VAPI Chatbot

VAPI AI Journaling Chatbot Workflow Overview This workflow automates a daily voice journaling session via a phone call. It is a two-part process. First, it receives a request to call a user, then it uses the Vapi voice AI platform to conduct a structured interview based on a predefined script. Once the call is complete, Vapi sends a full report, including the transcript, back to the workflow. The workflow then uses an AI model to analyze the transcript, extract key information into a structured format, and saves the data as a clean journal entry in a Supabase database. This automation creates a hands-free, conversational journaling experience for the user and ensures the captured data is structured, clean, and ready for analysis without any manual intervention. Technologies Used - n8n: The core automation platform coordinating the entire process. - Vapi: A voice AI platform used to manage the phone call, real-time transcription, and conversational flow. - Twilio: The underlying telephony service used by Vapi to physically place the outbound phone call. - OpenAI (GPT-4o): The Large Language Model (LLM) that acts as the conversational AI during the phone call, asking questions and engaging with the user. - Deepgram: The AI service providing high-accuracy, real-time speech-to-text transcription during the call. - ElevenLabs (11labs): The text-to-speech (TTS) AI service that generates the natural-sounding voice for the AI assistant. - Anthropic: The LLM used post-call for analyzing the transcript and performing structured data extraction. - Supabase: A backend-as-a-service platform used as a PostgreSQL database to store the final, structured journal entries. AI Integration This workflow uses AI in two distinct and powerful ways: 1. Conversational AI (During the Call): The Vapi service is configured to use OpenAI's GPT-4o as the brain of the voice assistant. It's given a detailed system prompt that defines its persona as a "supportive journal keeper" and provides a strict script of questions.