How to turn a lecture into flashcards

The rules of a good card

The example cards here come from our sample study pack, made from an invented lecture on cardiac output.

1. One fact per card

   If a card asks for two things, you will often know one and miss the other, and then you cannot mark it honestly. Split it.

   WeakWhat are the determinants of stroke volume and how does each one change it?Three facts and three directions in one card.

   BetterAfterload: what happens to stroke volume when afterload rises with preload and contractility unchanged?Answer: stroke volume falls.

2. A short question and a shorter answer

   If the answer runs past a line, it is more than one card. As a working limit, keep the question under 25 words and the answer under 12.

3. Say what the card is about

   Open with the topic, then a colon. A month from now the card will come up on its own, with no lecture around it.

   WeakWhat happens when it increases?What is “it”?

   BetterFrank-Starling law: what happens to stroke volume when end-diastolic volume increases?Answer: stroke volume increases.

4. Ask for recall, not a guess

   Avoid true or false, yes or no, and “which of the following”. You can get those right without knowing the answer. Make yourself produce it.

5. No lists

   A card that says “name all five” fails when you miss one, every time. Ask for one item, or hide one item and show the rest.

   WeakList the three determinants of stroke volume.Miss one and the whole card is wrong.

   BetterStroke volume: set by three things, namely preload, afterload and [...].Answer: contractility. One thing is hidden.

6. Do not give the answer away

   Read the question and check that the answer is not sitting in it. Give numbers their units: “5” is not an answer, “5 litres a minute” is.

7. Reword it, do not copy it

   A fill-in-the-blank lifted straight from the transcript tests whether you remember the sentence. Put the idea in new words and you test the idea.

8. Keep the source

   Note the minute, the slide and the lecturer's words on every card. When a card stops making sense, you can go back and see what was meant.

9. Make fewer cards

   Every card is a promise to review it many times. About 24 cards for an hour of lecture is a good set for most lectures. Pick what you would be sorry to forget.

Doing it by hand

No AI

1. Get the transcript

   Working from text is much faster than pausing a video. See how to get a transcript of a lecture. Saving a lecture's captions as a transcript is free.

2. Mark the ideas, not the sentences

   Read through once and mark what the lecturer stressed, repeated or said you need to know. Skip the housekeeping and the stories.

3. Write one card per idea

   Follow the rules above. Add the minute from the transcript to each card.

4. Cut

   Go through the set once more and delete every card you would not mind forgetting.

With a StreamSnag study pack

ProUses your AI

A study pack does the same job and applies the same rules. It is part of Pro, and it is written by your own AI account: a Claude or ChatGPT subscription you already have, or a pay-as-you-go key. The guide and quiz that come with the cards are covered in how to make a study guide and practice quiz from a lecture.

1. Save the lecture and its transcript

   The pack is written from the transcript and the text of the slides. If the lecture has no captions, Pro can transcribe it on your computer first.

2. Choose an AI account

   AI is off until you pick one in Settings, on the AI tab. It works with a Claude or ChatGPT subscription, or a key for Anthropic, OpenAI, Gemini, DeepSeek or OpenRouter.

3. Press Make study pack

   It is the fourth step on the lecture's path, which “Show the path” opens. Before anything is sent, a sheet headed “Before the AI reads this lecture” shows what it will make, what it sends (the transcript, and the text and speaker notes of that lecture's slides, never slide pictures or the video), to which account, and roughly how much AI use the run will take. You choose which parts to make there, and the number of flashcards in Settings.

4. Let it read, write and check

   It reads the lecture section by section and pulls out facts with the lecturer's exact words. Cards are written only from those facts. Then every card is checked against the rules above, and the ones that break a rule are dropped. A last pass removes anything the lecture does not support. It takes one to three minutes and carries on if you close the popup.

In the statistics sample, 17 cards were written and 9 were kept. Five were over the number asked for and three broke a rule.

Reviewing the cards

ProNo AI

StreamSnag has its own review screen, opened by Flashcards at the foot of the popup. It shows what is due across all your lectures, a few new cards a day, and brings the cards you have seen back on their own days. Your progress stays on your computer.

Every card has a Show me in the lecture button that opens the video at that minute and slide. If you use Anki, each study pack also saves a flashcards file: in Anki choose File, then Import, and pick that file. Here is the file from the sample.

What to watch for

• AI-written cards can be wrong. The checks reduce mistakes. They do not remove them. Every card points to its minute, so check anything important against the lecture itself.
• Check your course's rules. Some universities restrict putting lecture material into AI tools. Whether to do it on a given course is your choice and your responsibility. Writing the cards by hand needs no AI at all.