Using LivePerson Functions with a Bot

Introduction

Use the LivePerson Functions integration to invoke a function (lambda) that is deployed to the LivePerson Functions (Function as a Service or FaaS) platform. There are no constraints here; if there is some custom logic (a function) you want to invoke with a bot, you can do it with a LivePerson Functions integration.

To enable the use of LivePerson's Function (FaaS) integrations, please contact your LivePerson account representative (other contact options: chat with us on this page, or message Support) for assistance.

LivePerson Functions — Configure the function

1. Access LivePerson Functions from Conversational Cloud by clicking the menu in the lower-left corner.

   

2. Create and configure a function with the custom JavaScript logic that you need for your use case. As an example, the following function takes item and price from the input Object, and, based on the value of price, it decides which sentence to send back to the bot. (This sentence is subsequently displayed in the conversation.)

   

3. Once you’re happy with your function, deploy it.

   You should see the status of the function change from Modified to Productive.

Conversation Builder — Configure the integration

1. Within Conversational Cloud, open Conversation Builder and click the Integrations tab.
2. Provide a name for the function’s integration. In our example, we'll use UnsolicitedCommentFunction.
3. Set the Integration Type to FaaS.
4. Select the Productive function that you want to integrate.
5. Function Headers: This is where you should set particular key/value pairs like bot variables or slots.
6. Function Payload: This is where you should set a specific JSON body to send to the function. Note: In the example below, for demonstration purposes, the two slots are being set in both the headers and the payload. By doing this, you’re able to understand the correct notation to use. In a real world scenario, the best practice is to use slots, custom variables, and bot variables as Function Headers and a JSON body to be set as the Function Payload.
7. Transform Result Script: To process the response coming from the function, you can set the following code to extract it in a JSON format and then send the sentence (in this case, it's a string) directly as a bot message.

var faasData = botContext.getBotVariable("api_UnsolicitedCommentFunction");
var faasJsonData = JSON.parse(faasData.jsonData);
var jsonResponse = faasJsonData.api_UnsolicitedCommentFunction;
botContext.sendMessage(jsonResponse);

In this scenario the Transform Result Script is used because the function is simply returning a string. In the case where a function returns a complex JSON structure, it's up to you to use the same or to take advantage of the Custom Data Fields. For instance, if the response from the function were:

{“response”: “200$ is way too much for a bag!”}

You could set the following as Custom Data Fields:

Then, you could use the response within a dialog’s interaction by using the following notation:

{UnsolicitedCommentFunction.response}

Here’s a visual summary of what needs to be done within the Integrations tab:

Conversation Builder — Configure the dialog

1. Create an interaction of type Integration, and select from the dropdown the function’s integration.
2. Save the change.

Conversation Builder — Preview

This is the end user's experience in a conversation: