Vibe Sensei for Teams

Vibe Senseiis a desktop application — a multi-agent AI cockpit where a developer runs and supervises AI coding agents, each in its own visible terminal pane, working on the developer's own code on their own machine.

The Vibe Sensei for Teams integration is the two-way bridge between that desktop cockpit and your Microsoft Teams workspace:

• Inbound — talk to your agents from Teams. @mention the Vibe Sensei bot in a connected channel. Your prompt is delivered to your online Vibe Sensei desktop app via a Microsoft Graph change-notification subscription; an AI orchestrator agent does the work and the answer is posted back into the same Teams thread.
• Outbound — your agents keep the channel posted. The integration posts AI-generated status updates into your chosen channel: a long-running mission finished or failed, a usage cap was reached, a worker agent died, or a new release shipped.

The bot reads only messages that explicitly address it — an @mention in a channel you have connected. It does not passively read or archive your Teams conversations.

The inbound channel uses the Microsoft Graph change-notification API: when you connect, the integration registers a subscription for new messages in your chosen channel. Inbound prompts can be toggled off at any time from Settings → TEAMS inside the desktop app.

Vibe Sensei for Microsoft Teams turns a Teams channel into a remote control for your own AI development agents. After you connect your Microsoft 365 account and pick a team and channel, the integration pairs that channel with your Vibe Sensei desktop installation and the code repository you are working on. From then on, anyone in the channel can hand a task to your AI orchestrator — “summarize the open PRs,” “what broke in the last build?,” “kick off the release checklist” — simply by @mentioning the bot, and the orchestrator's real answer arrives back in the thread, including any relevant links (pull requests, deployments, commits). Meanwhile, the desktop app pushes proactive notifications to the same channel so the team can follow long-running agent work without watching a terminal.

The heavy lifting happens on the paired desktop app, which must be online: agents run locally against your own code and your own AI provider subscriptions. The Teams integration itself is a thin, secure relay — the Microsoft OAuth token is stored encrypted (AES-256-GCM) server-side and auto-refreshed; requests are verified server to server via the Microsoft Graph API; and message content is relayed to your own machine rather than stored.

The integration requires a Microsoft 365 work or school account with permission to add apps to a Teams channel. Personal Microsoft accounts are not currently supported.

The Teams connection lives inside the Vibe Sensei desktop app — there is no public one-click install button because the Microsoft OAuth flow requires a signed-in Vibe Sensei account to complete securely. Follow these steps:

1. Download & sign in. Download the Vibe Sensei desktop app and sign in to (or create) your Vibe Sensei account.
2. Open a workspace with a git remote. The Teams integration binds to a code repository. Open a workspace in the app that has a git remote configured — the TEAMS settings section becomes available once an eligible workspace is active.
3. Open Settings → TEAMS. Click the Settings (gear) icon in the app. Select TEAMS from the left-hand panel. You will see a Connect Teams section with a “Connect Teams” button.
4. Click “Connect Teams.” The app opens Microsoft's OAuth consent screen in your browser. Review and approve the requested permissions for your Microsoft 365 account (work or school account required).
5. Return to the app — it confirms automatically. After you authorize in the browser, the desktop app polls for the connection. Within a few seconds the status badge switches to Connected.
6. Pick a team and channel. Click Pick a channel, select your Microsoft Team from the first dropdown, then select the channel from the second dropdown, and click Set channel. The bot is now bound to that channel.
7. Enable inbound prompts. The Inbound prompts toggle is on by default — this allows @mentions in the bound channel to run on your orchestrator. You can toggle it off at any time from the same settings panel.
8. Talk to it. With the desktop app online, @mention @Vibe Sensei in the bound channel:
   • @Vibe Sensei summarize the open PRs on this repo
   • @Vibe Sensei what broke in the last build?
   The bot posts the AI orchestrator's answer back into the same thread.
9. Receive updates. The bot posts status messages on its own when agent missions complete or fail, usage caps are hit, or new releases ship.

The bot responds only to @mentions addressed to it in the connected channel. Plain channel chatter is not read or processed. If your desktop app is offline or the channel is not yet connected, the bot replies with a short message explaining that.

The Teams integration is designed to touch as little of your data as possible: the Microsoft OAuth token is stored encrypted (AES-256-GCM) server-side and auto-refreshed; messages are only read when explicitly @mentioned to the bot; relayed message text is not stored; and your Teams content is never used to train AI models or sold to anyone.

Read the full Privacy Policy — including the dedicated section on Microsoft Teams and messaging integrations — for how Vibe Sensei collects, manages, and stores data.

Our Terms of Service govern use of the Vibe Sensei service, including the Teams integration.

To disconnect Teams, open Settings → TEAMS → Unbind channel in the desktop app. This stops all inbound and outbound activity for the bound channel. To revoke the Microsoft OAuth permission entirely, remove Vibe Sensei from My Account → Apps & services → Apps and services with account access in your Microsoft account settings.

To request deletion of any data associated with your Teams workspace or Microsoft account, or for any support question about the Teams integration, email support@vibesensei.ai. We respond within one business day and complete deletion requests within 30 days.