AI Voice Chat Agent with ElevenLabs and InfraNodus Graph RAG Knowledge

Name: AI Voice Chat Agent with ElevenLabs and InfraNodus Graph RAG Knowledge
Rating: 5 (5 reviews)
Author: Free N8N

Intermediate

20 nodes connected

detail.loadingPreview

Free N8N Temples

340 views

45 downloads

AI & Machine LearningAIAgentChatbotElevenLabsInfraNodusKnowledge GraphRAGVoice

An n8n workflow that creates an AI voice chat agent using ElevenLabs for voice and an AI agent for reasoning, leveraging InfraNodus knowledge graphs as RAG (Retrieval Augmented Generation) experts.

About This Workflow

This workflow orchestrates an AI-powered voice chat agent. It integrates ElevenLabs for voice interaction and an n8n AI Agent node for intelligent decision-making. The core of its knowledge base relies on InfraNodus knowledge graphs, acting as specialized 'experts' that the AI agent can query using RAG principles. The workflow is designed to handle user queries, determine the most relevant expert (InfraNodus graph), retrieve information, synthesize it, and respond to the user, all while maintaining conversation history.

Key Features

Voice Interaction: Utilizes ElevenLabs for converting text to speech and potentially speech to text (though the JSON focuses on text output).
AI Agent Orchestration: Employs an n8n AI Agent node to intelligently select the best InfraNodus expert(s) for a given query.
InfraNodus Knowledge Graphs as Experts: Integrates multiple InfraNodus knowledge graphs (books) as specialized data sources, each with specific domain knowledge.
RAG Implementation: The AI Agent crafts prompts to query InfraNodus graphs, simulating Retrieval Augmented Generation to fetch relevant information.
Conversation Memory: Implements a Simple Memory node to maintain context and allow for follow-up questions.
Webhook Trigger: Designed to be triggered by external requests, likely from an ElevenLabs AI Conversational agent, via a webhook.
Structured Response: The workflow aims to provide a synthesized and conversational response back to the user.
Extensible: Allows for the addition of more InfraNodus experts.

How To Use

Set up ElevenLabs Agent: Follow the detailed tutorial provided in 'Sticky Note' node 0ac0d3d2-75b9-4173-a45a-56a0fe55892d to create a Conversational AI agent in ElevenLabs, linking it to the n8n workflow's webhook.
Configure InfraNodus Credentials: Ensure your InfraNodus API key is correctly set up in the n8n credentials for the httpRequestTool nodes.
Define Experts: Each httpRequestTool node represents an expert (InfraNodus book). Configure the toolDescription parameter within each node to accurately reflect the knowledge domain of the book. The name parameter in bodyParameters must match the specific graph name in InfraNodus (e.g., waves_into_patterns).
Configure AI Agent System Prompt: In the AI Agent node (e7ffd4c6-d725-4ac0-8aa9-09f52d9a1cb4), the systemMessage should clearly define the AI's role and instruct it to use the available tools (experts).
Configure LLM: Choose between Google Gemini Chat Model or OpenAI Model based on speed and precision needs. Ensure the credentials are set up.
Configure Chat Memory: The Simple Memory node (2e049f32-5d79-4a46-8c99-6a5a9c583c4f) is crucial for maintaining conversation context. Ensure sessionKey is correctly mapped.
Webhook Trigger: The Webhook node (56f27d69-5091-4a0e-abc9-49fea9a863b3) is the entry point. It receives requests, likely from ElevenLabs.
Respond to Webhook: The Respond to Webhook node (8b7a6506-4220-476e-8d94-fa62315ab807) sends the final synthesized response back.
Activate Workflow: Ensure the workflow is active to receive incoming requests.
Test: Initiate a conversation via your ElevenLabs agent to test the workflow.

Apps Used

Agent

Chatbot

ElevenLabs

InfraNodus

Knowledge Graph

RAG

Voice

Workflow JSON

{
  "id": "dc1da027-6854-4ab5-aeae-57f608c04555",
  "name": "AI Voice Chat Agent with ElevenLabs and InfraNodus Graph RAG Knowledge",
  "nodes": 20,
  "category": "AI & Machine Learning",
  "status": "active",
  "version": "1.0.0"
}

Note: This is a sample preview. The full workflow JSON contains node configurations, credentials placeholders, and execution logic.