Case study

Kipti

Student information platform — turning scattered teacher notes into an early-warning system for struggling pupils.

Client
Kipti
Delivered
2024 — ongoing
Engagement
12-week MVP build
Industry
Edtech / public sector
  • 0→live

    greenfield to pilot-ready product in twelve weeks

  • AI

    reads teacher notes and surfaces patterns no single teacher would spot

  • 3+

    schools onboarding in active trial across Germany

Challenge

When a student changes class or teacher, everything the previous teacher knew about them disappears. By the time someone notices a child is falling behind, months have already been lost.

German schools are remarkably under-digitised. Student information lives in paper folders, email threads, and individual teachers' memories — none of it shared, none of it searchable. When we started, the founder had a working prototype that demonstrated the concept but wasn't scalable or maintainable enough to put in front of real schools. The codebase needed a complete rethink before it could be trusted with sensitive student data.

On top of that, the product faced an unusually high compliance bar. Student data in German public schools falls under strict GDPR enforcement — meaning no third-party cloud providers, no data leaving German jurisdiction, and encryption requirements that ruled out many off-the-shelf solutions. Any shortcut here would make the product unsellable to schools entirely. Compliance wasn't a checkbox; it was a hard engineering constraint that shaped every infrastructure and integration decision.

The combined challenge was significant: build something simple enough for non-technical teachers to use daily, powerful enough to surface AI-driven insights, and locked down tightly enough to satisfy the data protection requirements of German public institutions.

Approach

Working directly with the founder, I rebuilt the platform from the ground up on a modern TypeScript stack — prioritising type safety end-to-end so the codebase could scale with the product and be handed off cleanly to future engineers. Every architectural decision was made with the constraint that teachers, not developers, would be the primary users.

  1. Rebuilding on solid foundationsReplaced the unshippable prototype with a Next.js + oRPC + Drizzle stack, giving the product end-to-end type safety and a maintainable data model that could handle real school data at volume.
  2. Teacher-first UIDesigned and built the core interface around the real teacher workflow: quick note entry after class and easy sharing them with colleagues. Adoption only works if it feels faster than a sticky note.
  3. AI summarisation and toolingIntegrated an LLM layer that reads accumulated teacher notes and produces plain-language summaries and early-warning flags. A teacher reviewing a student before parents' evening gets months of context in seconds.
  4. GDPR-compliant infrastructure from the ground upDesigned and deployed a self-hosted stack on German bare metal servers, with encryption throughout the stack and strict data residency controls. A compliant EU cloud such as AWS would have been legally viable, but full data sovereignty on German soil was a strong differentiator for schools evaluating vendors — worth the added engineering complexity.
  • TypeScript
  • Next.js
  • Hono
  • Postgres
  • Drizzle
  • oRPC
  • Tailwind
  • Kubernetes

Outcome

The platform launched to its first pilot schools in Germany and is actively onboarding teachers. For the first time, schools using the product have a continuous, structured record of each student that survives class changes and the end of the school year — and an AI layer that can read across that record to flag students who need attention before their difficulties become visible to the naked eye.

Critically, the product meets the data protection requirements that make it legally sellable to German public schools — a bar many edtech products never clear. The infrastructure is fully self-hosted on German servers with encryption throughout the stack, giving schools confidence in where their data lives and who controls it. This sovereignty work was expensive in time and engineering effort, but it is now a durable competitive advantage: schools can adopt the platform knowing student data stays on controlled German infrastructure.

The codebase is clean enough that the founder can hire and onboard additional engineers without requiring hand-holding.

Testimonial

Jonas has been with Kipti from the start and built the product with us. He doesn't just think in tasks — he thinks in the product and the company, and brings as much strategically as he does technically.

Jonas begleitet Kipti seit dem Anfang und hat das Produkt gemeinsam mit uns aufgebaut. Er denkt nicht nur in Tasks, sondern im Produkt und im Unternehmen mit – und bringt strategisch genauso viel ein wie technisch.

Björn SchriewerFounder, Kipti

Building something similar?

I take on 1–2 projects at a time. Let's talk about yours.

Get in touch →