See llms.txt for all machine-readable content.

Back to Templates

Manage Zendesk tickets and contacts via Placetel AI Voice Agent calls

Created by

Created by: Placetel || placetel
Placetel

Last update

Last update 2 days ago

Categories

Share


Quick overview

This webhook-based workflow lets a Placetel AI Voice Agent look up callers and manage Zendesk users and tickets by sending an operation in each request, returning a speech-ready response for the agent to read back.

How it works

  1. Receives a POST webhook request from Placetel with an operation and caller context details.
  2. Parses the request, extracts the requested operation, and normalizes the caller phone number to a German +49 format for consistent Zendesk matching.
  3. Routes the request by operation to the corresponding Zendesk action (search or create a user, create or update tickets, list open tickets, or update a user).
  4. For caller identification operations, searches Zendesk users by the normalized phone number and returns a speech response indicating whether a matching customer is found and/or verified.
  5. For ticket operations, creates callback, call-log, or regular tickets in Zendesk, or fetches/updates an existing ticket for status, internal notes, or escalation.
  6. Returns a JSON webhook response that includes a concise speech string and key IDs (such as customer_id or ticket_id) for the voice agent to continue the conversation.

Setup

  1. Add a Zendesk OAuth2 credential in n8n and select it in each Zendesk node.
  2. Activate the workflow and copy the webhook URL for the POST /placetel-zendesk endpoint.
  3. In Placetel, configure your AI Voice Agent (custom tool) to POST to the webhook URL and to always send an operation plus phone from %%caller_number%%.
  4. Ensure your Zendesk instance supports user search by phone and that your user phone numbers are stored consistently (the workflow normalizes to +49 / 49 formats).
  5. Review and adjust the identity verification logic to match your Zendesk user fields and what you consider acceptable verify_value inputs before using it in production.