How to Make Anki Flashcards 10x Faster with AI (for free!)

So I owe pretty much all the success that I had in life down to the fact that I used ENI and I'm not kidding when I say that ENI is a reason that I got into Cambridge University. But basically about 3 4 years ago when I was making ENI related videos on YouTube, we had to make ANI flashcards by hand. So we would have to read a book or we would have to watch a video, take screenshots and then make those flash cards.

But nowadays a lot of that process can be automated by AI which means that you can focus more on understanding the material instead of just typing notes. And I'm going to go over that in this video. But basically you might be asking why would you use ENI over some other programs that have AI built in.

And the reason is that ENI is free and it also has a lot of community support. So you can download ENI. I have a previous video about how to download and use it.

And if you go to add-ons, the add-ons page, then you can see literally hundreds of add-ons made by the community for pretty much anything you can imagine. So, there's like a pop-up Wikipedia add-on. There's a field autocomplete.

There's a Chinese add-on, text to speech add-ons, and you can go on YouTube and you can search on key add-ons, and then you can find add-ons for say uh language learning, medical school, and pretty much anything else that you can imagine learning. Now, the way to actually start out with this is you want something that you want to learn. So, for example, there's a video on transformers and how LLMs work.

And you want to go to Google AI Studio and you also want to copy this prompt that I have in the description. So this prompt is on a community that I have and you can just go to the page and then copy over this prompt. So this prompt specifically is for Google Gemini and there's another prompt for chat GPT which you want to pay attention to.

So copy the Google Gemini prompt and paste it into uh Google's AI studio and you can see it takes up this many tokens 435 tokens and we want to copy the URL for the video and then also paste it right at the bottom underneath message. And now you can see it's loading the tokens and it loaded another 473,000 tokens for this uh 27 minute long video. And the reason you want to be doing Google AI Studio is because Google own YouTube and they can download the video.

And this Google Gemini can actually watch the video. Chad GPT is unable to download the video. So it has to rely on the video's transcript to be able to figure out what the video is about.

Whereas this can actually look at the elements or like the visual elements of the video and then make you better flash cards. So this prompt basically makes flashcards. You can read through it and make any edits.

So it says identify the highle concepts and ideas presented relevant equations, facts, make question answer cards. You can also edit this yourself to make close cards. And over time I will be adding more types of prompts here for different types of cards.

But anyway we have this and we will then press run. So, we can have this running and it might take a couple of seconds or minutes depending on how long the video is. So, whilst it's running, you can go back to the beginning of the video or in case you haven't already watched it, you can start watching through the video.

And when you're watching the video, you want to be paying as much attention as you can to the actual concept and the overarching idea of the video. You don't want to be stopping and taking notes because the AI is taking notes for you. So you can focus more on like making connections between what you already know in your mind about the topic and these new ideas that are being presented to you.

Previously you would probably have to pause a video every minute or two minutes and then make some notes and make a flash card and then resume the video and then pause it again and now kind of interrupt the flow. So I think this is better because it means that you can focus more on the understanding aspect. So you can see now the model is thinking over here.

So it's thinking about all the elements of the video that it just watched. So now you can see it took about 2 minutes and this will be done by the time you finished watching the actual video. So now that you're finished with the video, what you can do is you can copy over all this text by pressing the button on the top right and then pressing copy text.

Now you want to be able to import this into ENI. But if you go to ENI and then you go to file and import, you'll notice you actually have to import a file. You can't just paste the text directly into ENI.

So what you want to do is you want to make a txt file. And the easiest way of doing that is going to a website called text editor. co code and then pressing create new text file and then paste the text into here and then we will press download in the top left and then just call this like flashcards and then it will download a flashcards.

txt file which is a text file. Now if you go back to key we can go to file import and then click on flashcards. txt and for the field separator you want to make sure it uses pipe.

If it uses anything else, then it will be messed up because we explicitly told Gemini in our prompt uh right over here. You should include each flash card on a new line and use a pipe separator to separate the question and answer. So, we want to make sure it uses pipe over here.

And now you can see it's laid out quite nicely. And now we will change a note type to basic. And we will choose our deck as well.

So, we'll choose a default deck. And then for existing notes, we can ignore this. And we can add any tags for these particular cards that we're importing.

So we can just type in like LLM and this will include an LLM tag. Free blue one brown as well as another tag. I actually didn't write the tag properly.

So we write LLM. Press enter. Then we write flee one brown.

