My friend in med school kept complaining about NotebookLM, so I built the fix
I'll be upfront: I'm not a daily NotebookLM user.
I'm a developer, and what I do read constantly is papers — mostly arXiv in ML. I go hunting for new ones pretty much every day, so I know the pain of wrangling sources and references firsthand.
But the actual reason I started building this is my friend in med school.
She lives inside NotebookLM to study. And she kept hitting the same walls, over and over:
- It generates flashcards, but she can't get them into Anki — which is where she actually reviews. Rebuilding them by hand is hours of work.
- The citations are vague. A number next to an answer, a highlighted chunk of a 40-page PDF, no page number, nothing clean she can drop into a reference list.
- Every notebook is its own island. No way to search across all of them once she had a dozen.
At first I figured someone must have solved this already. So I looked. There are a couple of extensions out there, but most of them just do plain export — nothing actually built around doing research and studying. And it's clearly not just her: dig around and you find the same complaints everywhere from students and researchers.
So I started building one.
What it does so far
It's a Chrome extension that sits on top of NotebookLM:
- Pulls citations out and turns them into real APA / MLA / Chicago / BibTeX references you can paste straight into a paper
- Searches across all your notebooks at once
- Exports the AI flashcards to Anki — but with the source printed on the back of each card, so when you review you can actually check where it came from instead of blindly trusting the AI
That last one is the piece I care about most, because it came from a real person's actual problem, not a feature list.
The part nobody tells you
The code was the easy part. NotebookLM has no public API, so everything works by reading the page's HTML directly. Which means the day Google redesigns something, parts of this can just break, and I'll be chasing it. Every tool in this space has the same problem — so, here we are.
Where I'm at
Being honest, because that's the whole point of posting this:
- Users: 0
- Paying: 0
- Revenue: $0
It's still in development. I'm putting it out this early because I'd rather build in the open and get it wrong in public than polish it forever in private and never ship.
So, a genuine question for anyone who uses NotebookLM for research or studying: what's the single thing that annoys you most about it? That's literally going to be my roadmap.
I don't have much experience with NotebookLLM, but what I was not excited about was the fact that I had to bring most of my own materials... I felt like I was delivering the next LLM training data sets.... I would much prefer a private support case that could data mine and reference my project folder.. Thank you for your post, as it has inspired me to also post my new project that I have spent more time polishing than getting early feedback.... sigh...........breath.......sigh....
Ha, the "am I just feeding the next training set" feeling is real. And yeah, having to bring all your own sources is exactly the friction — NotebookLM is basically a reader you have to feed, not something that goes and finds stuff for you. The "point it at my project folder and let it work" idea is a whole different (and honestly harder) product, but I get why you'd want it.
And hey — go post your project. Seriously. I sat on this way too long telling myself it wasn't ready, and the second I posted it I learned more in a day than in the weeks of polishing before. Ship the messy version. Nobody's judging as hard as you are.
I like that you're solving the point where NotebookLM stops being useful rather than trying to compete with it.
The strongest features you mentioned all bridge the gap between generating knowledge and actually using it—whether that's reviewing in Anki, citing sources, or finding something across notebooks. That feels like a much clearer workflow to own than simply adding more AI capabilities.
You put it better than I did in the actual post. "The point where NotebookLM stops being useful" — that's exactly the line I've been circling around and couldn't name.
That gap between generating knowledge and using it is the whole bet. Adding more AI is a race I'd lose against Google anyway; owning the workflow after the AI is done feels like the only defensible spot for a small solo thing. The cross-notebook search especially — that one came straight from watching someone drown in a dozen separate notebooks with no way to look across them.
Really appreciate you taking the time to think it through like this.