AI secretary for whoever runs tutoring at a school
PrototypeClient: School (private prototype)Sector: Education
A private prototype that organises sensitive student information and drafts documents to review. It uses synthetic data, and a person makes the final decision.
The problem
Whoever runs tutoring prepares family meetings and evaluation sessions from information spread across spreadsheets, exports, screenshots and notes. It has to be redone every evaluation, with careful tone and strict privacy. A generic AI tool does not work here, because it means pasting exactly the data that must not leave.
What was built
- Import of already-structured information, validated before anything is saved.
- Its own database with a record per student, evaluation period and context memory.
- A dashboard to read what is stored without searching five files.
- Draft reports generated with fixed rules and a repeatable structure.
- A privacy filter: internal notes never reach the text a family sees.
- Everything is a draft: the final decision stays with the person.
What you can see
A demo with fully synthetic data and recreated screenshots; no real data. There is a time-saving estimate from the workflow analysis, but it has not been measured, so it is not published as a result.
Shown with synthetic data, never with real client data.
What it shows
AI on sensitive data, with privacy and human review planned from the start.