We make video as queryable as a database.
The world records more video every day than anyone can watch. The knowledge inside it, the decisions, the numbers, the names, the moments, stays locked in a format you cannot search. Lirovo opens it up.
Why we are building this
Software has had structured access to text and tabular data for decades. Video never got the same treatment. To get a single fact out of a recording, someone still has to watch it, and the result is a wall of transcript text that is barely easier to query than the video itself.
Lirovo is a single layer that sits between any video and the systems that need its data. You define a schema for the things you care about, point it at a video, and get back typed JSON that fits straight into a spreadsheet, a database, or an agent. Every value carries a link to where it came from, so you can always check the source.
It runs in your terminal, built for developers and the AI agents they work in. No new app to learn, no manual review step, just data you can use.
What we believe
Four decisions shape every part of the product. They are not features we added later. They are the foundation.
Evidence first
Every value Lirovo returns points back to the exact second in the video and whether it was spoken or shown on screen. An answer you cannot trace is an answer you cannot trust.
Model-agnostic
Lirovo runs on the inference you bring, so you use the best model for each step and are never locked to one provider. As models improve, your results improve, with no change on your side.
Local first
Lirovo runs on your own machine and your own inference. Your video and your keys stay with you, and nothing is sent to a server you do not control.
Free and open source
The CLI is free and the pipeline is open. You can read exactly how each stage works, run it on your own terms, and trust what it does because you can see it.
From a video to data, in four steps
fetch a link or take an upload
transcribe audio, analyze frames
fuse both into one knowledge graph
fill your schema, with evidence
Where we are
Lirovo is a small, founder-led team based in Montreal. The harness runs the whole pipeline on your own machine, and we sharpen it every day: more efficient skills, tighter prompt orchestration, and a faster workflow pipeline, so each run gets better without you changing a thing.
We build in the open about how the system works, and we would rather ship something correct and traceable than something flashy. If that resonates, we would love to hear from you at hello@lirovo.ai.
