AI Chatbot Integration — Large-Scale Retail App

Context

A large-scale retail app with millions of active users, built on Flutter and BLoC.

Challenge

Integrate conversational AI into an existing, high-traffic Flutter app without disrupting the production architecture or introducing regressions for existing users.

My role

Associate Tech Lead, external consultant.

Technical approach

• Designed a platform channel architecture to bridge native Android/iOS AI SDK capabilities into the Flutter layer
• Built session and authentication management for the assistant on top of the app's existing auth layer
• Implemented real-time chat state with BLoC, kept isolated from unrelated app state
• Instrumented the feature with Sentry monitoring before and after launch

Outcome

Successfully shipped to production across iOS and Android with no architectural regressions.

What I learned

The hardest part of AI integration in a mature app isn't the model — it's making the new feature's state management invisible to everything that already worked. Isolating the AI feature behind a clean BLoC boundary meant the rest of the app didn't need to know it existed.