Why Video Search Is Broken (And How We're Fixing It)
You can search the entire internet in milliseconds, but finding a single moment in your own footage still means scrubbing through hours of timeline. Here is why, and what FrameQuery does differently.
You can search the entire internet in 200 milliseconds. You can find any email you've ever sent. You can locate a single line inside a million-file codebase.
But finding the exact moment someone said "let's go with take three" in yesterday's shoot? That still means scrubbing. Manually. Through hours of footage.
The problem is metadata, not video
Video files are opaque. A 40 GB R3D file and a 200 MB MP4 look the same to your operating system: a filename, a duration, a thumbnail. Nothing about what's inside.
Text is searchable by nature. Video is not. That gap has only grown wider as storage got cheaper and shooting ratios ballooned. The average production now generates 20-50x more footage than it uses. Finding the right clip is the bottleneck.
Current solutions fall short
Folder structures and naming conventions break down the moment more than one person is involved. Was it in B-Roll/Day2 or Selects/Interview_Final_v2?
NLE markers and bins help during the edit, but they are locked inside your project file. You cannot search across projects.
Manual logging works if you have the budget for an assistant editor spending hours on it. Most teams do not.
What if your footage indexed itself?
That is the core idea behind FrameQuery. When you process a video, we build a local index that captures:
- Transcription: every word, with timestamps
- Object detection: what appears on screen, frame by frame
- Face recognition: who is in each shot
- Scene descriptions: AI-generated summaries of what is happening
The index lives on your machine. Searching it is instant and free. You type a natural-language query like "Lena talking about the product reveal" and get timestamped results across your entire library.
Local-first by design
We deliberately chose a local-first architecture. Your index is yours. It works offline. If FrameQuery disappeared tomorrow, your indexes would still work.
Processing happens in the cloud (that is what you pay for), but the result, the searchable index, is saved locally. Search is always free.
What comes next
We are currently in early access. If you work with video and want to stop scrubbing, join the waitlist for priority access and locked-in pricing.