How do custom actions work in Chatref for marketing automation?

Custom actions are tool-calling steps that extend your AI agent beyond answering questions. When a visitor asks something that requires a real-world change - like "unsubscribe me from the weekly newsletter" or "what’s my plan status?" - the chatbot securely gathers the necessary details, then calls an external API or webhook you define. The response flows back directly into the chat, so the entire interaction stays within the widget and requires zero manual intervention.

This capability transforms Chatref from a support doc answer-bot into a true AI agent that carries out tasks. You design the input fields (email, order ID, campaign name) and specify where to send the request. The agent handles the rest, using your marketing automation platform's endpoints to fetch or update data immediately.

Set up a custom action once, and it becomes a reusable step your chatbot can invoke during any conversation. Here's the typical flow for a marketing automation use case:

1. Visitor asks: "Are my campaign stats live?" or "Unsubscribe me from the sequence."
2. The AI agent detects: The intent matches a defined custom action, so it switches from retrieval mode to tool mode.
3. Smart collection: The widget asks for any missing details - email address, campaign name, etc. - and validates the format inline.
4. Secure call: With the collected data, Chatref sends a request to your marketing automation endpoint (e.g., get subscriber info, update a list).
5. Instant reply: The external tool’s response appears inside the chat, phrased in your brand’s voice by the AI.

Because everything runs asynchronously and the agent retains full conversation context, a visitor never feels bounced between help articles and your dashboard. They get the answer or the action completed right where they are.

Creating a custom action takes only a few clicks in the Chatref playground. You name the action (say, "Check email sequence status"), define the required fields (email, campaign ID), and provide the endpoint URL and authentication method. The agent builder then lets you write a natural-language prompt that describes when to trigger the action. Once saved, the action is live across all channels your widget serves.

There is no need to write custom code in your frontend. The Chatref widget calls your API securely from the server side, keeping credentials hidden and your marketing automation data safe.

Your AI chatbot can take on dozens of account-related tasks, freeing up your customer support team and giving visitors instant results:

• List management: Let users subscribe or unsubscribe from specific segments without leaving your website.
• Campaign enquiries: Allow prospects to check open rates, last-send times, or their position in a drip sequence.
• Lead capture and enrichment: When a visitor asks a sales question, collect name and email, then push the lead directly into your CRM or automation platform.
• Subscription troubleshooting: If a contact complains about not receiving emails, the agent can verify their status and trigger a resend or single opt-in update.
• Trial or plan status: For SaaS products with email-based campaigns, let users ask "What campaigns am I in?" and pull the answer from your automation tool.

Each interaction reduces the load on human agents and cuts average resolution time from hours to seconds.

How to set up custom actions?

Log into your Chatref dashboard, open the agent builder, and navigate to the “Custom Actions” tab. Click “New action,” give it a name, define the fields your action needs (e.g., email, list name), and enter the endpoint URL of your marketing automation tool along with the request method and headers. Write a short description so the AI knows when to use it. Save, and the action is available immediately in all conversations.

Can AI chatbots perform account tasks?

Yes. With Chatref’s custom actions feature, your AI chatbot can securely call external APIs, update records in your marketing automation platform, fetch account details, and more - all inside the chat window. The agent collects the necessary information from the visitor, makes the request, and presents the result in natural language.

What are the best use cases for custom actions?

For marketing automation teams, the most impactful uses are: managing email list subscriptions, checking campaign delivery statuses, pulling engagement data for a specific contact, triggering a drip campaign restart, and capturing leads directly into your CRM. Anything that previously required a support ticket or a dashboard login can be handled automatically by the chatbot, reducing response times and freeing your team for higher-value work.