Integrations / n8n
n8n AI voice agents that call your workflows mid-conversation
A webhook tool posts the caller's details to an n8n Webhook node, and the workflow's response becomes the agent's answer.
Workflow agent
Sends caller requests to an n8n workflow.
Instructions
Collect name, phone and request, run the workflow, and read the result back plainly. If it returns an error, apologize and offer a callback. Transfer to the office line if the caller asks for a person.
- Handoff
- Office
Connected tools
- Run workflowWebhook
- Transfer to officeTransfer
What a n8n AI voice agent can do
Open one to watch a call. The grey lines are the tools the agent calls mid-sentence.
A self-hosted n8n keeps data on your side. The workflow's URL only needs to be reachable over public HTTPS.
Set the Webhook node to respond with the last node's data, and the agent reads that result back to the caller.
If the workflow returns an error, the agent says it can't reach the system and offers a callback, instead of guessing an answer.
Tool definitions
Webhook tools go to POST /v1/agents/tools. Transfers are a system tool on the agent. Replace the placeholder values in capitals with your own.
Run workflow
Posts the request to a workflow webhook.
Show the JSON definition
Transfer to office
Cold transfer to your office.
Show the JSON definition
Set it up
Through n8n's Webhook node. The agent has no n8n-specific node. A webhook tool posts to your workflow's production URL. Vendor details read in October 2026.
Add a Webhook node
Set HTTP Method to POST and use the production URL, not the test URL, once you go live.
Require a header
Set the Webhook node's Authentication to Header Auth with the same header name and secret the tool sends.
Respond with data
Set Respond to When Last Node Finishes, which returns the last node's data.
Add the tool and test
Paste the URL, add the header secret, and call the agent.
Before you build
Private networks are blocked
Cartesia blocks webhook calls to private and internal address ranges, so a workflow on your LAN needs a public HTTPS entry point.
Use the production URL
n8n registers the test URL only while you listen for a test event in the editor. Publish the workflow and use the production URL.
Use cases for teams on n8n
More integrations
Enterprise-grade security
HIPAA compliant

SOC 2 Type 2

GDPR
n8n AI voice agent FAQs
How do I build an n8n voice agent?
Pair a Cartesia voice agent with an n8n Webhook node. The agent answers the call and uses a webhook tool to post the caller's details to your workflow, and the workflow's response becomes what the agent says next. n8n has no Cartesia node, so the webhook is the whole connection.
Does it work with self-hosted n8n?
Yes, if the webhook URL is reachable over public HTTPS. Cartesia blocks webhook calls to private and internal address ranges, so a workflow on your LAN needs a public entry point.
Why does the tool fail with my n8n webhook URL?
It is often the test URL. n8n registers the test URL only while you listen for a test event in the editor. Publish the workflow, point the tool at the production URL, and use the POST method.
How do I secure the n8n webhook?
Send a shared secret header from the Cartesia tool, and set the Webhook node's authentication to Header Auth with the same name and value. n8n then rejects calls to the URL that don't carry the header.
Who bills what on an n8n setup?
Cartesia bills the call, and n8n runs on n8n Cloud or your own server. Agents run at $0.03/min, plus LLM usage and telephony. A self-hosted n8n adds only the cost of the server it runs on.
Get started today
Talk to an expert.
Connect with a member of our team and learn how Cartesia can help you build world-class voice experiences.
Start building.
Access our models via API and bring a voice agent into production in minutes.