You send a clip
You send us video from your camera — a whole day, an hour, or just the part you care about. We only look at footage you hand us. We never point this at video you didn't send.
You tell it what to find
You ask a plain question, like "When did anyone walk in the back door?" or "When were the boxes moved?" You don't need special words. You ask it the way you'd ask a person.
The AI watches every frame
A video AI model (Google's Gemini) looks at the footage many times per second — every part of it, start to finish. It never gets bored and never skips ahead.
You get a timestamped list
It writes down each moment that matches your question and the exact time it happened. You read the list, click a time, and jump straight to that spot in the video to check it yourself.
Why we show the cost and the time
On the demo page we put the real cost and the real time it took, right on the screen. Most tools hide that. We show it because it's the whole point: reviewing footage this way is fast and it costs very little.
The warehouse clip on the demo took about 27 seconds and cost about $0.006 in AI tokens, using the gemini-3.1-flash-lite model. That's the real run, not an estimate.