This text is 100% handmade, like the code I am writing...
Recent developments in the AI realm made me think a lot about my next moves in order to not fall behind. The pace is so fast that I am very picky about what to do, because I cannot waste much of my time learning new things that I might not finally use in my daily workflow.
I have to admit, that I ever was a late adopter. I am consciously waiting for hypes being over (or becoming a real thing) and watching technology to proof itself, before fully committing to it and throwing my money at it. I am not the type who buys the next new shiny thing, plays with it for a week and then throws it away if it does not serve the purpose I expected it to.
However, the situation is different now. While I am watching AI coding tools and agents to proof themselves to become a serious help in software engineering, everybody else is doing it already, despite all the downsides. I am still hesitating to ship code that I am not fully aware of, because the output of coding tools and agents totally exceeds my capabilities to review. Output speed triples but my QA seems to slow down by 10x.
At that point I am now standing in some kind of void, frozen from this situation and I cannot move in any direction. On the one hand, manual programming feels wrong in 2026 (damn, it used to be fun!). It feels like making fire with stones while everybody else uses a blow torch. At the same time I am hesitating to go all in on agentic coding as I am watching people burn their entire house, because they left the blow torch doing its work...
Is this just my perception or is this a real thing? Am I the only one who is not seeing a clear direction where to move next?
Top comments (9)
While it's important to be introspective and careful about the future, it's also equally important not to fall to despair and paralysis. Instead of focusing on the closed windows focus on the new open doors. What couldn't you build before but now can? what couldn't you do before and now can? Optimism is the new way forward, that's how I choose to roll nowadays.
I love the optimism, Harry, but in this case, I think Jan’s internal defense mechanism is spot on.
I used to feel that exact same paralysis until I deliberately cut out tech social media. Once you mute the hype machine, you realize that the reality of day-to-day software engineering actually hasn't changed that drastically yet.
I don't want to downplay LLMs at all. They are absolutely changing how we write software. But my fear is that when the dust settles, the influencers who are currently speed-running the industry into oblivion (by glorifying unverified code) will just move on to the next grift. The engineers who actually want to stay in this field long-term are the ones who will be left absorbing the aftereffects and cleaning up the massive technical debt.
To borrow from David Deutsch in The Beginning of Infinity: he warns against being a static society, which I completely agree with. But his core thesis is that we must always demand "good explanations." To me, the current hype around vibe / agentic coding does not provide a good explanation for how this settles into a sustainable practice.
There's a 50% probability I am dead wrong about this. But right now, Jan's caution feels like the only truly "pragmatic" engineering I've seen all week.
I’ve written many times about how coding used to be fun. However, if you want to keep your career moving forward, you probably need to start using AI in production codebases. Using AI to write code proved itself about a year ago.
Not fully “vibe coding,” though. There are already many tools that help ensure integrity and maintain quality, like SpecKit, Claude Superpowers, and methods like BMAD.
This pretty much sums up how I've been feeling lately. It's so difficult to see what will be next move in the AI field right now, that trying to decide what is the right direction to follow is so frustrating, as you never know what's about to be the next new announcement or innovation that could potentially affect your career choices.
You aren't frozen, Jan. Your BS filter is just working perfectly and stopping you from making reckless decisions based on hype. What you are feeling isn't paralysis; it's the natural hesitation of an engineer who actually understands accountability.
What is genuinely absurd right now is how the industry is actively glorifying the act of shipping code we don't even read. Treating "I didn't verify this" as a flex rather than a massive liability is just speed-running technical debt and security breaches.
Let them burn their houses down with the blow torch. You aren't alone in seeing through this.
I'm still in the process of figuring out what to use AI for and what not to use it for. When it's best to use it and when it's better not to. For repetitive work, I usually let AI handle it, but for tasks that require planning and deeper thinking, I prefer to do them myself. I still use AI during work, of course,but it's sometimes difficult for me to find the golden line between these two choices
"Totally relate! I’m also hesitating to fully rely on AI for coding. It speeds up output a lot, but reviewing everything carefully feels overwhelming. I’m trying to find a balance — letting AI handle repetitive tasks while keeping the critical thinking and planning parts for myself."
Late adopter here too, and honestly? 2026 feels like the worst year to be one. The QA slowdown you mentioned is real - output speed goes up but confidence goes down.
Curious what your current workflow looks like. Are you still shipping solo or working with a team that can absorb some of the review load?
Either way, glad someone said this out loud. 👏
I am a late adopter also. I am fully using AI for a sex months now and now I am more like code reviewer than programmer anymore, but I am also nervous too ship a code which I am not aware of.