Retell AI Phone Assistant

Retell AI Phone Assistant - Workflow Overview What This Automation Accomplishes This workflow builds a sophisticated, AI-powered phone assistant capable of handling real-world customer interactions. It automates two key business functions: providing information and booking appointments, effectively creating a 24/7 virtual receptionist. The system listens for incoming webhooks from the Retell AI voice platform. When a call ends, it automatically processes the conversation, summarizes it, extracts key contact details, and sends a notification to a Telegram chat. During a live call, the Retell agent can trigger other parts of this workflow to perform specific tasks, such as looking up information in a company knowledge base (RAG) or scheduling an event directly in a Google Calendar. This provides a smooth, intelligent, and immediate response to caller needs without human intervention. Technologies Used - n8n: The central automation and orchestration platform connecting all services. - Retell AI: A service for building and deploying conversational AI phone agents. - OpenAI: Provides the Large Language Models (LLMs) for understanding, summarization, data extraction, and response generation. - Qdrant: A vector database used to store company knowledge for the RAG system. - Google Drive: Serves as the source for documents that form the company's knowledge base. - Google Calendar: Used for automated appointment and event scheduling. - Telegram: Functions as a notification channel for sending post-call summaries. How AI is Integrated AI is the core engine of this entire workflow, enabling complex conversational tasks that would otherwise be impossible to automate. 1. Conversational Intelligence: The primary AI, the Retell AI Agent, handles real-time voice conversations, transcribing speech and understanding user intent. 2. Post-Call Analysis & Summarization: After a call, an OpenAI Chat Model (gpt-4o-mini) analyzes the full transcript. It generates a concise summary, identifies key points, and uses a Structured Output Parser to precisely extract entities like names, emails, and phone numbers. This turns unstructured conversation into structured, actionable data.