Before any school division deploys local AI in a meaningful way, the data question has to be answered. Not all of it at once. Just the part that matters most first.
Roughly 20% of a typical division’s data is personal and sensitive. Student records, IEPs, disciplinary files, personnel, health information. This is the data that carries FERPA obligations, the data that cannot leave the building, and the data that defines your compliance exposure. It is also the most contained, the most structured, and the most urgent. This is where you start.
The other 80% – curriculum, general administrative content, policy documents, correspondence – is largely already living in whatever Microsoft or Google ecosystem the division has been building for the last few years. That lane is not likely to change much. It will need some attention eventually, but it does not need to be prioritized.
Focusing on the 20% first accomplishes several things simultaneously. It protects the highest risk data immediately. It does not disrupt systems already in place. It gives staff a defined, manageable first project rather than an overwhelming full remediation. And once complete, it tells you exactly what local infrastructure you actually need, because you know the workload before you buy the hardware.
There is the local Library and the Big Campus Libraries (Cloud AI platforms). The Primary mission: Keep all sensitive data protected onsite in your own local Library. Simple.