Kenneth Stanley's Why Greatness Cannot Be Planned argues that fixating on a learning objective blinds you to the stepping stones that actually teach you something — which is why NotebookLM's free 50-sources-per-notebook tier and BeFreed's $12.99/month book-to-podcast plan only work when used to follow curiosity, not to fake having read a book.

I bought into "AI tutor" tools the way I buy into most productivity claims — skeptically, and only after they'd already been sitting in my browser tabs for three weeks. What changed my mind wasn't a feature. It was noticing that the two tools I kept actually opening, NotebookLM and BeFreed, are both built around the wrong metaphor for how I use them.

They're marketed as tutors — implying a syllabus, a start and an end, a completion state. But the way I've actually gotten value from either one looks nothing like finishing a course. It looks like Kenneth Stanley and Joel Lehman's argument in Why Greatness Cannot Be Planned: that ambitious objectives are deceptive, because the stepping stones that lead somewhere worthwhile almost never resemble the destination you set out for. "The greatest achievements become less likely when they are made objectives," they write, and I think that's true of learning too, at least for the kind of scattered, cross-domain reading a founder actually needs.

Here's the protocol I run now, which I'm calling the stepping-stone method because it's directly lifted from the book:

  1. Drop a genuinely wide source set into the tool — not one book, but five or six adjacent ones (I load NotebookLM's free tier, which handles up to 50 sources per notebook and 100 notebooks total, with a mix of a book PDF, two competitor blog posts, and a research paper).
  2. Ask it to surface connections you didn't specify, not to summarize what you already know is there.
  3. Follow whichever thread is most interesting, not whichever one maps to your original question — the interesting one is usually the actual stepping stone.
  4. Stop when you hit a decision-relevant insight, not when you've "covered" the source material. There's no course to complete.

NotebookLM — Google folded it into the wider Gemini app ecosystem this year, and it now runs on Gemini 3 across every tier — is free at the Standard level with no card required: 100 notebooks, 50 sources per notebook, 50 chats a day, each source capped at roughly 500,000 words. That's generous enough that I've never hit the ceiling doing research for a single decision. The paid tier that actually matters for founders is bundled into Google AI Ultra at $99.99 a month, which gets you 20TB of storage and 5x the usage limits — overkill unless you're running notebooks across a whole team, in which case it's cheaper than most alternatives at that volume.

BeFreed does something structurally different: it turns books, papers, and expert talks into personalized podcasts and short videos, built by a San Francisco team with backgrounds at Columbia, Google, and Pinterest. Its free tier gives you one summary to test the format; Premium runs $12.99 a month, or $6.58 a month if you commit annually at $89.99 a year. I use it specifically for the books I know I should read but won't sit down and read — the ones where a 20-minute personalized-depth audio version, listened to on a walk, gets me most of the value for a fraction of the time, which is a trade I'll take every time for anything that isn't going to become a primary source I quote from directly.

The honest failure mode: both tools are excellent at surfacing connections and terrible at telling you which connections actually matter to your specific decision. I've had NotebookLM confidently link two sources on a shared keyword that turned out to be coincidental, and BeFreed's condensed audio format occasionally smooths over exactly the caveat that would have changed my mind about something. Stanley's own caveat applies here too — following interestingness isn't the same as abandoning judgment. It's a search strategy, not a replacement for deciding what to do with what you find.

If you're picking one to start: NotebookLM's free tier costs nothing and handles your own documents, which makes it the better first move for research and decision support. BeFreed earns its $12.99 once you've identified the specific backlog of books you keep not reading.

Sources: felloai.com's NotebookLM pricing breakdown (2026); note.com's Gemini Notebook feature explainer (July 2026); kortex-notebooklm.com on free-tier limits; befreed.ai's pricing page and "12 Best AI Podcast Generators" review; Kenneth O. Stanley & Joel Lehman, Why Greatness Cannot Be Planned (Springer, 2015).

Related: How to Find Your Passion · Best Self-Improvement Books · How to Make Better Decisions · What University Will Not Teach You