May 31st, 2026

Articles of the week

This week I picked up these articles:

Learned/Notes

What's easy now? What's hard now?

I've talked about this with coworkers and friends regarding AI: AI is good at things like Web Development, not so good at things like Database internals. However, that's not necessarily true, in this article Marc presents a hypothesis called "The feedback loop" in which the output for a given input, depends on the feedback loop.

By the way, this is my understanding of the blog/article, and I'm open to learn from others if you think I am wrong about something. When using AI, the input is our initial prompt, the feedback loop is the whole interaction until we're done with the session — if we only used one prompt and we're okay with the first response, that's the output, if we continued iterating 3 more times, whatever we ended up with is the output. The feedback loop then, is everything in between, the constraints we set in our prompt, any other prompts, skills, files, data — anything that made the model change or refine the output.

Now, the overall point is that, web dev or database internals don't really specify how well the output is. It is more about the feedback loop (the constraints, rules, code reviews and modifications based off them) that we have. If the feedback loop is me when prompting the model to generate a website, and it's having to wait for me to tell it what I like, and what I don't, which colors and fonts to use, etc. The output is going to be based off me and the time it takes me to decide as well as how good I am at guiding it through correct design and what not. However, if I were really knowledgeable about database internals, and I wrote a very detailed PRD document, have a bunch of constrains to make the model write good and concise code, have ways for it to run tests, tell it how to navigate the whole thing, chances are, the output will be better since the feedback loop is more organized and doesn't depend on me, it's autonomous.

I agree with this thesis. It's pretty clear that models are capable of doing so many things, and it depends on us to be good drivers and steer them correctly. Think about it: how come there are engineers producing so much with AI? Useful tools, finishing projects left and right by themselves, while others are not able to get Sonnet 4.6 to fix a bug in their code?

How to stay resilient in a difficult job

I really enjoyed this article, the main points were that you need to focus on what you can control (which in my opinion is good advise for life in general), find some sort of support if possible, work up to your standards even if the system is broken, to maintain a belief that there is something better in the future, and to stay positive (but realistic).

I've gone through my 20s liking my employes and disliking them at times, and I've definitely gone through times where I don't want to wake up because I've got work, and it's usually because I'm super negative about it, I'm telling myself from the moment I open my eyes in the morning "god I have to go there again". However, it's always helped to be hopeful for a better future, to focus on the things that you like and can control at work, to be around people that make you happy and to be happy with yourself. All of those things definitely make you think about the negative less.

For anyone that's struggling with their workplace, this blog is great!

Essential books for product builders

This isn't really a blog but still something I consumed today and wanted to share. It's so funny how much I read nowadays, when I don't think I read much in my first 20 years of life. I think a lot back to when I first started reading, it was around the time I starting wanting more out of life and I saw that every successful person was a reader, so I decided to become one. From that time I have always read to learn. I definitely now read just to read as well, but most of my reading time is spent reading content that can teach me something I can apply to my life.

I really enjoyed this list by Lenny because he splits it into buckets of "I want to learn _".

I've read a few of the books in that list, and my favorite I'd say is Deep Work, it's a book that I've read probably 3 times by now and it's helped me tremendously to find time and ways to focus and learn things. I'm curious if anyone has read any of the books in that list and if so, which one is your favorite?

Domain expertise has always been the real moat

Last year I started thinking more broadly, I stopped thinking only about the tools and programming languages I use at work and started also thinking about the domain in which I'm using them. I don't think I've found a domain I want to specialize in yet, and I REALLY want to find it soon. But I agree that domain expertise is even more important now.