What are Apps
Apps give PandaOS both a set of actions and a live workspace UI for every connected service, tied together by Spotlight.
Two halves of every app
An app in PandaOS is more than an API connection. Every app has two sides:
- Actions - the things PandaOS can do on your behalf: send an email, run a SQL query, create a calendar event, move a Trello card. These are the tools the AI uses during a conversation.
- A workspace tab - a full, live interface for that service embedded directly in PandaOS. Gmail shows your inbox with threads, labels, and compose. Supabase shows your tables, rows, and schema. Google Drive shows your files and folders. Trello shows your boards and cards.
This is what makes PandaOS apps different from a plain tool integration. You are not just giving the AI permission to call an API behind the scenes. You get a real, interactive frontend for each service inside the same window where you chat. When PandaOS drafts an email, you can watch it appear in the Gmail tab. When it runs a migration, you can see the schema change in the Database tab. The AI works with these services and you see the results in the same place.
Spotlight: the AI shows its work
Spotlight is what ties the two halves together. When PandaOS uses an app during a conversation, Spotlight automatically opens that app's workspace tab so you can watch the work happen in real time.
Ask PandaOS to check your email and the Gmail tab opens alongside the chat, showing the inbox it is reading from. Ask it to query your database and the Database tab surfaces with the table it is working with. This is not a notification or a log entry - it is the actual app interface, live, showing the same data the AI is acting on.
Spotlight is on by default for every app that has a workspace tab. You can turn it off per app if you trust the output and do not need to observe every step. The work still happens; you just will not see the tab switch. There is also a global Spotlight toggle in settings that controls whether the feature is active at all.
Built on the MCP open standard
Under the hood, every app is an MCP server - a connection that follows the Model Context Protocol, an open standard for connecting AI systems to external tools and data sources. MCP defines a common way for an AI to discover what actions are available, call them, and get results back.
This matters because MCP is not proprietary to PandaOS. Any tool built as an MCP server works with any MCP-compatible client, and vice versa. PandaOS ships with 20+ built-in apps, but you are never limited to them. If a service you use has an MCP server (or you build one), you can import it into PandaOS and it becomes a first-class app with its own icon, settings card, and actions - no vendor lock-in.
Available integrations
PandaOS covers the tools most teams already use:
- Email & Calendar - Gmail, Outlook, Google Calendar
- Messaging - Slack, Microsoft Teams
- Docs & PM - Notion, Confluence, Jira, Asana, Trello
- Dev & Cloud - GitHub, Supabase, Vercel, Databricks
- Marketing & Analytics - Google Ads, Google Analytics, Meta Ads, LinkedIn Ads, TikTok Ads
- Files - Google Drive
- CRM - HubSpot
See the Integration Reference for the full list with connection methods and capabilities.
Importing MCP servers
Any MCP-compatible server can be imported into PandaOS as an app. Once imported, it gets its own settings card and its actions appear alongside the built-in ones. This is how you connect internal tools, niche services, or anything the built-in gallery does not cover - without waiting for official support. If an MCP server exists for it, PandaOS can use it.
Unattended Runs
How automations run while the app is closed: the background service, sleep catch-up, locked credentials, permission rules and pause-for-approval
Connecting an Account
How authentication works, and how you control what PandaOS can do with each app through Actions and per-tool permissions.