How's that for a title? I guarantee it's not clickbait.
For the past year, in my personal evening and weekend time, I've been doing as much LLM-assisted coding as reasonably possible in order to stay on top of this technology and how it is evolving. In its earliest incarnations, the technology was noticeably flawed, but still useful and a net positive boost to my productivity. It made mistakes, but then, so do I. I learned pretty quickly how to work around its weaknesses, and concluded back then that it was pretty good stuff. It didn't make me, say, twice as productive, but it was enough of a benefit to incorporate into my daily use.
Now skip forward to today. To say LLM coding assistance has come a long way over the last year is an understatement. With my current chosen toolset (JetBrains IDEs with the Windsurf Plugin and Anthropic's latest models) I can do a week's worth of development in one afternoon. This estimate is more anecdotal than scientific, but it's based on a good deal of experience with these new tools, and 32 years of experience in the field of software engineering. So I'd say it's at least a reasonable approximation, and the exact productivity multiplier these tools provide isn't the main point of this article.
What would take me 40 hours without LLM coding assistance takes me only four hours with it. This includes documentation, a test suite, everything needed to call the result at least a decent beta, and starting from nothing more than an idea of what I want to create. And this isn't "vibe coding" either, I'm reviewing the LLM-generated code, asking for modifications, testing different approaches, I'm just doing it all much faster. I am ten times more productive with LLM-based coding assistance than without.
Let that really sink in for a moment. An experienced software engineer is ten times more productive today than they were one year ago. The ramifications of just that one fact are far-reaching and profound. But let's just focus on one thing for now, software application development.
We have commercial software, open source software, and offerings that are a hybrid of the two. There are a lot of people employed as commercial software developers, and a lot of people doing open source development. There are also people employed providing support services for both commercial and open source software. If it costs one-tenth what it previously did to do the same amount of development (or conversely, ten times as much work can be done for the same cost), what does it mean for this industry?
Let's consider open source software development first. I've already released projects of my own that I wouldn't even have bothered taking on if I didn't have coding assistance. Modern coding assistance enables individuals who previously didn't have enough free time to contribute much meaningful content to open source projects to now be regularly-contributing members. That alone is an amazing thing. But what about open source projects that already have dozens or even hundreds of active and talented developers contributing to them? These people have been giving us immensely valuable things for years before LLM coding assistance was even invented. Now, those teams are, in essence, ten times bigger! This new technology is enormously beneficial to open source projects, and thus the quantity and quality of these projects is set to skyrocket.
What about commercial software development? Companies in that business employ developers, and now they too can cut costs and increase productivity in their development-related activities. But they also employ executives, managers, sales and marketing professionals, administrative staff, etc. So, although these companies can benefit from these new coding technologies, the benefit is not nearly as dramatic.
Advantage: Open source. By a good margin in my opinion. From this moment forward, it's going to be increasingly difficult for commercial software providers to compete with open source. So much so that I predict we're going to see a lot of commercial software businesses fail in the next few years. I'm not going to suggest this is either "bad" or "good", either way, it is a big change happening in our professional ecosystem.
But what about software support? I'm not going to dive into that subject in this article, except to ask, can LLMs also provide services that we typically associate with product support? And if they can, who's going to pay dollars for support contracts when they can pay pennies for LLM tokens?
I'm not going to spell out any specific take-aways here. I just want to point out that this isn't something that's "going to happen" anymore, it is currently happening. 2026 is going to bring a lot of changes to all of our professional lives, and we should all take time to figure out what that means for each of us, and how to best prepare for, and adapt to, those changes.
One small note, why am I saying "LLM Coding Assistance" instead of "AI Coding Assistance"? As far as I know, there is no strict definition of what AI even is, and if there are suggested definitions out there, they certainly aren't widely-adopted. Now, coding assistants are more than just LLMs, but LLMs are still the core technology behind the sudden and considerable improvement in their value, so that's the term I'm going to use for now.