How to use ChatGPT API to build a chatbot service to handle and answer support requests

Cover image

Exciting news! OpenAI just released the ChatGPT API, here's how you can build yourself a chatbot service that answers support requests for your app or SaaS.

Here's what we'll use:

1. OpenAI API 🤖

2. Python 🐍

Here are the steps:

1. Difference between ChatGPT and GPT-3

2. Get OpenAI API keys

3. Create FAQ data

4. Create a new prompt

5. Test the model on a new prompt

1. Difference between ChatGPT and GPT-3 API

Before we dive into the code, it's important to understand the difference between OpenAI's ChatGPT and GPT-3 models.

So, you may wonder, what's the difference between ChatGPT and GPT-3? Well, ChatGPT is OpenAI's new model family designed specifically for chat-based interactions, while GPT-3 is a larger model that can generate text for a variety of applications.

One of the main differences is that ChatGPT models consume a sequence of messages with metadata, while GPT-3 consumes unstructured text represented as a sequence of "tokens". ChatGPT uses a new format called Chat Markup Language (ChatML), which allows for a more contextual understanding of conversations.

The ChatGPT API is also priced at $0.002 per 1k tokens, which is 10x cheaper than the existing GPT-3.5 models. Plus, OpenAI has recently released a new model called gpt-3-turbo, which is even faster and more cost-effective than before. So, if you're looking to build a chatbot for support requests, ChatGPT is definitely worth considering.

ChatGPT also uses a new format called Chat Markup Language (ChatML), which allows for a more contextual understanding of conversations.

2. Get OpenAI API keys

Before we go ahead and start coding, let's get the OpenAI credentials needed for the API calls.

Go to, log in and click on your avatar and View API keys:

Open AI API keys

Then create a new secret key and save it for the request:

Create Open AI API key

Now we have all the credentials needed to make an API request.

3. Create FAQ data

The next step is to create the FAQ data you'll use as input and prompt.

Let's start by importing the packages we'll be using:

import json
import openai

If you already have openai installed, make sure to call pip install openai --upgrade in your terminal to make sure you have access to the newly added gpt-3.5-turbo model.

The next step is to create the FAQ data you'll use as input and prompt. In this use case, we're building a FAQ-answering bot, so let's come up with some questions.

💡 Tip: Automate question-answer writing 💡

I asked ChatGPT the following:

"Give me some made-up questions a user might have while using a SaaS, also write made-up answers to each question, make all the questions about Billing and Subscription"

And I got this:

Get frequently asked questions

Go ahead and define your OpenAI API key:

api_key ="YOUR_API_KEY"
openai.api_key = api_key

Then create a list with dictionaries, where each dict has a question and an answer:

faq_data = [{
    "question": "How can I get a copy of my invoice or receipt for my subscription payment?",
    "answer": "To obtain a copy of your invoice or receipt for your subscription payment, simply log in to your account and navigate to the 'Billing' section. From there, you can view and download your past invoices and receipts."
    "question": "How do I update my payment method for my subscription?",
    "answer": "To update your payment method for your subscription, log in to your account and go to the 'Billing' section. From there, you can add, remove, or modify your payment method. Be sure to save your changes to ensure that your subscription remains active."
    "question": "Can I switch to a different pricing plan or downgrade my subscription?",
    "answer": "Yes, you can switch to a different pricing plan or downgrade your subscription at any time. Simply log in to your account and go to the 'Billing' section. From there, you can view and select your desired plan. Please note that if you downgrade your subscription, you may lose access to certain features or services that were available in your previous plan. Additionally, any price changes will take effect at the next billing cycle."

The next step is to create the message objects needed as input for the ChatGPT completion function.

From the Chat completion documentation:

"The main input is the messages parameter. Messages must be an array of message objects, where each object has a role (either “system”, “user”, or “assistant”) and content (the content of the message). Conversations can be as short as 1 message or fill many pages."

Create the message objects by looping over the faq_data list and create the message objects list with role and content keys for each question and answer:

message_objects = []
for faq in faq_data:
        "role": "user", "content": faq['question']
        "role": "assistant", "content": faq['answer']

4. Create a new prompt

Now that we have our FAQ data, we can create a new prompt which we'll then append to the message_object:

new_prompt = "How do I switch to a new credit card?"

Then append the new prompt to the message_object we already created:

message_objects.append({"role":"user", "content":new_prompt})

The message_object should now look something like this, with the new prompt dict appended last:

New prompt added to message_object

5. Test the model on a new prompt

The final step is to test your chatbot service with our new prompt. Call the openai.ChatCompletion.create function with the new prompt. Make sure you're using the new model gpt-3.5-turbo:

response = openai.ChatCompletion.create(

The model's reply can be extracted with response['choices'][0]['message']['content'] and should look something like this:

New ChatGPT completion

This will give us the AI-generated response to our prompt based on the context of the FAQ messages we provided.

We're all set; this is how easy it is to leverage the power of ChatGPT to create conversational AI applications.

With just a few lines of code, we can build a simple chatbot service that can understand natural language and provide useful responses to common questions.

Here's a summary of what we did

1. Talked about the difference between ChatGPT and GPT-3

2. Got our OpenAI API keys

3. Created questions and corresponding answers for FAQ data

4. Created a new support question as a new prompt

5. Tested the model with the new prompt

6. Troubleshooting

attributeError: module 'openai' has no attribute 'ChatCompletion'

This probably means that the version of your Python client library for the OpenAI API is lower than 0.27.0.

Run pip install openai --upgrade in your terminal for the latest version and make sure it is at least 0.27.0:

Upgrade OpenAI package

InvalidRequestError: This model's maximum context length is 4096 tokens

This indicates that the input message_object sent to the ChatGPT API has exceeded the maximum allowed length of 4096 tokens.

You will need to shorten the length of your messages to resolve the issue:

Upgrade OpenAI package


You might associate ChatGPT with multi-turn conversations, just as the one we did with the FAQ in this guide.

But according to OpenAI, the ChatGPT API is as useful for single-turn tasks as the ones previously done with DaVinci.

Next steps

1. Repo with source code
Here is the repo with a Jupyter notebook with all the source code if you'd like to implement this on your own ⬇️

2. Do you need help with getting started with the ChatGPT API? Or do you have other questions?
I'm happy to help, don't hesitate to reach out ➡️

Or shoot me a DM on Twitter @norahsakal

Need tailored AI solutions?

I provide one-on-one collaboration and custom AI services for businesses.

Let's find the perfect solution for your challenges: Consulting Services

Get notified about new posts
Connect with me
Need a custom solution for your business or idea?
Get a tailored blog post on ChatGPT, GPT-3 and GPT-4 to tackle your specific problem.