My goal is to turn streams of user-interaction events from our Flutter mobile app into clear behavioral insights. The system you create will sit between the app and our backend, funneling tap-, scroll- and navigation data to an OpenAI-powered model that automatically analyzes patterns and flags notable behaviors in near real-time. Key tasks • Wire Flutter event logging through a secure API layer • Build or adapt an OpenAI prompt/embedding workflow that interprets the raw data and returns concise JSON summaries of user intent, friction points and engagement drivers • Expose an endpoint the app can call for on-device guidance messages based on the model’s output • Provide unit tests plus a short technical note on fine-tuning or prompt-engineering choices so we can iterate later Acceptance is straightforward: the demo app must stream interaction data, receive an analyzed JSON response from the OpenAI integration within two seconds, and display at least one context-aware suggestion per session. Feel free to reuse proven analytics or state-management packages as long as they mesh cleanly with Flutter 3.x. I’ll supply API keys and a minimal dataset once the structure is in place.