Sources

Drop in almost any recording.

Paste a public link or point it at a local file. Lirovo takes it from there: it fetches the recording, sorts out the formats, and turns it into one run you drive from your terminal. This is the first step in the pipeline, the ingest.

What you can feed it

Links from the places people actually record, plus your own local files. If it is a public link or a file on your machine, it is fair game.

YouTube

Paste any public YouTube link. Lirovo fetches the video for you, no download dance.

Zoom

Point it at a shared Zoom recording link and it pulls the session in.

Microsoft Teams

Share a Teams meeting recording link and turn the call into structured data.

Loom

Drop a Loom share link to capture a walkthrough or async update.

MP4 and video files

Run it on MP4 and other common video formats straight from your machine.

Audio files

No video track? Lirovo runs an audio-first path and works from the transcript.

Webinars

Recorded webinars and other long-form sessions are handled end to end.

Any direct URL

Have a direct link to a video or audio file? Point it at the URL and Lirovo takes it from there.

How ingest works

01step

Paste a link or point at a file

Start a run two ways. Paste a public link, or point it at a video or audio file on your machine. You also choose which of your own models to use for the run: your Claude, ChatGPT, or Gemini plan, or your own API keys.

02step

Lirovo fetches and normalizes it

Lirovo downloads the source, sorts out the formats and codecs, and handles long runtimes. You do not babysit it. When there is no video track, it switches to an audio-first path automatically.

03step

You get a run you drive from your terminal

Every source becomes one run with an id. Watch progress and pull results right in your terminal, or drive it from your agent. Lirovo runs standalone, or as a plugin inside the agent harness you already use.

It runs on your machine

Lirovo runs locally and only fetches the link you give it. Your recordings, your transcripts, and the results stay on your machine. You bring your own inference, so the model calls go to the provider you choose with your own plan or keys.

Nothing is sent to us to process, and there is no managed account to set up. For the full picture of how this stays on your side, see our Security page.

Point it at your first recording

Run a link or a local file through the whole pipeline and watch it become structured data, right in your terminal or from your agent.