Voice-driven AI assistant using VAPI and GPT-4.1-mini with memory

Send VAPI voice requests into n8n with memory and OpenAI for conversational automation

This template shows how to capture voice interactions from VAPI (Voice AI Platform), send them into n8n via a webhook, process them with OpenAI, and maintain context with memory. The result is a conversational AI agent that responds back to VAPI with short, business-focused answers.

✅ What this template does

• Listens for POST requests from VAPI containing the session ID and user query
• Extracts session ID and query for consistent conversation context
• Uses OpenAI (GPT-4.1-mini) to generate conversational replies
• Adds Memory Buffer Window so each VAPI session maintains history
• Returns results to VAPI in the correct JSON response format

👤 Who’s it for

• Developers and consultants building voice-driven assistants
• Businesses wanting to connect VAPI calls into automation workflows
• Anyone who needs a scalable voice → AI → automation pipeline

⚙️ How it works

1. Webhook node catches incoming VAPI requests
2. Set node extracts session_id and user_query from the request body
3. OpenAI Agent generates short, conversational replies with your business context
4. Memory node keeps conversation history across turns
5. Respond to Webhook sends results back to VAPI in the required JSON schema

🔧 Setup instructions

Step 1: Create Function Tool in VAPI

1. In your VAPI dashboard, create a new Function Tool
   • Name: send_to_n8n
   • Description: Send user query and session data to n8n workflow
   • Parameters:
     • session_id (string, required) – Unique session identifier
     • user_query (string, required) – The user’s question
   • Server URL: https://your-n8n-instance.com/webhook/vapi-endpoint

Step 2: Configure Webhook in n8n

1. Add a Webhook node
2. Set HTTP method to POST
3. Path: /webhook/vapi-endpoint
4. Save, activate, and copy the webhook URL
5. Use this URL in your VAPI Function Tool configuration

Step 3: Create VAPI Assistant

1. In VAPI, create a new Assistant
2. Add the send_to_n8n Function Tool
3. Configure the assistant to call this tool on user requests
4. Test by making a voice query — you should see n8n respond

📦 Requirements

• An OpenAI API key stored in n8n credentials
• A VAPI account with access to Function Tools
• A self-hosted or cloud n8n instance with webhook access

🎛 Customization

• Update the system prompt in the OpenAI Agent node to reflect your brand’s voice
• Swap GPT-4.1-mini for another OpenAI model if you need longer or cheaper responses
• Extend the workflow by connecting to CRMs, Slack, or databases

📬 Contact

Need help customizing this (e.g., filtering by campaign, connecting to CRMs, or formatting reports)?

• 📧 [email protected]
• 🔗 https://www.linkedin.com/in/robert-breen-29429625/
• 🌐 https://ynteractive.com