Yenetta
An AI study companion for Ethiopian Grade 9-12 students, grounded in the actual Ministry of Education curriculum and past national exam papers, not generic internet knowledge.
Yenetta is a study companion for Ethiopian high schoolers in Grades 9 through 12. The core is a chat-first RAG tutor: every answer cites the actual curriculum or a past national exam paper, and when nothing in the corpus supports a question it says “not in the curriculum yet” rather than guessing. Everything else is built around that grounded core.
What it does
Structured study tools (summaries, notes, flashcards, and spaced-repetition review), timed past-exam practice with graded explanations, personalized study plans that adapt over time, and progress analytics with weak-area detection.
Streaks, XP, and opt-in leaderboards for motivation, plus Amharic-language support so a student can ask it to “explain in Amharic.”
Cost-controlled AI
Every model call is routed to the cheapest capable model, cached, and gated by per-tier quotas. Timed past-exam practice is pure retrieval, so it costs nothing in AI spend. That keeps the free tier viable by default and the premium tier cheap to run.
Data-residency-aware architecture
Personal data and the curriculum embeddings live in separate databases. If personal data needs to move to an in-country host, that happens without re-architecting the system.
A shared design-token package keeps the brand identical across the web app and the native mobile app.
Offline first
The app ships as an offline-first Android build with local SQLite for study in low-connectivity areas, alongside the web app. Chapa handles payment for the premium tier; the free tier is the default.
Status
All milestones M0 through M7 are complete, including the phase-2 additions: long-term memory, personalized plans, deep analytics, gamification, and Amharic i18n. No public hosted demo is on file yet, so the repository is the reference.