Free tools. No account. Nothing stored.

Learn From My Notes

My notes are always complete and almost never useful. I sit through a lecture, I write down most of what gets said, and three weeks later I have a wall of sentences that remind me a lecture happened without telling me what it said. Re-reading does not fix it. Highlighting does not fix it. What fixes it is being asked one specific question about one specific word.

So this tool takes the notes you already have and turns them into a study set. Paste the text, and it splits your notes into concepts, pulls out the key terms, builds flashcards from your own sentences, and gives you a short retest made of the terms you marked weak. It runs in your browser and it is not an AI model. It is a set of plain rules about headings, colons, and repeated words, and nothing leaves your device. There is an optional AI mode as well, one checkbox and off by default, which sends the material to Google Gemini so a model groups your notes into concepts and writes the definitions properly — which I will explain properly below, because a study tool that hides its rules is just a machine you have to trust.

Paste your notes exactly as they came out of your head — lecture scribbles, a transcript, half a summary — and this turns them into a concept outline, a glossary of key terms, flashcards you can mark known, weak, or unknown, and a short retest built from whatever you marked weak. It is a set of plain text rules running in your browser, not an AI model, and your notes never leave this page.

Local only by default. This tool has no account, no server call, and no third-party script. The splitting, the glossary, the cards, the retest, and both exports are plain rules in your own browser. Your note text is never stored, never sent, and never seen by anyone — if you tick the box above, only your known / weak / unknown marks and the topic label are kept in this browser’s local storage, and the Clear button deletes them.

And if you tick the AI box below, the notes and any attached file are sent to Google Gemini so a model can group them into concepts and write the definitions, instead of the rules lifting sentences out of your own text. Nothing is saved here, and unticking it puts you straight back on the local version.

1. Concept outline

2. Glossary of key terms

3. Flashcards

No cards yet.

Build a study set first.

Turn the card over to see your own sentence.

4. What you marked

5. Weak-topic retest

Print it or take it with you

Export preview — only used if your browser blocks the download

Print sheet

Where it stops. The outline is guessed from short lines, lines that end in a colon, and numbered lines. The glossary takes the words your notes repeat, and every definition is a sentence copied out of your own notes — this is a study aid, not a reference, so check anything you plan to write down in an exam. Long pastes are read up to about 120,000 characters, and notes that are one giant paragraph give it no headings to work with.

What it actually does

Five steps, and you can stop after any of them. Most sessions go: paste, build, flip through the cards marking the bad ones, print or export, close the tab.

1. It reads your notes. Paste them into the box, or open a .txt file you already have. The reading is deliberately dumb. A line becomes a concept heading if it is short, or ends in a colon, or starts with a number. So this:

Regulation: the cell cycle is controlled by cyclins and CDKs.

becomes a concept called Regulation, with the rest of the sentence kept as its detail. That is the whole trick, and it is why writing your notes as short labelled lines makes this tool roughly twice as useful as writing them as paragraphs.

2. It pulls key terms out. Terms come from two places, in this order: phrases your notes repeat, then single words that appear in your own labelled lines. Stop words are thrown out, so you never get a card for the or because. Short all-capitals words survive, which is why DNA, RNA, and ATP still make the cut. Every term in the glossary comes with the sentence it was found in and the concept it belongs to, plus how many times you wrote it down.

3. It builds flashcards. Front is the term. Back is your own sentence. That is the whole card — there is no generated explanation, and I will not pretend there is. If your notes said it badly, the card says it badly, which is honestly useful, because a badly-worded definition is usually the exact thing you cannot defend in an exam.

4. You mark each card. Three buttons: I know this, still fuzzy, no idea. Click the same mark twice and it clears. As you go, the counts under the cards update and the glossary badges each term so you can see at a glance what you have already sorted.

5. It builds a retest from the bad ones. Press the retest button and you get a short list — eight questions plus a few topic questions — built only from what you marked weak or unknown, weakest topic first. If everything is marked known, it says so and covers the first few cards instead of pretending it found weak spots. Each question hides its answer behind a Show answer button, so you can actually test yourself before you peek.

Then print it, or export it as TXT or Markdown. The print button fills a print sheet with the outline, the terms, and the things you marked weak. The exports are written by your browser as a plain file download, and the Markdown one is set up so you can drop it straight into Obsidian, Notion, or a GitHub repo.

The rules it uses, and the rules it does not have

This is a text-processing tool, not a language model. There is no API call, no account, no key, and no third-party script on the page. Everything happens in the tab you are looking at. That has two consequences you should know about before you trust it with an exam.

The good consequence: it is fast, it works offline once the page has loaded, and it costs nothing to run. It also cannot hallucinate, because it does not know anything your notes did not say.

The honest consequence: it has no idea what your terms mean. If your notes say “checkpoint fails” in passing, the tool will happily make you a card out of it. It cannot tell a definition from a fragment. It does not know that metaphase plate is worth more than spindle fibres, and it will not correct you when you write something wrong. It reads sentences and counts words. That is the whole job.

So use it as a study set, not as a reference. If a card looks wrong, the fix is in your notes, not in the tool. When the glossary hands you a weak term, write a better sentence in your notes and rebuild — that is the actual study happening, and the tool is just the thing that keeps asking you to do it.

Other limits worth knowing: very long pastes are read up to about 120,000 characters, and anything past that gets cut with a note telling you so. Sentences longer than 240 characters are trimmed with an ellipsis, so a definition that runs on for half a page gets shortened and the tail is lost. Abbreviations get chopped at their periods, so “e.g.” and “etc.” will split a line in a slightly silly place. None of this breaks the tool, but it does mean the output is a draft, not a product.

