Video2Tutorial — a recording that becomes a written procedure
Screen recordings are how knowledge actually moves inside a company — and how it gets lost. This is the tool I built to turn a video into a structured document: timestamped steps, the text visible on screen, a transcript excerpt per step, and a GIF for the one gesture that is hard to describe.
Five steps, no magic
Import
MP4 / MOV / MKV / WebM up to 2 GB, or a public YouTube URL (720p maximum, 90 minutes).
Capture
Frames sampled at intervals, each one timestamped and linked to the nearest subtitle line in French or English.
Read
OCR on the captured frames so the text visible on screen lands in the document instead of being retyped.
Structure
Five fields per step: context, detailed action, expected result, verification, command and notes.
Export
A GIF of the gesture and a Markdown procedure that can be reviewed, versioned and handed over.
- No generative model. It retrieves subtitles, reads the screen and pre-fills a framework — you validate and complete it.
- Captures are sampled at regular intervals: they are not semantic detection of every click.
- YouTube subtitles may be absent, disabled or in another language. Private or restricted videos are not supported, and nothing tries to get around that.
- Use it only on content you are entitled to process.
It is the same discipline I apply to a release: take an unstructured flow, make each step explicit, name the expected result, then prove it happened. A cutover runbook and a tutorial are the same artefact with a different domain.
The tool runs on its own subdomain, behind a login, and is not part of the public site.
Try it on a recording you already have
Upload a video or paste a YouTube link, review the generated steps, export the GIF and the Markdown procedure.