2026 Resolution

Past and Now

The last 3 years were amazing, and the pace of change and improvement was blazingly fast. The world we’re living in (and coding in) is a different world. Almost no one could imagine where we are today—maybe a few did.

LLMs and agents changed everything. I don’t see many developers who write code anymore; developers are usually planning, asking, and steering. Writing software is becoming an agent task. Even “reviewing” is starting to look like a mix of LLM help and human reading. In the near future, you might not read the code at all—you might only look at some metrics and the output.

So I think we should plan for this now (and for what’s next), and prepare for this shift. You might feel fear or joy, but I believe that as long as you’re a problem solver, you can make a good living—and be one of the people who actually enjoys this wave of change.

Future

Human part

I created this blog a few months back. I only posted once and then… nothing. It should be a place for me to publicly express what I’m learning. The main user of this blog is me, but I’d be happy if people find it, come here, and read.

Why am I saying that? Because what humans write is still one of the best things we have. You can ask LLMs about many things, but when you’re reading a real human opinion—someone who wrote something (even with help from LLMs)—it’s different. So I want to be part of that good feeling.

One of my 2026 goals is to write more than I code. Maybe I already achieved that by the end of this post, but what I really mean is: I want to write as much as I wrote code in previous years. Writing and sharing findings is a big part of this year for me.

People are learning a lot by using LLMs and agents, and they share their findings. I learn a lot from them. I also want to contribute, and to get a better understanding of what agents are doing—and how we can use them more, and learn more.

AI part

I said I think writing code is becoming something for agents. So I believe that if you want to build a library, you should build it for agents too. I want to spend more time on GitHub and open source, but my goal is to write tools that help agents build better.

The most important part for me is making sure what I write is readable by agents—so they can use it well, and help developers discover something new.

I actually started doing that. My plan is to recreate the Jalali Pandas Python library and make it something usable for agents. I also want to create a really good LLMs.txt document that helps agents use it—and makes human life easier too.

In general, I want to focus more on building tools. Even if they’re only slightly different from existing ones, I think it will help me understand a lot of things better. I’ll build in public as much as I can.

Side Project

I want to ship a few ideas I built last year. I sometimes postpone shipping. This year I want to ship—and fail a few times—so I can learn more. At the end, I want to say I enjoyed it and built a few good things.

The key point for me: I’ll ship, and I’ll fail. Nothing should be dead code on my machine.