I learn best when I slow down, write, and understand things visually. Recapping is a way I stay connected to what I learn by running it over again and again. It gives me a sense of visibility to remember, especially when everything around me is moving faster than my own mind can keep up.

I recently tried something small but motivating. I picked up manim, the library behind 3Blue1Brown videos. I cloned the community fork, set it up to create a simple animation. It worked on the first try and I open sourced it. You can access the final video I created here , details here.That feeling stayed with me. It reminded me why slow learning still matters. It gives space for curiosity instead of pressure. So, I will be posting a lot of educational youtube videos that I always wanted to do and open source the video code thats behind. Anyone else can try working on top of it too.

Technology is growing faster than our mind can adapt. Our monkey brain still needs time. This is why learning has to be visual and conceptual. We have to understand what sits under the layers we keep adding. Otherwise everything becomes noise and shortcuts.

I have also realised something else. Claude gives a strong start when I want to bootstrap a project, but it does not always let me build on top of what I already have due to context issues. At this stage no agent is perfect. They help, but they do not replace the way we learn.

People say that in a few years agents will take over the hard work(I recently came across this definition of work that emphasised whats needed). Maybe they will. But it will take time. They need data from us. They need our mistakes. They need our corrections. There are a subset of developers who work towards making better Agents. But until then we live with the errors and move forward. They will learn quietly in the background. I am fine with it.

What worries me is how the future of knowledge is shaping. Large companies will decide who gets agency. A small group will have deeper access while the rest will depend on limited versions. That is not how the internet was imagined.

Google in the olden days used only page indexing and still gave the most relevant results to anyone who searched. Knowledge should be open. Should not be limited to usage.

We do not need giant models for everything. A small model is enough to translate our language to a code or help us understand a concept. Learning should stay accessible.

This is why I keep writing. To remember. To reflect. To stay grounded. To stay curious. And to not let speed replace the joy of understanding something from the root.

PS: I removed like feature on my blog because it created a sense of reward that I did not want. I prefer knowing if someone actually read the piece. I use to mark myself to revisit things later with a clearer mind. If you finished this, you can mark it as read or leave a comment if you want. It stays in your browser until you clear it. :)