May 3rd, 2026

Articles of the week

This week I picked up these articles:

• Finishing things - Dave Gauer
• Is AI making developers lazy? - Fran Soto
• Agents memory patterns - Tim Kellogg
• What every dev should know about AI sandboxes - Engineer's Codex

Learned/Notes

Finishing things

As someone who has read many self-help books to learn about time management, how habits work, mindset stuff, I really dislike the victim mentality. I'm sorry, but I cannot stand that, this article was full of it so I did not get to finish it. Even so, it did have some good points about how you can go about finishing things you start that I want to share (and feel free to read the article if you want!)

• Dave uses a physical object made of cardboard to store post-it notes - think a card holder. This way he can only see the one at the top, he uses this box as a FIFO queue, which helps him to only work on the note on top. I thought this was good because it seems to work for him. Habits work when you make it easy to do, and difficult to cheat it. By having this item in front of him, he has to do nothing but look at it to get reminded of what to work on. He also can't cheat it to try to look at the next item because he has to physically like open it.
• The whole time reading this I kept thinking "He could definitely benefit from reading some self-help books" which made me think of something: If you are someone who wants to get everything done, sadly, my friend, you can't, I can't, we can't. Our time is limited, prioritize what's important to you, and understand there is no way to do everything you want.
• If you are also someone who is perhaps struggling with setting habits for yourself or getting things done, these books are good (in my opinion):
  • The power of habit by Charles Duhigg
  • Atomic Habits by James Clear
  • Getting things done by David Allen

Is AI making developers lazy?

This was my favorite read today. This has been on my mind for the most part of this year (can't believe it's May already :/). I kept thinking... I enjoy using AI, but... is it hindering my learning? Am I over-relying on it?

Some good points from the article:

• Yes, there have been studies (the article links to one done by Anthropic) and if you are using AI to generate the code for you without questioning, reviewing, or understanding the code. You are 100% learning a lot less than you would if you were not using AI, or if you were using AI differently.
• The article gives a really good way to use AI to be faster but learn better than if you full-on rely on it, that I want to start doing myself. Basically you use AI as a buddy to bounce ideas off of, validate implementations/ideas, for explanations. You then either implement the code yourself or understand what it generates for you. The article states that if you cannot explain a piece of code, then you aren't learning it.
• The article also has a ton of negative impacts that AI has on developers that are relying on it, and it's worse the more junior you are.

Agents memory patterns

Unless you're brand new to using AI and sort of like giving it memory, you don't need to read this. Basically three ways (and one isn't real memory imo):

• Files (anything it can reference to use to remember something)
• Data in the prompt (system/user prompt)
• Datastores (like a knowledge graph)

What every dev should know about AI sandboxes

I am not into this at all, I just knew of containers and VMs as sandboxes, it was the first time I read about other kinds of sandboxes like GVisors, MicroVM, Simulated, etc. Definitely just like an article to get an overview on different kinds of sandboxes.