Integrating OpenAI API for Comprehensive Knowledge of My Web App

Hello,

I’m looking to integrate the OpenAI API with my web app in a more advanced way. I’ve already implemented it using Retrieval-Augmented Generation (RAG) and fine-tuning, but now I want to take it a step further.

I want to teach ChatGPT everything about my web app—its features, why and when they were added, how long they took to implement, and other details (organized by tasks). This includes a large amount of data: features are documented across hundreds of tasks and thousands of comments.

The goal is for ChatGPT to have complete and logical, chronological knowledge of my app. I want to ask questions like:

• When and why was a specific feature added?
• Has a particular feature already been implemented?

RAG doesn’t seem suitable here, as I need ChatGPT to generate answers with full context, not just partial knowledge.

Is this possible? If so, what approach or tools should I use to achieve this?

2 Likes

With assistants tool, you can upload these data as a text file (pdf, txt etc.) in a file search. Or uploading direct from the local disk might be possible programatically. These data which you have uploaded in the openai platform (playground) will be used in model to give the suitable answer. It uses the specific assistant_id where you have uploaded your files to give the response. https://platform.openai.com/docs/assistants/quickstart

Unfortunately, the Assistant feature on the OpenAI platform still depends on embeddings and dividing text into chunks to process and retrieve information. As a result, the model doesn’t develop a complete or cohesive understanding of the entire dataset but instead works with individual segments to generate responses.

1 Like