Make flashcards from your notes

Most tools that promise flashcards from notes do one thing well: they turn a list into term-and-definition pairs. Paste a page of notes, choose a deck, get a grid. For a history reading that is the whole job.

For a language lesson it is the easy half. A pair such as mercato and “market” is a dictionary entry with a timer attached. It tells you the word exists. It does not tell you the form you heard, the article that came with it, or the sentence where the word was used, and those three are what you actually have to recall. A card built only from the pair rewards recognition of a translation and tests nothing you will need at the till.

What is missing is not generation speed. It is the lesson itself: the specific sentence, the tutor’s correction, the exercise you got wrong twice. A generator that never saw the lesson cannot put that on the card, so the learner adds it by hand, which is the work they were trying to avoid.

Three things, and only one of them is copyable from a word list:

1. The form you met. il mercato, not mercato: the article is part of the word and it is the part learners drop. For a verb, the infinitive stays on the card while the reviewed form is the one in the sentence.
2. The meaning in your own language. One translation is enough; a list of five senses makes the card a reading exercise.
3. The sentence it came in. This is the retrieval cue. On its own, a word can be matched from memory of the list. Inside the sentence from your lesson, it has to be produced.

The card is deliberately plain. There is no illustration to decode and no second question on the back, because the review is already doing one thing: asking you to bring the word back before showing it.

The import is one pass over the notes, followed by one decision from you:

1. Paste the notes, or on a phone photograph the page or choose a PDF. Photo and PDF text is read on the device, so the image is not uploaded for that step.
2. The notes go once to Google Gemini, which proposes the word list: each word with its article, a translation and a sentence from the lesson.
3. You read the proposal and delete what you do not need. This is the part that keeps the deck yours; the model is a fast first draft, not the editor.
4. Save. Each kept word becomes a card with the sentence it arrived in, and enters the review schedule.
5. Review in short recaps when the schedule asks. Nothing calls a model during review.

That boundary is the honest description of the AI use here: one model call per import, none afterwards, no chat window during practice. The page on apps without an AI tutor states it in more detail, including what the model does not do.

Worth knowing before you move a term’s worth of cards:

• No Quizlet or Anki import. Existing decks do not come across today. This is a live gap for anyone with a set they like, and we would rather name it than have you find it after signing up.
• No card authoring screen. You cannot design your own card types, add images, or hand-write a card outside the import and edit flow. Cards are for vocabulary and conjugation from lessons, not for arbitrary facts.
• No shared deck marketplace. Cards are personal, and a shared lesson gives each member their own copies and their own schedule. There is nothing to browse and nothing to sell.
• No course. It does not decide what to study next, does not teach a grammar sequence, and does not replace a lesson. It assumes the lesson happened.

Vocabulary review is available for English, Italian, Spanish, French, German, Latvian and Swedish; conjugation review covers English and Italian. Italian is the deepest case today, which is why the public demo is Italian.

The point of a flashcard is the third and fourth visit, not the first. After a lesson the words are still fresh; the risk is next Thursday, when they have gone. A short recap on the day the schedule asks for is the whole routine: a handful of cards, produced from the sentence, marked right or wrong, and rescheduled.

Recaps are built to finish. When the queue is empty the app says so, and it does not ask for a longer session or a streak to protect. A week with three short recaps does more for a set of lesson words than one long session on Sunday, and the schedule is the part that makes the difference rather than the effort.

If you want to see the loop before making an account, the demo runs a fixed Italian sample lesson in the browser, with nothing to install and nothing saved.

If the question is whether cards from your own notes beat a ready-made deck, the guide to language learning apps compares the categories, and the page for tutors covers handing the same workflow to a student.

LanguageRecap is not affiliated with Anki, Quizlet or any other app mentioned on this site. Written November 2026.