Press enter. Now if you press import, you can see all of the cards are imported. We can press show.

And then it has all the cards over here. And you'll see one thing that I did for each of these cards. If you go from bottom to top is you can see a timestamp from where in the video this was extracted from.

And the reason I like this, you can actually remove this if you want by removing the part of the prompt that includes this. So number five over here, if you remove that, then it won't include it. But the reason I like doing this is because I can then have the video side by side after I finish watching the video.

If I realize there are any diagrams that I want to be adding, then I can go to that particular point in the video. So I can go to 1 minute and 23 seconds into video and then for example 1 minute 23 seconds. There's no diagram that I want to add over here.

And let's actually find something where a diagram makes sense. So where it says what is that embedding matrix? We might want to go into 13 minutes and 3 seconds into video and then copy over like a screenshot of the embedding matrix.

So I will use my shortcut for this and then I can just press Ctrl +V and paste it over here. So now I can preview the card by pressing the top right preview button and then it looks pretty good. And the nice thing here is we can use videos that are also not on YouTube.

So if you download a video, so for example, I have a lecture for my university which is astrophysical fluid dynamics. If I drag and drop this video or lecture onto here, then it starts uploading and after it's finished uploading, it will show how many tokens this video is worth. And we can also use it to make flashcards based on PDF documents as well.

So we can go to a new chat by going to a chat button on the left. And then in this new chat we can find some lecture notes that we have or any other PDF. So I'm going to use uh my quantum field theory PDF or handout that we got in our final year of Cambridge.

So this is part two uh quantum field theory and I can just drag and drop this handout onto here. And then I can copy the prompt that we have as well. Pasting into here.

We can write a message at the bottom saying uh please make flashc cards for the entire PDF document attached. So now we have 9,400 tokens and let's use the most powerful model which is Gemini 2. 5 Pro and then we can press run.

So it seems it's actually pretty comprehensive because it's going through every single page. So page 14, page 13 and so forth. I guess if you have a really long document, you want to specify which pages you want it to go through.

So you might only specify five pages and then it will be faster. But I think this is pretty good because it's going through all the pages for us. And you can see it took about 3 minutes.

So we can press copy to clipboard down here. We can go to text editor and then press Ctrl +V to paste it. And then we can press a download and call it like quantum field theory.

And then we go back to on. And then we can go to file and then we can press import. And then click on quantum field theory over here.

And you can see made a small mistake over here. So I added another column even though we only have two columns. And the reason is because pipe the pipe operator is actually already used in equations for quantum field theory anyway.

So we should actually go back and then change the operator to something else such as we should change it to semicolon. So we can scroll back up as edit where it says uh use a pipe operator in the output format. We can change it to say use a semicolon separator and then put the semicolon and then say don't use a semicolon for anything else.

And that prevents it from getting confused. Uh we would have to press uh rerun this turn and then it would do it again. But we won't wait for that to finish because this is just a demonstration video.

So going back to an we would then use a semicolon operator. But in this case, since we use pipe initially, you would use pipe. And then add it as a basic note type.

And then for the deck, we would choose the quantum field theory deck that we made. And yeah, then we would press import. And now you can press show.

And you can see it's slightly broken because of the pipe operator that we used. But all the other math was imported just fine, such as all of this, like selfcorrelation equation and so forth. Um, but if you use a semicolon instead, if you're doing something quite math heavy, then that is probably better for you.

But yeah, now you can see it's using the semicolon instead to separate the question and answer section of each new line. And basically, this is quite useful because when you're reading any PDF or reading any book or watching a video, you can focus more on understanding the higher level picture and not have to worry about writing down flashcards and thinking about which flashards to make because this does it all for you. And then you might actually want to make the flash cards first and then as you're reading through it, you would look at the flash cards and then maybe add any pictures to your flash cards as well to help illustrate the point home because unfortunately currently now Google's AI studio does not support adding like extracting photos from a video or PDF and then adding it to any particular flash card that is relevant.

So my workflow usually is before watching a video I would put in this prompt. I'd put in the video or I'd add a PDF and then I would make the flash cards and as I'm watching the video I'd focus as much as possible and when the video was finished then I would go over the flash cards and then I would see the timestamps and I would add in any of the pictures that it's referring to in the timestamps or from the PDF as well. I would add in any of that and then that would improve my flash cards and then I would go through the flash cards because it helps me remember the finer details but I still have the highle picture overview of what is happening.