Video Search
How to Search Inside Videos (Not Just Their Titles)
Searching a video usually means searching its filename or description. Searching inside a video means finding the exact moment something happens. Here's the difference and how it works.
Two different things people call "video search"
Most video search is metadata search: a filename, a title, a description, a tag someone typed in by hand. It finds the file. It can't tell you where in that file something happens, because nothing about the file's metadata describes its contents frame by frame.
Searching inside a video means the system has actually looked at the footage — every frame, every line of dialogue, every on-screen action — and indexed what's there, so a query like "the part where he explains the pricing" returns a timestamp, not just a file.
Why this matters more as footage volume grows
A single interview is easy to scrub through by hand. Eight hours of raw multi-camera footage isn't. Once a library grows past what one person can hold in their head, metadata search stops being enough — you need the system to know what's inside the file, not just what it's called.
This is the same capability that makes agentic editing possible: an AI can't build a story from raw footage unless it can first find the moments that belong in it. Search-inside-video is the retrieval step underneath the editorial one.
How Broll approaches it
Broll indexes every frame of your raw footage on import — objects, actions, dialogue, on-screen text — so you can search it the way you'd search a document, and the agent can find the moments it needs to build your edit without anyone logging footage by hand first.