I’ve relied on AI tools to learn since they’ve been available! The interactive approach that tools like Claude use has always felt better than going through a full textbook — especially when I need a quick refresher. However, as you’d have understood by now, asking something like Claude to “explain thermodynamics” can only do so much to help you understand complex concepts.

Recently, however, I’ve been trying quite a few ways to study with Claude, and they’ve been a game-changer. The strategy, as I realized, was to ask Claude to build things that help you study rather than to ask the AI tool plainly to “explain.” I also combine these methods with skills you can use in both the free and paid versions of Claude.

A flashcard deck that grades itself and remembers what I missed

Spaced repetition without opening another app

Spaced repetition is one of the most effective ways to study, and Claude can help you create the right materials for this purpose. Instead of asking Claude to show flashcards as usual, I usually bring some gamification into it. At first glance, the flashcard set is normal, but each card will ask whether you got it right. Now, the cards you got wrong don’t really disappear into oblivion; instead, they show up in the end.

As you can guess, these flashcards use session-level spaced repetition, and I’ve used this method to study a variety of subjects. Keep in mind that the flashcards will be based on the prompt, and you can also use a Claude Project for better results. This is what I do: I create separate Claude Projects for different subjects and upload the necessary syllabi and study materials. With the right custom instructions, Claude will base all its output on the uploaded material. So I don’t have to worry about where the questions come from.

Here’s the prompt I use to create an interactive flashcard set with Claude:

Build me an interactive flashcard deck on [subject]. Self-grade each card, and if I get one wrong, resurface it again before the session ends instead of just moving on.

Clicking through a concept beats reading about it

Step-through diagrams for anything with a sequence

Most of the time, learning is more than memorizing a concept; instead, it requires you to understand related ideas. This is where I use step-through diagrams you can create with Claude. Over time, I’ve realized that trying to reconstruct an idea or understanding through step-by-step diagrams leads to better retention than explanations or flashcards.

This option has worked great for me while studying concepts so interconnected that you can’t understand one without the other. You can use the prompt given below to make Claude create these pointers:

Walk me through [concept] as a step-through diagram, one stage at a time, not a single static image.

Compared to an interactive flashcard set, however, step-through diagrams require more tokens and time. Therefore, you should be more conscious of what you ask Claude to create a step-through diagram out of.

Dragging a slider taught me supply and demand faster than a textbook

Simulations for the “it depends on X” problem

Simulations are the next great artifacts you can create with Claude to study more effectively. I’ve been using this prompt to create simulations that teach me concepts where X depends on Y and has a different impact on the outcome. For instance, if I want to understand supply and demand and how they impact outcomes, like surplus. Here’s what I used for that:

Build a slider-driven simulation of [concept]. Let me drag [variable] and show me how the outcome changes in real time.

Similarly, I once created a simulation to explain how the sunk cost fallacy works. I’ve also noticed how Claude is intelligent enough to choose a design system that converts, so that you don’t have to worry about the prompt much. Here’s a sample prompt you can use to create such a simulation:

Build an interactive HTML artifact that simulates the core sunk cost theory.

You’d want to try the second prompt if the concept involves too many variables. This way, Claude could choose the best way to show the idea to you through a simulation. As I said earlier, however, you can get better results by running these prompts within a Claude Project rather than a general conversation.

How the /learn skill decides to build instead of explain

The trigger conditions that actually matter

You may already be familiar with Claude’s /learn skill. It is similar to ChatGPT’s Study Mode in that Claude will optimize its responses to help you learn a concept. Most of the time, responses using the /learn skill are better than those without it, and it makes a huge difference when building interactive learning material. Therefore, I’d also recommend invoking the /learn skill before you enter the prompts I mentioned above.

I’ve tried this option with the same prompt, and the results have been impressive. Sure, Claude builds the tool it asks you to, but the way the content is arranged remains unique. Therefore, if you expect the learning materials to use instructional language, you might want to use the /learn skill.

Where this may not be perfect

All these methods have improved the way I use Claude. I get to study various subjects in interactive ways while also leveraging Claude’s excellent reasoning skills. Nevertheless, there are times when I need maximum source fidelity and prefer alternatives like NotebookLM. That is especially the case when I want video or audio overviews.