How to Pass Custom Data from a Webex CC Voice Flow to an Autonomous AI Agent

One of the most common questions I get when working with Autonomous AI Agents in Webex Contact Center is this:

“How can I pass custom data from the voice flow to my Autonomous AI Agent?”

For a long time, this was a challenge because the feature was not supported. Recently, Cisco changed that, and now it is possible. Let me show you how it works.

To demonstrate this, I will modify one of the lab flows we use in our training classes. The original logic of the flow was very simple: the call arrives in the flow, the Autonomous AI Agent is invoked, and then the call is transferred to one of the queues with live agents.

Now let’s enhance this flow by adding a database lookup to retrieve customer information based on the caller’s number (ANI).

Here is what a record in our lab database (Airtable) looks like:

Between the Start activity and the Invoke AI Agent activity (Virtual Agent V2), I inserted an HTTP Request activity. This activity connects to Airtable and retrieves customer data such as the customer’s first name and last name. These values are then parsed from the HTTP response and assigned to local flow variables: custFirstName and custLastName.

Here you can see an example of the settings for this activity (of course it depends on the names of your DB columns etc. fName and lName are columns in our lab DB - first name and last name):

The next step is to pass these values to the Autonomous AI Agent through the Invoke AI Agent activity. In this activity, you can use the Event Data settings to provide the values you want to send.

The Event Name field should be left blank, and the data should be formatted as JSON using key-value pairs. This JSON payload is then sent to the AI Agent. The key names you define in the JSON will be available on the AI Agent side as variables.

Once the data is passed, you can use these custom values inside the Autonomous AI Agent, for example in the Welcome Message or in the Instructions. In my example, I configured a personalized greeting by adding {{customer_first_name}} and {{customer_last_name}} to the welcome message.

Don’t forget to save and publish your AI Agent after making the changes.

Now let’s test it. When the call arrives and the Autonomous AI Agent is invoked, the caller hears a personalized greeting such as:

“Hello Dmytro Benda, I’m Julia, your customer service representative at ACME Order Management department. How can I help you today?”

This is a simple but very useful enhancement. It allows you to make interactions with the Autonomous AI Agent more personal and more context-aware by passing customer information directly from the flow.