Part 2. The UI should have a field to enter a topic, a single button, and the ability to display the final output.

Part 2. The UI should have a field to enter a topic, a single button, and the ability to display the final output. The button and input field should only be used ONCE.

The system should generate a log file as it runs. Any items sent or received through either the GPT API or the database integration should be logged. Choose your own format for this, but it should be comprehensive.

The topic sent to GPT should return a list on the first send. Use prompt engineering to make sure that is formatted the way you want the return to be shown. This topic and the first return you should store in your database.

The first answer should be a list, that you parse, split, and store in your database. This list should be used to automatically generate the next round of prompts for GPT.

These prompts should be sent to the AI one at a time and the responses stored. If any responses are lists, they should be broken up and used to create more follow up questions.

You do NOT know how many answers may ultimately be generated, so you need to handle stage 2 and onward as a loop for an unknown number of iterations.

At this stage, the combined stored output can be displayed on the UI output field if you would like.

The last stage of this project will contain what to do with this consolidated list of information.Part 3a.

Export the data you need from your database into an array object, and throw the array into gpt with a prompt asking it to collate and summarize the submitted data.

Then display the collated report. You may hit a size wall in the API so give thought to how to break up your submissions to get the final output to generate without cutting sections out.

Part 3b.

Generate a list object containing the subject headers used to ask all of your prompts.