Case study · AI product · Live on Google Play

Outspokn

AI English learning app — practice that adapts to how each student actually learns.

React NativePythonDjangoOpen LLM modelsRAG / retrievalAdaptive learning tracks

Problem: Generic English courses treat every learner the same. Students lose motivation when lessons ignore their level, weak spots and practice history — and most apps still ship static content catalogs instead of adaptive paths.
Solution: Outspokn is a React Native English learning app backed by Django. Open LLM models power conversation and feedback. A tuned RAG layer retrieves the right lesson and practice material from the course corpus, then builds and adjusts courses against each student’s learning tracks — so practice stays relevant as they improve.
Technology: React Native · Python · Django · Open LLM models · RAG / retrieval for course data · Learning-track driven personalization
Challenges: Keeping RAG answers grounded in course content (not generic chat), shaping retrieval so practice matches learning tracks, and delivering a mobile UX that feels like a teaching product — not an LLM demo wrapped in screens.
Result: Live on Google Play as Outspokn: English Speaking App — AI conversation practice, learning tracks and personalised courses for real spoken-English fluency.

App screens