NotebookBloom: Why I do not guess PDF page numbers from NotebookLM citations

NotebookLM can point to a supporting passage, but a passage location is not always the same thing as the original PDF page boundary.

1. PDF extraction can reorder two-column text.
2. OCR can change words and punctuation.
3. Headers and repeated text can produce convincing false matches.

NotebookBloom keeps the source title and quoted passage, then asks the user to verify the page in the original PDF. A missing page is safer than a precise-looking wrong page.

Research tooling should preserve uncertainty instead of hiding it.

I built NotebookBloom around this workflow: https://notebookbloom.com/blog/notebooklm-zotero/