ChatGPT using SAP CPI

[bhaskrsaishankr]

To the beginners in CPI, for hands-on this blog would be helpful understanding how can we integrate ChatGPT using CPI basic Iflow with steps.

ChatGPT: It's a chatbot powered by OpenAI's GPT (Generative Pre-trained Transformer) language model. It's capable of generating human-like responses to various questions and prompts. It can be integrated into messaging platforms, customer support systems and other apps.

Prerequisites: OpenAI account, SAP CPI, Postman

Step 1:

First, we need to create a secret key on OpenAI website

https://platform.openai.com/account/api-keys

Step 2: 

To authenticate we need to download certificate provided by OpenAI and upload it into CPI system.

https://api.openai.com/v1/chat/completions - Open this URL in the browser and download the certificate into your desktop.

Step 3: 

Go to CPI Monitor -> Manage Security -> Keystore and the upload the downloaded certificate

Step 4:

Create the Iflow using HTTPS as sender adapter as we are going to test the Iflow from Postman.

• Content Modifier - Preparing Request Body (Headers)


• Authorization value is the secret key that we created in Step 1, it should be in the format Bearer {secret key}

• Content Modifier - Preparing Request Body (Body)

{

    "model": "gpt-3.5-turbo",

    "messages": [

        {

            "role": "user",

            "content": "${in.body}"

        }

    ]

}

Here we are building the body in JSON format to post the request content on ChatGPT API.

• Request Reply - HTTP adapter Configurations

• JSON To XML Converter - Converting JSON body received from the ChatGPT API to XML for extracting the content from the required field (using XML format is easier to extract the content from required field)

• Groovy Script - Extracting the content from required field using groovy script and storing in the body

import com.sap.gateway.ip.core.customdev.util.Message;

import java.util.HashMap;

import groovy.xml.*;



def Message processData(Message message) {

    

    def new_body = new XmlParser().parseText(message.getBody(java.lang.String) as String);

    

    def set_body = new_body.choices.message.content.text();

    

    message.setBody(set_body);

    

    return message;

}

 

Step 5:

Testing the Iflow using Postman

This is a basic example shows how to use ChatGPT with CPI. I could find it using very useful in building Iflows real-time.

Furthermore, we can use ChatGPT for many other use cases like building XSLT mappings, groovy scripts, building UDFs which helps in building Iflows with complex logics.

Happy Learning 😊