspace and draws on workspace credits (OpenAI), so its cost follows how your company licenses ChatGPT. Runbear charges by usage, not by seat: one subscription covers every teammate who uses your agents in Slack or Microsoft Teams. Team is $99 per month with 20,000 credits and Business is $399 per month with 100,000 credits (Runbear pricing). If you are weighing seat costs, see Cutting ChatGPT Team Plan Costs.
Which option should you pick?
- Your company is standardized on ChatGPT Business or Enterprise and one admin-managed assistant is enough: start with @ChatGPT.
- You want help with your own work across your own apps: use a dot, if your account is eligible.
- Several teams need their own named agents in Teams channels, with model choice and per-user tool permissions, and you are not on ChatGPT Business, Enterprise or Edu, or you want pricing by usage rather than by ChatGPT plan: use a shared agent such as Runbear for Microsoft Teams. For a broader look at how shared agents work, see shared AI agents for teams, and for a head-to-head with OpenAI's options, Runbear vs ChatGPT.
FAQ
Is there an official ChatGPT app for Microsoft Teams?
Yes. As of October 2026, OpenAI offers @ChatGPT in Microsoft Teams for ChatGPT Business, Enterprise and Edu workspaces. An admin sets it up, and people mention it in an approved channel or message it in a one-on-one chat (OpenAI Help Center).
Does @ChatGPT in Teams use my own permissions?
Not by default. OpenAI says @ChatGPT uses its own service account and the company connections configured for it, and it does not automatically inherit each requester's access to apps or private files. Some requests may ask a person to authorize separately (OpenAI).
Can I add ChatGPT to Teams without coding?
Yes. @ChatGPT is installed by a workspace admin, dots are connected from your dot's profile in ChatGPT, and a shared agent platform such as Runbear deploys to Teams from a dashboard: install the app, add it to a team or group chat, and complete setup. Building your own bot with code is still possible but means maintaining it yourself.
Can one AI agent answer everyone in a Teams channel?
Yes, that is what a shared agent is for. A Runbear agent added to a team or group chat can be used by the people in that conversation who @mention it, and tools can be set so the agent acts with each asker's own account where that matters.
Sources
- OpenAI Help Center, ChatGPT with Slack and Microsoft Teams (accessed October 24, 2026): https://help.openai.com/en/articles/20001536-chatgpt-with-slack-and-microsoft-teams
- OpenAI Help Center, Setting up and managing ChatGPT in Slack and Microsoft Teams (accessed October 24, 2026): https://help.openai.com/en/articles/20001538-setting-up-and-managing-chatgpt-in-slack-and-microsoft-teams
- OpenAI, Dots documentation (accessed October 24, 2026): https://learn.chatgpt.com/docs/dots
- Runbear docs, Microsoft Teams: https://docs.runbear.io/channels/microsoft-teams
