October 2025
The trigger
A friend's child fell ill during a school trip. The parents received simultaneously: three identical push notifications, one email in their inbox, one message in Smartschool, and no phone call. Nobody knew where the child was — every communication channel worked separately, none integrated. That's when I knew this had to change.
November 2025
Market research
Three weeks of interviews with teachers, principals and IT coordinators from primary and secondary. The pattern: everyone complains about the same things, nobody expects it can change. "Smartschool is what we have." That resignation was the market opportunity.
December 2025
Architecture plan
Three weeks of design work: data model (44 tenant-scoped tables), RLS strategy (FORCE on all), API layer (FastAPI with Problem-JSON), mobile strategy (Expo, not PWA). Decision: everything in one codebase, three native apps. No framework soup.
January 2026
First pilot tests
200 fictional students and 12 fictional classes on a staging environment. Schedule worked. Reports worked. Attendance worked. AgODi XML export worked. The parent portal was still rough — three extra weeks on that.
March 2026
Privacy audit
Self-conducted DPIA based on the AgODi template. Result: three red flags (loket export flow, under-12 messaging, evidence retention policy). All three resolved before the first pilot.
May 2026
Ready for one pilot school
44 tables with FORCE Row-Level-Security, 391+ passing tests, all four AI features production-grade (self-hosted Ollama + LibreTranslate). 56,000+ lines of code. Looking for one pilot school for the 2026-2027 academic year.