How to Trigger an n8n AI Agent from a Webhook Node

n8n ships an AI Agent node and a Respond to Webhook node, which together let you build a tiny self-hosted AI endpoint. A caller POSTs a question, the agent thinks, and your workflow answers in the same request. No extra hosting beyond your n8n instance is needed.

What you need

• An n8n instance (cloud or self-hosted, version 1.x)
• Credentials for a chat model, for example OpenAI or Anthropic
• A REST client such as curl, Postman, or Insomnia

Step 1: Add a Webhook node in Respond mode

Create a new workflow and add a Webhook node. Set the HTTP Method to POST and the Path to ask. Under Respond, choose Using Respond to Webhook Node so you can return the model output later.

Step 2: Add the AI Agent node

Add an AI Agent node and connect a Chat Model sub-node to it (pick your provider and credential). In the agent's Text field, reference the incoming question with an expression so each request drives the prompt.

{{ $json.body.question }}

Step 3: Return the answer

Add a Respond to Webhook node at the end. Set Respond With to JSON and build a small body that pulls the agent output.

{
  "answer": "{{ $json.output }}"
}

Step 4: Activate and test

Click Active in the top right to switch from the test URL to the production /webhook/ask path, then call it.

Result: you now have a private AI question endpoint that lives entirely inside n8n and answers callers in real time.