Flashcards that come from what you actually got wrong

The expensive part of a deck is not making it

Making a card takes a minute. Reviewing it takes a minute a week for a year. That is the real price, and it is why a deck of twenty thousand cards someone else wrote is not a head start — it is a standing daily tax on facts you may already know, paid in the hours you needed for the ones you do not.

The opposite of that is a small deck where every card corresponds to a gap you have confirmed. There is exactly one moment you can be certain a gap is real: the moment you answered a question and got it wrong, or read an explanation and realised you would not have produced it. That moment is also the moment you are least likely to stop and type out a card.

The best card is the one you made the second you got something wrong, because that is the only second you know for certain the gap is real.

How a card gets made

• By hand. Write the front, write the back, save it. Or write one sentence and blank out the part worth remembering: select it and tap Blank selection, or type {{c1::…}} around it. Every blank becomes its own card, showing the sentence’s other blanks filled in, so a sentence’s new cards are spread out, about one a day, instead of arriving together. Nothing clever, and still the right tool when you know exactly what you missed.
• From a question you got wrong, in one tap. When your pick was wrong, Make this mistake a card sits under the explanation. The card shows the question as you saw it and the answer you picked, and asks why it was wrong; its back gives the right answer, why your pick was wrong and why the answer is right, and the textbook pages the AI analysis cited, when it cited any. It goes to a deck of your mistakes, one card per question: miss the same question again later and one more tap updates the card to your newest wrong answer, keeping its place in your schedule.
• From a practice question you have answered. Add to my cards sits under the explanation, never beside a question you are still working on, because the back of the card is the answer. It opens the card already filled in — the objective or the start of the question on the front, the correct option and its explanation on the back — and you save it as it is or change it first. No model writes either card, so nothing is reworded or invented, and a deck never gets the same question’s card twice.
• From an answer in the chat. Ask the assistant about a mechanism, then tap Make cards from this under its answer. It drafts three to five cards from that answer alone, one fact each — a question and its answer, or a sentence with the key term blanked out — and each keeps the textbook pages the answer cited for its fact, when it cited any. Untick any you would not have written, edit any that need it, pick a deck, and add them; nothing is saved until you do, so a card you did not choose never enters your reviews.
• From a diagram. Some things are learned as a picture: the branches of a nerve plexus, the parts of a cell, a labelled figure. Upload a diagram or a screenshot, or open the figure of a question you have answered, or the book page a source points to, and draw a box over each label you want to know. Every box becomes its own card: the front hides that one label, the back shows it, so you practise naming each part in its place instead of reading a list.
• From a book page. Open the page a source points to and tap ＋ Card: the editor opens on the passage the source quotes, ready for you to blank out the term worth keeping, and the card names the book and the page it came from.
• A deck from the question bank. Build one from the questions you have answered, or only the ones you got wrong, for a subject or a system: one card per question, and Refresh adds any that match since. Your weak-spot deck is the same thing for the subject you miss most; build it again after new misses and it adds them.
• From Med Guru’s own decks. Enrol in a deck when you want coverage of a subject rather than a record of your own mistakes.
• From another student. A student can share a deck they wrote, once Med Guru has read it; Add to my decks makes you your own copy, which you can change as you like.

Cards can carry images and audio, which matters for the material where a description is a poor substitute for the picture. Editing a card later changes the text only — fixing a typo or sharpening a phrase does not reset its schedule, because you did not forget the fact, you improved the wording. Take a blank out of a sentence and its card is set aside with its schedule intact, ready to return if you put the blank back.

What the four ratings actually do

The study runner shows one card. You commit to an answer, flip it, and rate yourself: Again, Hard, Good or Easy, by tap or by keyboard. It advances immediately rather than waiting on the network, so a review session on a bad connection moves at the speed you think rather than the speed the signal allows.

The rating feeds a spaced repetition scheduler that keeps a small model of how well you know each card. A new card is introduced in the same sitting: Again brings it back in a minute, Hard in about five and a half, Good in ten, and two Goods in a row graduate it into review. From there it comes back around the day you would be about to forget it, and every rating updates when that is. Miss a review card and it drops into relearning and re-graduates.

• Again on a card you had learned brings it back in ten minutes, in the same sitting, while the reason you missed it is still in your head. It does not start over as a new card: what the schedule knows about it is kept.
• Each card sets its own pace. How long a card waits depends on how well you have remembered it so far and on how much you aim to retain — 90% by default; aim higher and cards come back sooner. A card you keep getting right spaces out, up to a year between sightings. Hard still counts as remembering it: only Again sends a card back to relearning.
• There is no fuzz. The day a card is due is the day it is due, with no randomisation smearing it across the week — so the wait printed under each rating button comes from the schedule itself, not a guess.

Rate yourself honestly. The schedule is only as good as the ratings you feed it, and pressing Good on a card you guessed is a lie you will pay for in three weeks.

Caps, and a badge that tells you the truth

There is a daily cap on new cards and a daily cap on reviews, counted per UTC day. Caps are what keep a deck survivable: without one, a strong week of adding cards becomes an unstudiable pile a fortnight later, and the pile is what makes people quit.

The number on the Cards tab is applied after the caps. It is not the total of everything theoretically due — it is exactly what Study will hand you if you open it now. A badge that promises forty and then serves a hundred and twenty is worse than no badge, so this one does not do that. The one exception is yours to make: open a single deck on its own and it can give you more of its new cards than the daily cap allows — you asked for that deck.

Decks that nest, and enrolments that follow them

Med Guru’s own decks form a tree: a system, the subjects inside it, the topics inside those. Enrolling in a deck enrols you in its whole live subtree, so cards added to a child deck later reach you without you going back to the catalogue to find them.

The catalogue is explicit about where you stand. Each deck reads Studying, or Included via the parent that already covers it, or Add. Each enrolment carries its own new-cards-per-day setting, so you can take pharmacology at twenty a day and something you are only shoring up at five.

What this does not do

It does not make cards a substitute for questions. Recall of an isolated fact and the ability to use that fact under a stem are different skills, and only one of them is what the exam tests. Cards are for the facts that keep slipping; practice is where they get put to work.

The scheduler is a model of forgetting, not a promise about your memory — it gets the timing roughly right across many cards, not exactly right on any one. And Med Guru is a study aid for exam preparation and revision, not medical advice: a card you wrote from an explanation is a revision prompt, and anything that matters clinically should be checked against a primary source.

The same rule that governs the answers governs the cards made from them. If a card came out of a conversation, it inherits whatever that answer was worth, which is the argument for answers that carry their source and for checking one before you revise from it. Verify the page first. Then make the card.