Intra AI works inside the systems you already have
Every member of staff connects their own account. An agent works with those credentials and sees exactly what that person is already allowed to see – never through a shared collective login.
Available today
Microsoft 365
Mailbox, calendar, Teams and SharePoint – more than 90 individual actions, reading and writing: create a draft, reply to a message, add a calendar entry, file a document in SharePoint.
Google Workspace
Gmail, Calendar, Drive, Docs and Sheets in their core functions: send email, create calendar entries, search files, read documents and spreadsheet ranges.
GitHub
29 actions across repositories, issues and pull requests – from code search to a comment in the right place.
GitLab
30 actions across repositories, issues and merge requests. Your administrators set the base URL, so the integration works against gitlab.com and against your own self-hosted instance alike.
FILERO (document management)
Query files, contracts and receipts in natural language, using each user’s own FILERO credentials and permissions. The connector is in beta: it is running in production with our first customers and law firms, and its scope is still growing.
In preparation
These connectors are being built. Tell us through the form below which one you need first — that is what decides the order.
Asana
Confluence
HubSpot
Jira
monday.com
Notion
Slack
All product names on this page are trademarks of their respective owners. Naming them says that Intra AI can connect to that system; it implies no partnership, certification or endorsement by the companies named.
Your system is not on the list? Then we build the integration.
We are not limited to a fixed catalogue. If your system has a documented interface – inventory and order management, time tracking, industry-specific software – we build a dedicated integration for it. It ships to you alone, runs only on your instance, and your instance does not have to be rebuilt for it. For straightforward interfaces we are talking weeks, not quarters.
Review the interface
We read your system’s documentation and agree with you which actions the AI should be able to perform – and which it expressly should not.
Build the integration
Every action is defined individually, with a validated input schema and a clear description. Nothing is passed through generically.
Check it automatically
Every integration goes through the same automated checks as our standard ones: type checking, conformance tests and security scans, before it ships.
Ship it to you only
The finished integration goes into the image of your instance and nowhere else. No other customer receives it, and the credentials stay encrypted per user on your side.
Tell us the system it all hangs on
Bring the name of the software and, if you have it, the interface documentation. In the intro call we will tell you whether an integration is feasible and roughly what it involves.
Request an integration
We will look at the interface and come back to you with an assessment of whether and how your system can be connected.
Pick a slot directly
By video or in person in Munich – you pick the slot, we bring the questions.