First, you must create a .env file and add your OpenAI API key. You can use the template.env as a reference.

You can run the application by first downloading Docker and using the command docker compose up --build. There are two components to the app, the front-end which you can find at http://localhost:3000/, and the backend. Navigate to the URL, and upload a .txt file, such as test_file.txt. In a few seconds, or longer if you've uploaded a larger file, you will see the summary of your document.

Due to this application just being a POC, many components of it are simplistic. For example, the chunking technique to reduce a large document into smaller, important chunks is done if the document length exceeds an arbitrary word count. An improvement would be to use token counts, and quantitivate metrics to determine when this necessary. Further, the chunking mechanism is quite simple. The API logic is also not concurrent, so if multiple users upload documents it will take a while to complete.