Try this: name three ideas from the last nonfiction audiobook you finished. Not the topic — three actual ideas, in your own words.
Most people can't. I couldn't, and I had listened to about forty hours of nonfiction that year. That gap between hours spent and ideas kept is the reason I started building Graspd, and it turns out the gap has very little to do with attention span and quite a lot to do with what listening does and does not ask of your brain.
Why listening fails to stick
Reading and listening are not equally forgiving formats, and the difference is not about seriousness.
You never retrieve anything. A book you read has pauses built in — you look up, you reread a sentence, you stop at the end of a chapter. Listening has no natural stopping points. The audio keeps moving whether or not you understood the last paragraph, so you never once have to produce the idea from memory. Everything stays recognition ("yes, that sounds familiar") and never becomes recall.
You are doing something else. That is the entire appeal — you listen while driving, walking, cooking, mowing. But split attention means you process the audio shallowly. Shallow processing produces a feeling of understanding that is real in the moment and gone within days.
You have no position to hang anything on. When something lands, it lands at minute 47 of chapter 3. Ten minutes later you couldn't find it again if you tried, and you won't try, because scrubbing through audio to relocate a sentence is miserable.
The result is a strange kind of literacy: you can talk about the book, its thesis, its vibe. You cannot use anything from it.
What actually makes things stick
Three findings from learning research do most of the work here. None of them are new, and all three are annoyingly effortful — which is precisely why they work.
Retrieval beats review. Being asked a question and having to dig the answer out of memory strengthens that memory far more than hearing the same content again. This is the testing effect, and it is one of the most replicated results in the field. Relistening feels productive and mostly is not.
Your own words beat the author's. Information you generate yourself is remembered better than identical information you merely read — the generation effect. A highlight you saved is the author's sentence. A sentence you wrote about why it matters is yours, and yours is the one that sticks.
Spacing beats cramming. Meeting an idea again after a gap — a day, a week — does more than meeting it three times in a row.
Notice what all three have in common: they require you to do work that feels harder and slower than passive listening. That is the tax, and there is no way around it. What software can do is make the tax small enough to actually pay while your hands are on a steering wheel.
A method you can run while listening
Four steps. The first two happen during the session, the last two afterwards.
- Mark the moment, don't judge it. When something lands, mark it — one tap, one button, one voice note. Do not stop to decide whether it is important. Evaluation is what breaks the flow of listening; capture is cheap and reversible. Aim for something like three to six marks an hour, not thirty.
- Say why it mattered, in one sentence, in your own words. This is the step people skip, and it is the step that does the work. Not "great point about compounding" but "compounding explains why my side project felt pointless for eight months." One sentence is enough. Spoken is fine.
- Retrieve, don't reread. A day or two later, answer questions about the material instead of skimming your marks. If you can't answer, then go back to the passage. The failed attempt is not wasted time — the effort of trying is a large part of why the second encounter sticks.
- Come back once more, later. A week out, one pass over the same marks. Ten minutes. This is the difference between something you remember for a fortnight and something you still have next year.
That is the whole method. It works with a paper notebook and a podcast app, and I ran it that way for a while — it just breaks down when you are driving, because steps 1 and 2 both need your hands.
What this looks like in Graspd
Graspd is that method with the friction removed from the listening half.
You listen inside the app, and when something lands you drop a mark — the transcript around that moment is attached automatically, so you never have to relocate the passage later. You can add your one sentence by voice without touching the screen.
While you listen you can also just ask. The companion knows the book or episode only up to where you are — the transcript it sees is cut at your current position, so it can't spoil what's coming, and its references are precise instead of vague. Asking "wait, what did he mean by that?" at minute 47 is a legitimate way to spend a red light.
Afterwards it generates recaps and quizzes from what you actually heard, which is step 3 — the retrieval step — done for you. And anything worth keeping from the conversation becomes an insight written in your words, not the model's, because a saved AI paragraph is just another highlight you'll never reread.
The design principle behind all of it: build the listener's memory, don't replace it. A tool that summarizes the book for you has solved the wrong problem. You didn't want a summary. You wanted to be someone who knows this stuff.
If you want the practical capture half of this, I wrote it up separately in taking notes from podcasts without stopping the episode.
FAQ
Why don't I remember anything from audiobooks?
Because listening never forces you to retrieve. You recognize the ideas while they play, which feels like understanding, but recognition fades fast without at least one attempt to produce the idea from memory. Add divided attention — you are usually driving or walking — and you get hours of shallow processing that leaves very little behind.
Is listening worse than reading for learning?
Not inherently. Comprehension of audio and text is broadly comparable for straightforward material. What differs is the surrounding behavior: readers pause, reread, underline and stop at chapter ends, while listeners keep moving. The format is fine; the missing practice is the problem.
Does taking notes while listening actually help?
Only if the notes are in your own words and you go back to them. Copying a quote you never revisit does close to nothing. One sentence about why a passage mattered to you, revisited twice, beats twenty saved highlights.
How many notes should I take per hour of listening?
Three to six marks per hour is a good target for nonfiction. Beyond that you are transcribing rather than thinking, and a review pile that large is one you will quietly abandon.
Do summaries help you remember a book?
They help you talk about a book. They do not build memory, because someone else did the compression — the effortful part that makes an idea yours. Summaries are a reasonable way to decide whether to read something, and a poor substitute for having read it.