AI, Software Development, and Centralization

January 18, 2026 12:12pm

Two posts on AI that caught my attention recently:

I love writing software, line by line. It could be said that my career was a continuous effort to create software well written, minimal, where the human touch was the fundamental feature. I also hope for a society where the last are not forgotten. Moreover, I don’t want AI to economically succeed, I don’t care if the current economic system is subverted (I could be very happy, honestly, if it goes in the direction of a massive redistribution of wealth). But, I would not respect myself and my intelligence if my idea of software and society would impair my vision: facts are facts, and AI is going to change programming forever.

Antirez is the creator of Redis, which is well respected cache software that I’ve used professionally. Writing this kind of software requires extra levels of care and thoughtfulness to maintain performance and reliability goals of the project. So, I find it interesting that LLMs are very useful to Antirez in this codebase.

I do think that LLMs will be a standard tool for programming going forward. But that’s probably been obvious for folks working in the field for a while now.

What caught my attention more is his views on the ecosystem around LLMs:

However, this technology is far too important to be in the hands of a few companies. For now, you can do the pre-training better or not, you can do reinforcement learning in a much more effective way than others, but the open models, especially the ones produced in China, continue to compete (even if they are behind) with frontier models of closed labs. There is a sufficient democratization of AI, so far, even if imperfect. But: it is absolutely not obvious that it will be like that forever. I’m scared about the centralization. At the same time, I believe neural networks, at scale, are simply able to do incredible things, and that there is not enough “magic” inside current frontier AI for the other labs and teams not to catch up (otherwise it would be very hard to explain, for instance, why OpenAI, Anthropic and Google are so near in their results, for years now).

This aligns well with how I feel about LLMs. I think anyone who cares about computing and the impacts it has on society has a vested interest in seeing that AI does not become centralized. And I agree that there’s too much practical use for LLMs in coding for things to totally evaporate in a cloud of hype. (Don’t take this to mean there isn’t too much hype around LLMs and a ton of worthless AI slop. I think the industry as a whole will see the bubble burst, but, like previous tech bubbles, we will some aspect of the technology make it through.)

Birchtree blogged about using LLMs to quickly build personalized software:

LLMs have made simple software trivial

I was out for a run today and I had an idea for an app. I busted out my own app, Quick Notes, and dictated what I wanted this app to do in detail. When I got home, I created a new project in Xcode, I committed it to GitHub, and then I gave Claude Code on the web those dictated notes and asked it to build that app.

About two minutes later, it was done…and it had a build error. 😅

What’s happening here, I think, will quickly be a major shift in the industry and it’s one that I’m excited about: software development will become increasingly decentralized and personal. For ages, we’ve all had incredibly powerful computers (and phones!) accessible to us on a daily basis. Yet, our dominant computing experiences have become increasingly centralized. Just think how much of our digital lives runs in the cloud rather than on our personal machines. Now, don’t get me wrong, I love the web. But what I love about it is how it is open and accessible to everyone as both a reader and a writer. However, today the web is dominated by a handful of companies (Google, namely). And it’s the same situation for our personal devices: Apple, Google, and Microsoft retain so much power over what’s allowed to run on their operating systems.

But, I think there’s a path into the future where LLMs help chip away at the centralization we are currently experiencing. Apple and others will lose their positions as gatekeepers and toll extractors if people with no development skills are able to use LLMs to build whatever idea they have into functioning software that runs on their personal devices. I think the trick here is to ensure we don’t just replace Apple, Google, and Microsoft with OpenAI, Anthropic, and, well, Google again.

To that end, it’s become a hobby of mine to experiment with opensource LLMs and the ecosystem around them on my Framework Desktop which can easily run large models on it’s shared memory architecture. I hope to write more about what I’ve been experimenting with here in more detail. But for now, here’s a few pointers if you are interested in this space:

clawd.bot: Think open source Siri with much more capability and integrations. I’ve only just started playing with this, but it’s super interesting.
opencode: A Claude Code like CLI tool that can utilize LLMs running in the cloud as well as models you are running locally, which is how I’ve been using it.