We need to build a Proof of Concept (POC) for a real-time Voice AI Chatbot using the Pipecat framework that can handle live voice conversations and seamlessly hand off the call to a human agent via Amazon Connect when required. The system should support real-time speech-to-text, LLM-based response generation, and text-to-speech to create a natural conversational experience. It should also include logic for intelligent escalation (e.g., user requests human support, low confidence responses, or specific intents), which will trigger a smooth call transfer to an Amazon Connect agent without dropping the session context. Key requirements: * Implement a real-time voice chatbot using the Pipecat framework * Integrate STT, LLM, and TTS pipeline for live conversations * Design escalation logic for handoff scenarios * Integrate with Amazon Connect to transfer the call to a live agent * Maintain conversation context during the handoff * Provide a simple demo setup (local or cloud) to showcase the end-to-end flow The goal is to demonstrate a working prototype where a user can speak with the AI bot and, when needed, be seamlessly transferred to a human agent through Amazon Connect.