Recall

Everyone reading this has a read-later app containing several hundred articles they will never open again. Saving is easy, costs nothing, and feels productive — which is exactly why it does not work.

The uncomfortable truth is that reading something once and filing it away leaves almost nothing behind. Within a month you retain the vague impression that you read something about the topic, which is not knowledge and cannot be used.

Recall's answer is the one the study literature has recommended for decades and nobody applies to their own reading: test yourself, on a schedule. The saving and summarising are the setup; the quiz is the product.

Recall: Platforms And Content Types

Recall: A Second Brain That Quizzes You

• Spaced repetition is the differentiator — automated quizzes on a personalised schedule, not just a place to store things.
• It saves almost anything: YouTube videos, podcasts, TikToks, Wikipedia articles, PDFs, articles, recipes, books and movies.
• Smart Tags auto-organise and improve over time, so filing is not a task you have to keep doing.
• Connections link related ideas automatically across everything you have saved.
• Augmented Browsing resurfaces connected knowledge while you browse, described as local-first.

Why Does Spaced Repetition Belong In A Read-Later App?

Recall schedules automated quizzes on what you have saved, at intervals designed to build long-term memory. That is spaced repetition, and it is the most consistently supported finding in the study literature.

The mechanism is retrieval. Pulling something out of memory strengthens it in a way that reading it again does not — and reading again feels far more productive, which is why almost everyone chooses the worse method.

The spacing matters as much as the testing. Being asked about something just as you are about to forget it produces the strongest effect, which is why the schedule needs to be computed rather than chosen. Nobody can guess their own forgetting curve.

Applying this to general reading rather than to exam revision is the unusual part. Students use flashcards; almost nobody applies the same technique to the articles, videos and books they consume for work — despite that being where most adult learning actually happens.

The honest caveat is that generated questions test facts more reliably than understanding. A quiz asking what a study found is easier to generate than one asking what its limitations imply, and the second is what makes knowledge usable. The tool gets you retention; the application is still your job.

Students have applied this for years through card systems like Quizlet. The unusual move is pointing the same mechanism at ordinary reading rather than at exam material.

Reading something again feels far more productive than being tested on it. That feeling is why almost everyone chooses the method that works worse.

Smart Tags And Connections Do The Filing

Recall auto-tags saved content with a system it says improves over time, and automatically links related ideas across everything you have saved.

This solves the failure that kills every personal knowledge system: manual organisation. People start with an elegant folder structure, maintain it enthusiastically for two weeks, and then save things loose forever after.

The reason is that filing costs effort at the moment of saving and pays off at some unknown later point, which is exactly the shape of task humans reliably skip. Automating it removes the decision entirely.

Connections are the more interesting half, because they surface something you could not have filed for. Two articles read four months apart that make the same argument from different fields is a link no folder structure would ever have produced.

The realistic expectation is that automatic connections are noisy. Some will be genuine insight and many will be superficial keyword overlap, and the value depends on the ratio being good enough that scanning them is worth the time.

Augmented Browsing Is The Local-First Bet

Recall describes Augmented Browsing as local-first functionality that resurfaces connected knowledge while you browse, with extensions for Chrome, Safari, Edge and Firefox.

The idea is to close the gap between having knowledge and having it at the right moment. A note about a company you read six months ago is useless when you are reading about that company today and do not remember writing it — and surfacing it automatically is precisely when it becomes valuable.

The local-first framing matters more than it sounds, because the feature necessarily involves watching what you browse. A system that resurfaces relevant knowledge has to know what page you are on, which is an enormous amount of information about you.

Processing that locally rather than sending it to a server is the difference between a useful feature and a comprehensive browsing log held by someone else. That is a genuine design decision and worth crediting.

It is also worth verifying rather than assuming. Local-first describes an architecture, not a guarantee, and anyone whose browsing includes confidential material should read the privacy documentation before installing an extension that reads every page.

That is a different promise from a grounded notebook. Gemini Notebook answers questions from sources you deliberately assembled; this surfaces things you had forgotten you saved.

• Save less than you want to — a knowledge base you cannot review is a read-later app again.
• Answer the quizzes when they arrive, since the schedule is the mechanism.
• Scan the automatic connections weekly rather than trusting them individually.
• Read the privacy documentation before installing a browser extension that reads every page.
• Write your own note on anything important, since your words retain better than a summary.

Building A Knowledge Base You Return To

The failure mode is obvious and near-universal: saving everything and reviewing nothing, which is the original problem with an AI layer on top.

Restraint at the point of saving is what makes the rest work.

1. Decide what you actually want to remember, and save only that — not everything you found interesting.
2. Write one sentence in your own words when you save something, because your phrasing retains far better than a generated summary.
3. Answer quizzes the day they arrive rather than batching them, since the interval is the whole point.
4. Review the automatic connections once a week and promote the two or three that are genuinely surprising.
5. Delete things you no longer care about, because a knowledge base with dead weight in it gets abandoned.
6. After a month, check whether you can explain three things you saved without opening the app — that is the only real test.

Where Recall Sits Against A Notes App

The obvious comparison is with a general notes or workspace tool, and the distinction is about what each is for.

A notes app holds what you wrote. It is an output tool — the place where your thinking accumulates — and it is judged on how easily you can get things back out.

Recall holds what you consumed. It is an input tool, and it is judged on whether any of that consumption survived contact with your memory. Those are different problems and both are real.

The overlap causes confusion because both end up as a searchable archive. But searching a notes app finds something you wrote and half-remember; the quiz in Recall is aimed at things you would never think to search for, because you have forgotten they exist.

For most people the honest answer is that both have a place, and the mistake is trying to make one do the other's job — a notes app used as a read-later pile is the most common version of that mistake.

A workspace such as Cleve sits on the output side of the same divide, building context from what you have written rather than what you have read.

Is Recall Your Second Brain?

A good habit when

• You save a great deal and remember very little of it.
• You have abandoned a manual folder system at least once.
• You consume across formats — video, podcasts, PDFs, articles — and want them together.
• You will actually answer a quiz when it appears, which is the whole mechanism.
• You want connections surfaced between things you read months apart.

Not your system when

• Your problem is capturing what you write rather than what you read.
• You need a shared team knowledge base rather than a personal one.
• You will not review anything, in which case this is a read-later app again.
• You need grounded answers with citations rather than retention.
• You are uncomfortable with an extension that reads the pages you browse.