laihenyi/NBLM2PPTX ? reverse-engineered prompt

Build me a simple browser based tool that turns NotebookLM exported PDFs into editable PowerPoint files.

I want to open a single HTML page, paste in my Google Gemini API key, upload a PDF, and have the app process the pages into a PPTX where the background image is separated from the text so I can edit the slides afterward. It should handle multiple pages, let me choose which pages to convert, and show clear progress while it works. If the PDF text can be extracted directly, use that for better positioning, and if not, fall back to OCR with Gemini.

Please make the interface friendly and easy to understand, with a clean modern look, and include the setup flow for first time users. It should store the API key in the browser so I do not have to enter it every time. If you need to check current Gemini or browser docs, go ahead and look them up.