Five users at a time was the whole problem
The Nashville Rescue Mission's old camera server let five people log in at once. Its IT manager on why the shelter switched, and why the deciding features were price and thumbnails.
Five people at a time could log into the Nashville Rescue Mission's cameras. That was the hard ceiling on the on-premises server the shelter ran until it switched systems. The mission is a non-profit that serves people who are hungry or experiencing homelessness, and like most shelters it has a lot of cameras and a lot of staff.
According to Spot AI's published account, that limit was the mission's main problem. Five concurrent logins across a facility with many staff meant that access to the cameras was unreliable, and whoever arrived sixth waited.
Unlimited logins
The replacement, in the mission's telling, removed the ceiling. Unlimited users, sign-on through the mission's existing SSO provider, remote access, and several people watching the same cameras at once.
Ben Grady, the mission's manager of IT and database administration, describes the result in terms of sharing.
Probably the biggest benefit of Spot is the ease with which we can look up incidents and see footage and then be able to share it and collaborate. It really has reduced the amount of time that employees have to spend looking through the footage.
An incident in a dormitory or a dining hall involves the staff who were there, their supervisor, security, and sometimes case managers. A camera system that admits five people is a system most of them never use.
Thumbnails
The second problem was search. Finding a specific incident meant scrolling through footage, and the account says this took a long time.
Grady's explanation of why the mission chose the system does not mention the AI at all.
The two things that caused us to go with Spot over the other ones that we looked at was the price, which was fair, and the use of thumbnails while searching to be able to quickly go through lots of material.
Thumbnails are the strip of still frames that lets a person scan hours of video by eye, without scrubbing. The account also credits AI filters for people and vehicles, which narrow a search before the scanning starts. But the feature Grady names is the visual one.
What a non-profit buys
Two things stand out in this account, and neither is a headline number. Spot AI publishes no dollar savings or time figures for the mission.
The first is that the decisive constraint was concurrent users, which is a licensing decision. Nothing about the old server's hardware stopped a sixth person from logging in; the licence did.
The second is that a non-profit IT manager, asked why he chose a video AI system, named price and thumbnails. The AI filters run underneath and narrow a search to people or vehicles before anyone starts scanning. If you asked your own IT manager why they picked your camera system, what would they name?