Using OpenAI with Azure.

Introduction

Developing the tool requires a series of subproducts/prototypes to make sure development and deployment of the AI models run smoothly. These subproducts are important steps that help the integration and use of AI within the tool.

Abstract

The first step in using the AI models (GPT-4 and DALL-E 3), was to make a deployment in Azure. Inside Azure, I could manage the models versions, capacity, and rate limit.

Next I needed to use these models in Python. I made use of OpenAI's SDK that allows you to easily connect and interact with a variety of different models on different services. Inside Azure, I created a private token that I could then use to connect and authorize myself to my AI deployments. I created the connection, left default system prompt and simply asked the AI: "Once upon a time" to see what it would do. I configured the code in a way so that the response is streamed back, this way you don't have to wait for the AI to fully complete. Instead, you can watch it generate its response.

Demo

Conclusion

This prototype will allow me to do the most important thing: interact with the AI models. Using my prompting research, I can now add and change the system prompts in order to try and generate pages.