In the world of AI-powered conversations, real-time integrations can dramatically enhance customer experience and workflow automation. With Voice AI’s Custom Actions, you can trigger webhook calls to external systems, fetch live data, and update third-party platforms—all during an active call.
This guide will walk you through the step-by-step process of creating a Custom Action for Voice AI using webhook integrations, dynamic parameters, and real-time testing.
What Is a Custom Action? #
A Custom Action in Voice AI allows you to trigger an external Webhook call during a conversation. These calls can be configured with authentication, custom headers, and dynamic values captured from the conversation in real time.
Custom Actions enable seamless integration with external systems, such as order management platforms, scheduling tools, and more, directly from the AI call flow.
Why Use Custom Actions? #
- Live Data Fetching: Fetch customer-specific data (e.g., contact records, appointments) while the call is ongoing.
- Enhanced Automation: Automate repetitive tasks, such as lookups or order verifications.
- Zapier & API Integration: Trigger Zapier automations or push updates to any system with an open API.
Creating a Custom Action #
Before you can create a custom action, you must first determine when and why your AI agent should trigger an external process. Common examples include:
- Checking a customer’s order status
- Verifying appointment details
- Updating a contact record
- Triggering a downstream Zapier workflow
Set Up a Conversation Trigger #
After identifying your use case, head to the Voice AI Agents settings and create a new agent or open an agent from the list.
You can now set up your action to occur during or after the call by clicking the “+ New Action” button in the relevant window.
From the available items, click the Custom Action tool.
A new window will open, where you can assign a name to the action, then define the trigger condition. This could be a keyword, phrase, or structured input detected during a call. When the condition is met, the Custom Action will execute automatically.
Once these are added, enter the phrase you want the agent to say before executing the action.
Configure the Webhook Request #
Next, choose the relevant method (POST or GET), then paste the destination API endpoint you want to interact with.
Add Authentication & Headers #
Set up required headers (e.g., Content-Type: application/json, Authorization: Bearer [token]) and include API keys or tokens for secure access, if needed.
Map Dynamic Parameters #
Voice AI can collect and pass live data from the conversation to your webhook call. These dynamic parameters include:
- Customer name, email, or phone number
- Order numbers or service IDs
- Numeric values or custom phrases
Use these parameters in your request body to personalize and automate the API interaction based on the caller’s input.
Test the Webhook in Real Time #
Use the testing interface to verify your webhook works as expected. This involves simulating the conversation trigger, reviewing request/response details, and ensuring the external system is receiving and processing the data correctly.
This step is critical to prevent errors once the automation goes live.
Save and Activate Your Custom Action #
After successful testing, save the custom action, link it to the appropriate voice flow or scenario, and monitor its performance during live calls using activity logs or external system responses.
Whether you’re triggering updates to external platforms, fetching live data, or launching workflows, Custom Actions ensure your AI conversations are intelligent, contextual, and action-driven.
Frequently Asked Questions #
What is a Custom Action in Voice AI? #
It is a configurable webhook call that the agent can trigger during a conversation to integrate with external systems, fetch live data, or update third-party platforms in real time.
When does a Custom Action run during a call? #
It runs automatically when the trigger condition you define, such as a keyword, phrase, or structured input, is detected during the conversation.
What HTTP methods can a Custom Action use? #
You can choose POST or GET, then provide the destination API endpoint you want to interact with.
Can I secure the webhook with authentication? #
Yes. You can set required headers such as Content-Type and Authorization, and include API keys or tokens for secure access.
What live data can be passed to the webhook? #
Dynamic parameters such as customer name, email, or phone number, order or service IDs, and numeric values or custom phrases can be mapped into the request.
Should I test before going live? #
Yes. Use the testing interface to simulate the trigger and review the request and response, ensuring the external system processes the data correctly before activating.
