Super Siri | Ask Sammy Voice AI Assistant

Super Siri | Ask Sammy Voice AI Assistant Workflow Overview This workflow creates a sophisticated, voice-enabled AI assistant named "Sammy." It functions as an API endpoint that can be integrated with voice assistants like Siri (via Shortcuts) or other applications. When a user asks a question, the workflow intelligently decides if it needs to access a specialized knowledge base about "projects" stored in a Supabase database. It then uses this external knowledge to formulate an accurate, context-aware answer, preventing AI "hallucinations" and ensuring responses are grounded in factual data. Technologies & Automation Technologies Used - n8n: The core automation platform hosting and executing the workflow. - Webhook: Serves as the API endpoint to receive user queries. - LangChain: The underlying framework used by n8n's AI nodes to create an intelligent agent capable of reasoning and using tools. - OpenAI: Provides the Large Language Models (LLMs) for both general reasoning and for synthesizing answers from retrieved data. - Supabase: Acts as a vector database, storing the specialized knowledge base (in this case, from YouTube transcripts) for efficient retrieval. What It Automates This workflow automates the process of answering complex user questions that require access to a specific, private knowledge base. Instead of a human manually searching through project documents or transcripts, "Sammy" can instantly retrieve the relevant information and provide a natural language answer. This provides an on-demand, 24/7 expert on the data it has been trained on, significantly improving efficiency and information accessibility. AI Integration: Retrieval-Augmented Generation (RAG) This workflow is a powerful example of Retrieval-Augmented Generation (RAG). The AI is not just a conversationalist; it's an agent that can use tools to find information it doesn't already know. Here’s how the AI is integrated: 1. AI as a Router: The primary AI Agent receives the user's query. It uses an OpenAI model to understand the user's _intent_. 2. Tool-Use Decision: Based on the query, the agent decides if it needs to use its available tools.