GPT Functioning Calling Tutorial Part II- Automate your workflow

Ever find yourself overwhelmed with the flood of business emails? Tired of manually sifting through every inquiry to extract crucial details? You're not alone. But now, thanks to OpenAI's game-changing feature, "Function Calling", you can automate these daunting tasks. Follow this auto-GPT tutorial to revolutionize your business automation strategy using Function Calling, an advanced capability introduced by OpenAI.

This GPT functioning calling tutorial is meant to provide an easy, no-code solution, allowing even beginners to enjoy the perks of language model technology. To better understand the function calling feature, you might want to check out my previous tutorial where I delve into this topic in depth.

Step 1: Test GPT's Information Extraction Capabilities

The first step in our no-code auto-GPT tutorial involves testing GPT's ability to extract structured information from unstructured data, like emails. This process involves creating a Dov file in Visual Studio and setting up your OpenAI API key. Once done, you can import the necessary libraries and load the Dov file.

Following this, you can create a function to pass onto GPT called 'extract information from email'. This function will attempt to extract valuable information from your incoming emails, giving you a sense of how adept GPT is at extracting data.

Step 2: Try Out a Sample Email

To see this feature in action, use a sample email. The email used in this tutorial detailed a potential client's inquiry about purchasing company t-shirts. As you watch GPT process this message, you'll be amazed by its ability to extract key details from the email, including the company name, product details, quantity, and categorize it appropriately. Moreover, GPT can even suggest the next best step to advance the conversation!

Step 3: Convert into an API Endpoint

Next, you need to convert your model into an API endpoint that can be called every time a new email arrives. This is made easy with the open-source library, FastAPI. Creating an API for this model ensures it automatically categorizes and extracts information every time a new email hits your inbox.

To test your server, you can create a 'Hello World' message and run it locally using a new terminal. Once this is running smoothly, you're ready to move on to the next step.

Step 4: Deploy Your API to a Cloud Service

After confirming that your server runs appropriately, it's time to deploy your API to a cloud service. For this, you can use platforms like, which allow easy deployment of apps. Once your project is uploaded to GitHub and linked to your render account, your API can be deployed directly. Testing the server again should show you that your API is running smoothly.

Step 5: Create the Workflow

The final step involves creating a workflow that initiates whenever a new email arrives in your Gmail account. To automate this process, you can use Zapier, which connects different services together seamlessly. Once you set up the workflow, every incoming Gmail email will automatically send a request to the API endpoint created earlier, generating structured information that is automatically added to a Google Sheet.

And there you have it! In less than 10 minutes, you've built an auto-GPT model that can extract crucial information from your emails, categorize them, and add them to your Google Sheet. By integrating Langflow, Langchain, and other platforms, your business automation game just got stronger.

Click here to view the first part of GPT function calling tutorial.