2025 was, at least for me, a sea-change. I’ve been getting paid for programming since about 2011, I began learning programming properly in 2009, and my first contact with writing code was probably trying to mod Knights of the Old Republic in about 2005. 2025 was the first year in 16-plus years where I didn’t personally write one single character of code, and yet my code output was easily an order of magnitude higher than the last several years combined.
I started the year using Cursor, switching to full-time Claude Code sometime in May, and as the Anthropic models have improved, I’ve had less and less contact with the code itself. Since Opus 4.5’s release in late November, I haven’t read a single line of code, let alone written any. (Although this is only true for personal projects - for client work, I review Claude’s code a lot, and act more as the pilot-not-flying than as a passenger. I don’t see this changing.)
After some wrestling with this, and its perceived impact on my self-worth and the value of my labour, I have found some peace, at least in the short-term. I think I’ve decided to allow my actual programming skills to atrophy and just to let Claude take the reins. There is plenty of joy to be found in other places in my life and I don’t need to find it in the act of programming - and honestly, I don’t think I found much joy in the act of programming anyways.
I also feel like I’ll still be able to find a job in this industry for a couple more years at least. Even though there’s utterly gigantic change at the personal level, many organisations that depend on bespoke software move slowly. Any group of people working towards any goal naturally creates a political structure, and that political structure creates friction (some of which is valuable and necessary, of course). I think the job of a sufficiently senior software developer is often really about managing that friction, even in the pre-LLM times.
In the post-LLM times, I believe there will still be roles for software developers, though the responsibilities and day-to-day activities of those roles are sure to change over time and become more overtly about managing the friction than it has been so far. I think it’s likely the job will evolve into a much more politically-active role within an organisation - spending more time directly with stakeholders and end-users and trying to determine higher-order issues for a given program design, more time worrying about Conway’s law and about intra-organisation political whims than about DRY or encapsulation or linting or what have you.
As a side note, it’s possible that the biggest winners of this paradigm shift will be very small organisations. Right now the power and agility of a sales/software dyad is probably higher than it’s ever been.
Anyways, I believe someone will still need to take ultimate responsibility for the software, and I do think that will be a specialised role for some time to come, and it’s probably a role I can do, so at least there’s that. I don’t think I’ll be finding any more joy in that than I did in writing code - at least programming was an activity where you were left alone - but I should be able to put food on the table for a while yet.
I’ve kind of lost my thread here, because I’ve got the flu, so I will leave it there. Here’s a quote I found quite apt:
Claude Code has the potential to transform all of tech. I also think we’re going to see a real split in the tech industry (and everywhere code is written) between people who are outcome-driven and are excited to get to the part where they can test their work with users faster, and people who are process-driven and get their meaning from the engineering itself and are upset about having that taken away.
— Ben Werdmuller via Simon Willison