Your back catalog is an asset you are not using
Old drives full of finished-job footage are dormant money, and batch AI logging is what finally makes them queryable.
Somewhere in your office is a drive with masking tape on it. Mine said "2019 MISC". Inside: four finished jobs, two dead pitches, a festival's worth of b-roll, and roughly nothing I could name without mounting it and scrubbing. Every working shooter owns several of these. We call them archives, which is generous, because an archive you cannot search is not an archive, it is a pile with a power connector.
The verdict: that pile is an asset, it is currently earning nothing, and the only thing between those two states is description.
Why unlogged archives never get touched
The economics are stacked against the pile. Any single retrieval is a maybe (maybe the shot exists, maybe it is usable, maybe it is cleared), and the cost of finding out is hours of scrubbing. So every rational person declines the search, every time. The client asking "do you have any footage of the waterfront" gets told "probably" and then nothing. The anniversary edit gets built from whatever was already on the laptop. The reshoot gets booked for material that exists three feet away. Unlogged archives do not decay because the footage gets worse. They decay because the price of asking them anything stays higher than the expected answer.
That is also why "we should organize the archive someday" never survives contact with a calendar. As a manual job it is weeks of unpaid watching, and nobody sane budgets weeks against a maybe.
What a described archive pays
Once a library answers questions, money shows up from directions you stopped considering. Licensing: a producer emails asking for storm footage from your region, and you reply the same day with timecoded options instead of a shrug. Anniversary and retrospective cuts: the ten-year edit becomes a filter across a decade, not a month of archaeology. Social cutdowns: "we need three verticals this week" turns into selects in an hour. Pitches: showing a prospect their own industry, already sitting in your archive, closes work. And the quiet one: b-roll you stop reshooting because you finally know you own it.
None of this requires new footage. It requires the footage you have to become queryable, which until recently was the expensive part.
An old drive, described in an afternoon
This is the part that changed. Batch AI logging means the first pass of watching is no longer your job. My protocol with ClipLogger: mount the drive and open the folder in place (nothing is imported, nothing moves, the index is only a rebuildable cache), then run one Prepare pass over the lot. On the free local engine the Mac grinds through it overnight and frames never leave the machine. When I want a drive done between lunch and school pickup, I use Rush, the paid cloud lane: parallel workers on larger models, flat pricing at one credit per clip (up to 15 minutes each), and the exact cost shown before anything runs. A thousand-clip drive is a $90 pack. Either lane ends the same way: transcripts you can search as text, on-screen text read off slates and signs, descriptions and people proposals staged for review.
Staged is the operative word. Nothing the model proposes becomes the record until you confirm it, so my afternoon is a fast review pass (accept, accept, reject, accept) instead of a week of viewing. The machine watches everything. I decide what is true.
Keep the description on the drive itself
One rule keeps all this durable: the description must live with the footage, not inside an application. Confirmed metadata should be written into plain files beside the media (ClipLogger's sidecars are readable JSON, one per clip), so the drive carries its own catalog. Plug it into another machine years from now and the record is right there next to the files, whatever apps exist by then. A catalog that lives only inside somebody's database, cloud or local, is a second archive waiting to be lost.
A weekend protocol for the pile
The standalone version, no particular software required. Saturday: inventory. List every drive, its label, capacity, and one line on what you believe is on it, and mount each one just long enough to screenshot the top-level folders. Pick the single drive with the highest expected value, usually the one tied to your best client or your most licensable subject, and ignore the rest of the pile. Sunday: describe that one drive as shallowly as honesty allows. Folder-level notes, a keep-or-kill pass on the obvious junk, real names on the twenty best clips. One described drive that answers questions this month beats a five-year plan for all of them.
The money for that footage was already spent, so describing it is the only step left between a box of old drives and a working asset.