Video intelligence for product catalogs
See what's in the shot.
Write the listing from it.
Mirova turns a product video into structured catalog data: detected objects, dominant colors, shot changes, and AI-written title, description and tags — grounded in frames it actually analyzed.
No signup. Object detection runs on your device; nothing is invented.
Real on-device object detection, running in your browser. Nothing is uploaded.
How it works
Three steps. Two run on your device.
Point a camera, or upload a clip
On-device object detection runs live in the browser — nothing leaves your machine for this step.
Frames become structure
Shot changes, dominant colors and detected objects are sampled across the clip — real signals, not guesses.
Claude writes the listing
The sampled frames and detections are handed to Claude, which drafts a title, description, tags and highlight clips grounded in what it actually saw.
What Mirova does
- ✓Runs real object detection in the browser, frame by frame
- ✓Flags likely shot changes from actual pixel differences
- ✓Sends only sampled frames to Claude for language and vision reasoning
- ✓Grounds every catalog field in what was detected — no invented specs
What it never does
- ✕Invent brand names, prices, materials or certifications
- ✕Upload your raw video file to a server for this demo
- ✕Publish anything without you reviewing it first
- ✕Claim to detect something it didn't actually see
Have a product video sitting in a folder?
Turn it into catalog copy in under a minute.
Try it now →