What kind of skill does it require to let LLMs write 100% of your code? I'm genuinely asking, what's the hard part that a pre-LLM developer is fundamentally incapable of doing? Is it running the agents in a loop? Or along a state machine? Running them in parallel? Because honestly none of that sounds like anything an experienced software dev shouldn't be able to pick up in two weekends.
How is that measured? Is his stuff maintainable? Is it fast? Are good architectural decisions baked in that won't prevent him from adding a critical new feature?
I don't understand where this masochism comes from. I'm a software developer, I'm an intelligent and flexible person. The LLM jockey might be the same kind of person, but I have years of actual development experience and NOTHING preventing me from stepping down to that level and doing the same thing, starting tomorrow. I've built some nice and complicated stuff in my life, I'm perfectly capable of running a LLM in a loop. Most of the stuff that people like to call prompt/agentic/frontier or whatever engineering is ridiculously simple, and the only reason I'm not spending much time on it is that I don't think it leads to the kind of results my employer expects from me.
I'm letting it write (type out) most (80-98%) of my code, but I see it as an idiot savant. If the idea is simple, I get 100 lines of solid Ruby. Good, saves me time. If the idea is complicated (e.g. a 400-LOC class that distills a certain functionality currently scattered across different methods and objects) and I ask 4 agents to come up with different solutions, I get 4 slightly flawed approaches that don't match how I'd personally architect the feature. And "how I'd personally architect the feature" is literally my expertise. My job isn't typing Ruby, it's making good decisions.
My conclusion is that at this point, LLMs are not capable of making good decisions supported by deep reasoning. They're capable of mimicking that, yes, and it takes some skill to see through them.
Follow the trendline. It went from autocomplete to agentic coding. What do you think will happen to your “good decision making” in a couple years?
As of right now the one shot complex solutions AI comes up with are actually frequently extremely good now. It’s only gonna get better and this was in the last 6 months. You could be outdated on frontier model progress. That’s how quick things are changing.
The thing is that the HOW of today becomes the context of someone else's tomorrow session, that person may not be as knowledgeable about that particular part of the codebase (and the domain), their LLM will base its own solution on today's unchecked output and will, inevitably, stray a little bit further from the optimum. So far I haven't seen any mechanism and workflow that would consistently push in the opposite direction.
Technically that's true, but unless you literally write every single line of code, the LLM will find a way to smuggle in some weirdness. Usually it isn't that bad, but it definitely requires quite a lot of attention.
Nope. Pairing with Claude Code makes programming smoother for me and makes it easier and faster to check different solutions (i.e. "sketch out this approach and let's see how that works out"), but there are no lootbox moments. The LLM does exactly what I tell it in pretty small chunks.
How I work today is still very similar to how I worked in 2023, but now I'm typing a TON of English and very little Ruby. But the overall vibe is nicer and starting on difficult stuff is significantly easier.
I had lots of woah moments where I'd plan changes and the agent would actually find inconsistencies in my understanding and complexities I wasn't considering. It's also been very good on keeping surrounding documentation and other code references updates. Maybe in time I'll get used to it and it'll just be another something I'll take for granted, but right now it feels just like gambling for me LOL
Works just fine for me in Prague, Czechia. The running costs of my EV are exactly the same as those of my previous car (a 1-liter econobox). And it’s a 200hp RWD.
It was Onshape for me, but the same idea. The concepts take just a few hours to "click" (the idea that you're stacking changes chronologically, which is different from e.g. layers in photo editing), but then you can suddenly build like 80% of all tools and mechanisms that you've ever seen. Yes, slowly and usually using less efficient tools and approaches, but you can make most things look and work right SOMEHOW.
reply