Your data never has to leave your machine.
Lirovo runs locally, on your own inference. Your videos and your keys stay with you, and because the harness is open source, you can verify exactly what it does. Here is how that works.
Runs on your machine
Lirovo downloads, transcribes, reads frames, and builds the knowledge graph locally. Your video is processed on your own hardware and never uploaded to a Lirovo server.
Your inference keys stay local
Your provider logins and API keys are stored on your machine and used only to call the models you chose. Lirovo never receives them and never proxies your inference.
It only talks to where you point it
The only network calls Lirovo makes are to fetch the video you give it and to reach the model provider you selected. There is no Lirovo backend quietly collecting your data.
Open source and auditable
The pipeline is open. You can read exactly what each stage does, what leaves your machine and what does not, and confirm it for yourself instead of taking our word for it.
You own your output
Extractions, the knowledge graph, and the evidence behind every value are written to plain files on your disk. They are yours to keep, move, version, or delete.
Resumable and reproducible
Each stage caches locally and a run resumes from the last completed step, so a re-run reuses finished work instead of redoing it, and the same input gives you the same result.
It runs where you run it
There is no Lirovo cloud in the middle. The harness runs as a local process on your machine, calls out only to fetch your video and to the inference provider you chose, and writes its results to your disk. The data path is short, and it is yours.
Because you bring your own inference, your video content is governed by the data terms of the model provider you already use, not by a layer we put in between. You decide which provider sees what.
Honest about where we are
Lirovo is an early-stage, open-source project, and we would rather be precise than impressive. The model on this page is simple: the work happens on your machine, so the data we could lose, leak, or be compelled to hand over is data we never hold in the first place.
If your team needs to review how it works before adopting it, the source is the documentation. For how the website and the waitlist handle the little data they do collect, see our Privacy Policy.
Responsible disclosure
Found a vulnerability? We want to hear about it. Report it to us privately, give us a reasonable window to fix it, and avoid accessing data that is not yours. We will not pursue good-faith research that follows those rules.
