Why I built Linkiige for the way I actually learn with AI.

I use AI to learn almost every day now. Most of the time it's casual: a quick question and a grab-and-go answer. But every now and then I get pulled into something I actually want to understand, and that's where my chats used to start falling apart.

And honestly, learning with AI is kind of great. It's the closest thing I've had to a patient tutor who's always around and never gets tired of my follow-up questions.

I usually start by asking for a high-level view of the subject, just enough to get a feel for the shape of it. Then I work through it the old-fashioned way: one topic at a time, one chapter at a time, usually in the order the AI suggests.

It feels great. Right up until it doesn't.

The moment my chats start falling apart

Here's what kept happening. I'd read an AI response and there'd be one concept in it I didn't really get yet. So I'd ask about that. Which would expose another concept I didn't get. Which would lead to a third. By the time I looked up, the conversation had drifted pretty far from the question I originally came in with, and I had no clean way to get back.

I actually like learning this way. Following curiosity is the whole point. The problem is that one chat ends up holding five half-finished threads, and none of them are easy to revisit later.

I tried using Projects in ChatGPT to force some order onto it. It never really stuck for me. Opening a new chat every time a topic branched broke the flow. I'd lose context, paste things around, and pretend I was starting fresh when I clearly wasn't. And after a few rounds, those supposedly clean chats got tangled too. Eventually I gave up on that idea.

What I really wanted

The image I kept coming back to was a game map. At the beginning everything is dark. As you explore, the path behind you lights up, and you can always look back and see where you came from, which route led where, and which places you still haven't returned to yet.

That's what I wanted learning with AI to feel like. I wanted to be able to roam around, go down side paths on purpose, follow whatever looked interesting, and still come away with a visible trail of what I'd actually learned. Not just inside one chat. Across all my chats. Even across Gemini, Claude, and ChatGPT.

So I started building Linkiige for myself.

How I use it, day to day

The way I use it now is pretty simple, and that's on purpose. I didn't want it to add friction to the part I already enjoy.

When I'm in a chat and the AI says something that really makes a concept click, I tag that response right there in the panel. That's the whole capture step. I don't break flow, open a new tab, or paste anything into a note app. The reply stays where it happened, and now it's attached to a concept I can actually find again.

Then comes the part that changed how I learn the most: I write a short note in my own words. Not a summary of the AI's answer. More like what just became clearer to me, and why it mattered. It's a bit like that famous Feynman line: “If you can't explain something simply, you don't understand it well enough.” If I can't put the idea into plain words, that's usually a sign I'm not really there yet.

A Claude conversation with the Hashtag Panel open, showing the current concept cluster, a concept memo, and a saved response with an anchor note.
In the chat itself, I can keep the exact reply, the current concept, and the note I'm shaping from it in view at the same time, without breaking flow.

When I wander into a sub-topic, which I basically always do, I just create another hashtag for it and link it back to where I came from. Over time that builds something I didn't expect: a small map of how my understanding actually grew, with the upstream concept on one side and the downstream concepts on the other. When I come back a week later, I'm not staring at a wall of saved answers. I'm looking at a trail.

The Hashtag Knowledge Workspace showing many linked concept cards spread across the canvas with a pinned trail and zoom controls.
Later, the workspace lets me zoom out and see the trail: which idea led to which sub-topic, what's still unresolved, and where I might want to return next.

And when something stops making sense, I can jump straight back to the exact AI reply that originally explained it, in the conversation where it happened. That matters to me more than I expected. The AI's words are the source. My note is what I made of them. Both stay attached.

A full-page focused reading view for one concept, showing the saved response content in a larger, calmer reading layout.
When I need to slow down and really re-read one concept, I can open a focused view instead of trying to recover it from a long chat scroll.

What I notice now, that I did not before

I stop more often. Before, I'd read a clear-sounding AI answer and move on, feeling like I'd “got it.” Now the small habit of writing an anchor note makes me pause and ask whether I actually did. Half the time the answer is no, and I end up asking the follow-up I would've skipped before.

I deviate without anxiety. I used to feel this little resistance to going down a side path, because I knew it would pollute the chat. Now I just tag the side path and keep going. The trail takes care of itself.

I revisit. This one surprised me the most. I used to never reopen old chats because they felt like archaeology. Now I open the workspace, scan the concepts I've been building on, and pick up threads I would've forgotten I ever started.

If you want to try it the way I do

If any of this sounds like your own experience with AI, here's my honest suggestion: don't try to use Linkiige for everything at once. Pick one real thing you're actually trying to learn right now. Save one reply that made something click. Write one anchor note. Link it to one related concept.

That first loop is usually where it starts to make sense. Everything else builds from there.

I'll keep writing here about how I use the tool and how it's evolving. Next time I'll probably go into upstream and downstream concepts, because I think the shape of knowledge matters more than just storing it.

Thanks for reading. If you want to follow along, you can try Linkiige here. And if you do, I'd genuinely love to hear what your own trail starts to look like.

Support

If this way of learning sounds familiar and you want help trying Linkiige yourself, email [email protected].