Everything you need to
study smarter

What each feature actually does

AI flashcards, and how many you get

One upload produces 22 to 40 flashcards, usually in under two minutes. The count follows the number of distinct testable facts in the source rather than its length, so a 40-page chapter with eight definitions yields eight definition cards rather than forty generic ones. Every card carries a question, an answer, a hint and an explanation, which means a card you get wrong tells you why instead of only that you missed it. Review runs on spaced repetition, so material you find hard returns sooner. Cards are fully editable, and Pro exports to Anki and PDF if you already run a review system elsewhere. Free accounts get one upload to test the whole pipeline before deciding whether any of this fits how you already study.

AI notes, and what they leave out

Notes are generated from the same upload as the cards, so the two never contradict each other. They arrive structured into chapters you can skim, with definitions, formulas and worked examples kept intact and the connective tissue of speech removed — greetings, admin, repetition and asides. That is the deliberate part: notes that take as long to read as the lecture took to attend have failed at their job. Key terms are highlighted, each section is anchored to where it came from, and the whole document exports to PDF. Notes can be edited in place on web, iOS and Android. Length scales with the source: a fifteen-minute recording produces a page you can read in three, not a transcript with headings added.

AI quizzes and practice tests

Quizzes are built from what the source actually taught, with multiple-choice questions whose wrong options are drawn from elsewhere in the same material rather than invented. That matters more than it sounds: an obviously absurd wrong answer tests nothing, while a plausible one drawn from the next chapter tests whether you can tell them apart. Feedback is immediate and carries an explanation, and the questions spread across the whole source rather than clustering at the beginning. If the material does not support a full quiz, fewer questions are written rather than padding it with questions the source cannot fairly answer. Questions are written to stand alone as subject knowledge rather than asking what the recording said, so they work as exam practice.

AI study podcasts

Any upload can become a two-host audio podcast discussing your own material, available on every plan including free. It exists for the time you cannot read — commuting, the gym, walking between classes — and it is generated from the same source as the notes and cards, so it covers the same ground rather than a summary of a summary. It is available in seven languages and works from PDFs, lecture recordings and typed notes alike. It is the one feature that is genuinely additive rather than a faster version of something you would otherwise do by hand. Episodes run a few minutes rather than the length of the source, so a fifty-minute lecture becomes something you can finish on a walk.

What Laxu AI does not do

It does not write your essays, and it does not check its own accuracy. Everything is generated from a source you provide, and where that source was misread — a photographed formula, a mistranscribed name — the error carries through to the cards and notes built on it. Anything that matters should be checked against your slides or your lecturer, which is why every section is anchored back to where it came from. It also does not replace attending the class. The material it produces is only as complete as the recording or document you gave it. Nothing you upload is used to train AI models, and audio is deleted from our servers once your notes have been generated.

Which feature to use for which material, ranked by fit

1. Lecture recording → notes first, then cards. An hour of speech is too much to memorise directly; the notes tell you what is worth carding.
2. Textbook chapter → flashcards first. The structure is already there, so the cards land cleanly and the notes add less than they do for audio.
3. Revision week → quizzes. Testing yourself finds the gaps faster than rereading, and the explanations close them on the spot.
4. Dead time → podcast. Weakest for first-time learning, best for keeping material warm between sessions.