Your own AI brain

Your own AI brain

Cloud AI is handy, but everything you type ends up with a corporation. The onpit AI brain works with your data and runs on your hardware. What you ask stays with you.

Cloud AI is handy, but everything your team types ends up with a corporation, often including customer data. The onpit AI brain works with your data and runs on your hardware. What gets asked stays in the building.

What it does

  • Answer questions about your own documents, such as “What does the lease say about notice?”
  • Draft, summarise and translate email and text
  • Search notes, manuals and cases, even when you can't remember the exact term
  • Take over recurring tasks, together with the automations in the stack

How it is built

  • Open language models, for example from Mistral AI in Paris, run locally on the onpit.
  • Your own data as knowledge: The AI reads what is already on the onpit, such as in Nextcloud. Nothing gets uploaded for it.
  • One login: Access only through your own onpit login, with the same permissions as your files.
  • Swappable: When a better open model comes along, it gets swapped in. No subscription, no change of provider.

Two ways to your AI brain

Set it up yourself

The AI brain guide is part of the open onpit documentation. You need a running onpit and a little patience.

To the guides

Have onpit set it up

We set up the AI brain on your onpit or in your business, connect your existing data and keep models and software up to date.

Ask about setup

What you should honestly know

For some tasks, large cloud models are still stronger than anything that runs on a small device. For everyday questions, writing and searching your own files, local models do well today. Fast answers from large models need more compute, such as an extra node with a graphics card. We tell you up front what works with which hardware.

in progress The AI brain is the next component in the onpit stack. The guides will appear once it runs on our own cluster.