A worked example

Say I paste in 180 words of mitosis notes, written the way I described above: a title, then one labelled line per stage. The tool reads them and gives me back 8 concepts — Cell cycle and mitosis, Prophase, Metaphase, Anaphase, Telophase and cytokinesis, Regulation, Checkpoints, and Cancer link — each with its own line of detail underneath.

The glossary comes back with things like cell cycle, daughter cells, and opposite poles at the top, because those are the phrases I actually repeated. Underneath them sit the single-word terms from my labelled lines: checkpoint, chromosomes, spindle, metaphase, anaphase, mitosis, cdks, kinetochore, cytokinesis, replication. Twelve cards in medium mode, twenty if I ask for the full set.

Then I flip through them honestly. cell cycle, known — I could say that in my sleep. checkpoint, weak — I know there are three but not what the third one checks. kinetochore, no idea — I wrote the word down because it looked important, which is exactly the case this tool exists for.

Press the retest and I get two questions and one topic question, and the tool tells me the weak topics by name. That is a fifteen-minute session built out of notes I already had, and none of it left the tab. I then hit Export Markdown and put the file in my course folder, where it will still be there next month when the exam announcement finally lands.

What stays on your device

Your note text is never stored. Not in local storage, not in a cookie, not anywhere. It lives in the text box until you close the tab or press Clear. In AI mode it is sent to Google for the one request that builds the set, and nothing is written to this site’s database.

There is one optional checkbox, and it is off by default. Tick it and this browser keeps your topic label and your known / weak / unknown marks so a refresh does not wipe an hour of sorting. That is all it keeps — the marks and the label. The notes themselves are still not saved, which is why coming back tomorrow means pasting your notes again. I think that trade is the right one: lecture notes can contain other people’s names, unpublished course material, or an exam question you should not be circulating, and I would rather you paste twice than have a browser hold a copy of it.

Press Clear and everything goes — the text box, the topic, the marks, and the saved copy in browser storage. Both export files are written straight to your downloads folder by your own browser, and neither one is sent to this site or to me.

How to write notes this can actually use

Three things, in order of how much they help.

Put each idea on its own line and start it with a name. “Prophase:” on the front does more for you here than any formatting trick. If you cannot name the idea, that is information — you did not understand the slide yet.

Use the words your professor uses. The tool can only pull terms that are in your notes. If your notes say “the sticky thing at the centre” and the lecture said “kinetochore”, you will get a card about the sticky thing, and you will still not know the word you need in the exam.

Write one real sentence per term, not a fragment. The card back is your sentence, so a sentence is what you get to revise from. “Cyclin concentration rises and falls” is revisable. “Cyclins” is not.

And do not try to make it pretty. It reads unformatted text, it does not care about your bullet characters or your numbering, and it will not be impressed by anything. Ten lines of plain sentences beats a beautifully formatted page you never wrote.

Frequently asked questions

Does this send my lecture notes anywhere?

Not in the default mode. That page has no network call for your text, no account, and no third-party script. Splitting your notes, the glossary, the cards, the retest, and both exports are all plain JavaScript running in your own tab. If you tick the AI box, the notes and any file you attach are sent to Google Gemini to build the set, and the answer comes straight back to your browser. If you tick the remember box, the only things kept in your browser are your topic label and which cards you marked known, weak, or unknown — never the note text. Clearing the tool deletes those too.

Is this an AI tool? Will it write my definitions for me?

No, and I want to be clear about that because the page could be mistaken for one. Nothing is generated. Every definition on every card is a sentence copied out of your own notes, and the tool says so on the card. That is a real limitation: if your notes explained something badly, the card will be badly explained too. I would rather show you your own sentence than hand you a confident one I made up, especially for something you are about to be tested on.

How does it decide which words are key terms?

Three rules. First, phrases you wrote down two or more times win, because repetition in notes usually means the lecturer kept coming back to it. Second, single words that appear in your own labelled lines come next, sorted by how often you wrote them and then by length. Third, ordinary English words are filtered out, along with generic verbs like “shows” and “starts” that would make useless cards. All-capitals words of three or four letters are kept on purpose, so DNA and ATP survive.

Will my marked cards still be there tomorrow?

Only if you tick the remember box, and only the marks — never your notes. Without it, closing the tab throws everything away, and you rebuild from a fresh paste. With it, a refresh keeps which cards you marked, and the tool reapplies them to the same terms when you build again. Your note text always has to be pasted back in, on purpose. If you would rather have nothing stored at all, leave the box alone; the tool works exactly the same, it just forgets.

Can I use it on my phone, or print it, or take it with me?

All three. It is a plain responsive page, so the paste box and the card buttons work on a phone, which is the situation where you are most likely to review. Print gives you a study sheet with your outline, your terms, and the things you marked weak, and it goes to your browser’s print dialog so you can save it as a PDF. The TXT and Markdown buttons write a real file to your downloads folder, which is the better option if you want the set in your own notes app rather than on paper.

How this fits with the other tools

Notes are only half the problem — the other half is the number at the end of the term. Once you have marked your weak terms and know which topics you are shaky on, the rest of this site turns that into an actual grade.

The pairing I use goes like this. Run your notes through this page first, because the weak terms tell you what to study. Study them. Then check what the marks are actually worth with the GPA calculators, so you know whether the effort is going into the right course. Nine times out of ten it is not the hardest course that needs the work — it is the one where you are quietly guessing.