I love learning.

AI kinda complicated that.

Whenever I thought of something I want to understand, one of the first things I do is open ChatGPT and start asking questions. It is incredibly good at explaining difficult ideas, answering follow-up questions, giving examples, and adapting its explanations to what I already know.

For quick learning, it is hard to beat.

The problem starts when I want to learn something seriously.

There is a strange contradiction here: we now have access to an AI that can explain almost anything on demand, yet it's rare in practice that we can actually remember it.

Recent research is beginning to show why. A 2025 randomized controlled trial found that university students who used ChatGPT as a study aid performed worse on a surprise knowledge-retention test 45 days later than students who studied without AI. The authors suggest that unrestricted AI assistance can reduce the cognitive effort that helps create durable memories. Read the study.

This does not mean that ChatGPT is bad for learning. In fact, research also finds that generative AI can improve academic performance and other learning outcomes when it is used appropriately. The important question is therefore not “Should I use AI to learn?” but rather “How should I use AI if I actually want to remember what I learn?”

For me, the answer comes down to three things.

1. You need a plan

Serious learning needs some kind of structure.

I'm sure I'm not the only person who has experienced this with ChatGPT.

You are in the middle of an interesting conversation and suddenly think, “Actually, I want to learn this properly.”

So you ask ChatGPT to create a complete learning plan.

It gives you a beautifully organised syllabus with ten sections, several subsections, recommended resources, and a suggested progression.

It looks great, but you never actually finish (or in some cases even start) it.

The problem is not necessarily the quality of the plan. It is the fact that the plan is sitting inside a long chat thread, lost alongside dozens of other conversations.

You have to find it again. You have to remember where you left off. You have to decide what to study next. You may end up starting another conversation because the original one has become too long or confusing.

Eventually, the “plan” becomes another piece of information that you generated rather than something you actually followed.

That is one of the problems I wanted Planverse to solve.

2. You need active recall

Reading something and understanding it in the moment feels like learning, but feeling like you understand something is not the same as being able to recall it later.

This is where active recall comes in. Instead of repeatedly reading information, you try to retrieve it from memory. A simple example is closing your notes and asking yourself: “What were the three main reasons for this?”

The evidence for this is remarkably strong.

In a well-known study by Henry Roediger and Jeffrey Karpicke, students who repeatedly tested themselves retained substantially more information on delayed tests than students who simply restudied the material. The researchers called this the testing effect: testing is not merely a way of measuring learning; it can actually produce better long-term retention. Read the study.

Another study by the same researchers found that repeated retrieval produced substantially better long-term retention than repeated studying, with the benefits becoming particularly clear on delayed testing. Read the research.

This is where I think AI can become particularly powerful.

Instead of asking an AI to simply give you the answer, you can use it to systematically ask you to retrieve it.

3. You need to be able to come back

The final piece is consistency.

Even a well-designed learning session is not particularly useful if you forget about it tomorrow.

When you are learning something substantial, you need to be able to return to it days or weeks later without having to reconstruct what you were doing.

This is difficult when your learning history is spread across separate AI conversations, motivation and interest just kinda depletes.

You might remember that you were learning something about history, mathematics, programming, or biology a few weeks ago. But which conversation was it in? What had you already covered? What were you supposed to learn next? Which questions did you get wrong?

The friction involved in answering those questions is enough to make many people simply start again.

I wanted learning with AI to feel more like having a personal study system than having an endless conversation.

How Planverse works

Planverse is built around these three ideas: structure, active recall, and continuity.

You start by spending roughly a minute answering a few basic questions about what you want to learn. Planverse uses those answers to create an organised learning plan for you.

You can create your first plan here.

Once it has been created, your plan is saved in one place. You can return to it whenever you want rather than having to search through old conversations.

From there, each part of the plan becomes something you can actually work through.

Click on a section and Planverse will generate source-backed reading material for that specific subtopic, alongside a question based on the topic. If you already know the answer, there's no need to read the material. Just prove it and move on.

The important part is that the quiz is not an optional extra sitting somewhere else. It is connected directly to the structure of your learning plan.

You only read if you want to.

You are asked a question. You think.

You retrieve.

You find out what you actually know.

You move on.

Your progress is tracked as you work through the plan, so when you come back later, you do not have to remember where you left off or prompt the system with your context and progress. The system already knows... that is main the difference I wanted to create.

ChatGPT is an incredibly useful agent for learning. I still use it myself for quick questions.

But an agent built for chatting is not necessarily a learning system. Planverse is a system built to essentially stop you having to ask and prompt the AI, instead, Planverse prompts you.

The value of Planverse is not that it can explain things better than ChatGPT. The value is that it takes the power of AI and puts it inside a structure designed around actually learning and retaining information.

If you want to try it, you can (make your first learning plan for free)[https://www.planverse.io/plan/new]. You don't need an account to get started.

If you decide you want more from Planverse, you can see the available plans and pricing here.

I'm excited to share this with you and to hear what you think. If you have thoughts to share with me, I'd love it if you could share your feedback. Thanks for reading.