I learned to code the usual way. Small apps, quick wins, things that worked. Until they didn’t.
That’s when I realized most projects don’t fail be...
For further actions, you may consider blocking this person and/or reporting abuse
I know that - building tesseract.gscloud.cz for 7 years, 3 rewrites, finally got to v2.5.0 - mostly REFACTORING now, no major new features
and guess what - you must use your own software to make it superior
This whole topic reminds me of an old banker's joke: "I can't be overdrawn, I still have checks left!" The surface is meaningless without the integrity of the underlying systems.
I haven't figured out vibe coding yet, but based on the resuts I've had just asking AIs technical questions, I can't imagine AIs generating code I could trust for anything longer than a screen or two of code.
I'll be really impressed the first time I ask an AI a question and it replies, "I don't know the answer."
This shift is incredibly important and often underrated. The distinction you're making is between building for novelty versus building for durability.
What interests me most is your point about AI integration. The moment you start working with LLMs or autonomous agents, system design becomes urgent. A monolithic feature-driven codebase becomes a nightmare—because AI can't understand implicit contracts. It needs:
What you're describing—"I stopped moving fast and started moving intentionally"—is exactly what production systems require when you add an AI component. The system has to be stable enough to handle the agent's uncertainty.
This is why I believe the future belongs to architects, not just builders. Building fast is commoditized. Building systems that can be extended and understood by others (or machines) is the rare skill.
"Building fast is commoditized. Building systems that can be extended and understood by others (or machines) is the rare skill." I like this great insight..
Speed gets you to a demo. Systems that can be extended and understood are what get you to production and keep you there. Especially once other developers or machines are involved, clarity becomes a force multiplier.
Really appreciate this perspective. You articulated the novelty versus durability shift better than I could have.
The point about AI not understanding implicit contracts especially resonated. Humans can work around unclear boundaries, but agents need structure, clear failure paths, and visibility or things fall apart fast.
I’m glad you mentioned observability and testability too. Once AI is part of the system, “mostly works” stops being enough. You need designs that can be understood, monitored, and extended with confidence.
Builders aren’t going anywhere, but the role is evolving. The people who think in systems and long-term constraints are the ones building things that actually survive.
Very well-said.
Thank you, I appreciate that. I’m glad it resonated.
This is actual really important, right now due to free or paid AI tool becoming available and they being able to generate the code. Every one just stop thinking and jump on road with fast track.
Without even thinking for moment, let ask AI system to design our own thought process are now behind the drive seat or sometime not even there if other people say lets ship feature or thing this in a week.
With AI making code so accessible, it’s easy to confuse speed with progress. When tools can generate solutions instantly, the temptation is to skip thinking altogether and jump straight to shipping something that “works.” The problem is that without deliberate system thinking, you’re not really designing anything. You’re just assembling outputs. AI should accelerate reasoning, not replace it. If we stop owning the thought process, we lose control of the system the moment things change or break.
Shipping fast still matters, but only when it’s guided by intent rather than pressure. Otherwise, the fast track leads straight to technical debt.
This is actual really important,
Thank you. It’s encouraging to see others recognize how important this shift really is.