Voice Assistant with Persistent Memory

I built a working voice assistant prototype that demonstrated conversation memory persistence across phone calls. The system integrated Twilio for phone connectivity, ElevenLabs for AI voice processing, and a custom Flask backend that stored conversation history in file-based storage.

Implemented a webhook architecture where the AI queried the backend at call start to retrieve previous conversations and saved new conversations at call end, enabling the assistant to remember context across multiple interactions.

Through building this MVP, I gained practical understanding of conversational AI architecture, webhook-based integrations, and the technical constraints that influence AI product decisions.