The pitch was simple: Google Analytics, but for your parking lot. Every car gets an entry and exit record, and the operator sees how the lot is actually used.
I wrote the software myself, start to finish, as a small venture called Lot Insights.
A camera catches the car coming in. The software finds the plate in the frame, reads it, and writes a row: plate, lot, time in. When the same plate passes the exit camera, the row gets its time out.
The dashboard shows cars in the lot right now, dwell time as a spread, peak hours, and utilization by hour and day.
Past the posted limit, a plate becomes a flagged violator with its entry time, and the alert goes to the operator. Known cars work the other way: an employee is checked in, a best customer triggers an alert.
Daylight accuracy held. Night reads weren't reliable without the right cameras, and plate placement differs by state, so the matching rule has to be per state.
It stopped because good plate-reading cameras cost too much to roll out in 2016. The software was the cheap part. Today the cameras are cheap and plate-reading APIs exist.