https://simonking.dev Notes on software, leadership and engineering judgement. en-GB https://simonking.dev/writing/architecture-decisions-cost The most expensive decisions in software aren't about tools or people. They're about architecture — and you don't pay for them once. You pay for them every time the system changes. Tue, 05 May 2026 00:00:00 +0000 https://simonking.dev/writing/architecture-decisions-cost https://simonking.dev/writing/ai-tools-missed-the-brief I ran the same brief through ChatGPT and Claude. The brief was the test — not the deliverable, not the polish. Neither tool could see what I hadn't said and recognise it as a decision that needed making. Sat, 02 May 2026 00:00:00 +0000 https://simonking.dev/writing/ai-tools-missed-the-brief https://simonking.dev/writing/ai-speed-without-discipline AI didn't change the trade-off between speed and quality. It just made the choice more visible by removing the excuse that time was the constraint. Thu, 30 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/ai-speed-without-discipline https://simonking.dev/writing/production-cost-thinking Engineers who develop the sharpest instincts aren't the ones with the most resources. They're the ones who never had the option to ignore the bill. Tue, 28 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/production-cost-thinking https://simonking.dev/writing/claude-design-not-a-production-tool Claude Design commits to a direction before you have had the chance to explore — that is a signal about what the tool is actually for, not a gap in capability. Sat, 25 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/claude-design-not-a-production-tool https://simonking.dev/writing/ai-code-that-looks-correct AI presents working code and broken code in exactly the same way. That confidence problem is a leadership problem, not a tools problem. Thu, 23 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/ai-code-that-looks-correct https://simonking.dev/writing/the-knowledge-moat When AI makes building cheaper, the advantage moves to whoever understands the problem deeply enough to build something that actually fits — that is a knowledge moat, and it is much harder to copy. Tue, 21 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/the-knowledge-moat https://simonking.dev/writing/environments-you-couldnt-hide-in Engineers who develop judgement fastest are the ones who chose environments where consequences were visible, feedback was short, and there were no layers to absorb the impact. Sat, 18 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/environments-you-couldnt-hide-in https://simonking.dev/writing/developers-not-expensive-decisions-are The salary debate misses the point. Good engineers reduce the cost of decisions — and that has always been where the real expense lives. Fri, 17 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/developers-not-expensive-decisions-are https://simonking.dev/writing/developing-developers Once you understand that the real job is creating conditions for the team to do their best work, everything about how you spend your time changes. Sat, 11 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/developing-developers https://simonking.dev/writing/nobody-tells-you-when-you-become-tech-lead The thing that got you the role is the thing you have to learn to step back from. The real job isn't writing code anymore — it's making sure good decisions compound over time. Fri, 10 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/nobody-tells-you-when-you-become-tech-lead https://simonking.dev/writing/best-interview-question The best interview signal isn't on a CV — it's how someone responds when asked about a decision that turned out to be wrong. Thu, 09 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/best-interview-question https://simonking.dev/writing/if-everyone-agrees-with-you The most valuable person on a team is often the one who tells you when you're wrong. AI tools make this harder to notice — they're designed to agree with you. Sat, 04 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/if-everyone-agrees-with-you https://simonking.dev/writing/consistency-over-brilliance The same trade-off you'd make in hiring — consistent reliability over occasional brilliance — applies directly to how you choose and build with LLMs. Fri, 03 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/consistency-over-brilliance https://simonking.dev/writing/smaller-focused-ai-model More capability isn't always better. In production, a smaller, more focused model you can understand and trust often outperforms a general-purpose one that's harder to control. Thu, 02 Apr 2026 00:00:00 +0000 https://simonking.dev/writing/smaller-focused-ai-model https://simonking.dev/writing/hiring-fewer-junior-developers Optimising for short-term delivery by not hiring juniors is a trade-off that doesn't show up on a sprint board — but it costs you the pipeline that builds your future team. Sat, 28 Mar 2026 00:00:00 +0000 https://simonking.dev/writing/hiring-fewer-junior-developers https://simonking.dev/writing/ai-thinking-not-just-coding For neurodivergent engineers in particular, AI's real value is in helping structure thinking and make communication easier — not generating more code. Wed, 25 Mar 2026 00:00:00 +0000 https://simonking.dev/writing/ai-thinking-not-just-coding https://simonking.dev/writing/quality-in-the-right-places Good engineering isn't about ignoring quality — it's about applying it where it actually improves the product. Time spent perfecting the wrong thing is still wasted. Tue, 24 Mar 2026 00:00:00 +0000 https://simonking.dev/writing/quality-in-the-right-places https://simonking.dev/writing/ai-tools-vs-ai-judgement Access to AI tools is becoming a commodity. The real advantage is the judgement behind how they're used — knowing when to use them, what to trust, and when not to build at all. Mon, 23 Mar 2026 00:00:00 +0000 https://simonking.dev/writing/ai-tools-vs-ai-judgement https://simonking.dev/writing/prototype-vs-production AI is making prototyping faster — but it doesn't change the cost of building something that can actually run in the real world. Prototypes show the idea. They are not the thing. Sun, 22 Mar 2026 00:00:00 +0000 https://simonking.dev/writing/prototype-vs-production https://simonking.dev/writing/what-senior-engineer-means Early on, seniority looks like speed and output. Over time, you realise it's about judgement — knowing what's worth building in the first place. Wed, 18 Mar 2026 00:00:00 +0000 https://simonking.dev/writing/what-senior-engineer-means https://simonking.dev/writing/betting-google-wins-ai Most people frame AI as OpenAI vs Anthropic. But winning usually comes down to distribution, unit economics, and the ability to ship at scale — which points to Google. Sat, 14 Mar 2026 00:00:00 +0000 https://simonking.dev/writing/betting-google-wins-ai https://simonking.dev/writing/what-makes-a-good-dev-in-2026 The industry is splitting into camps around AI. The developers who thrive are the ones who can do both — ship and think — with range across fundamentals, judgement, and stakeholder fluency. Thu, 12 Mar 2026 00:00:00 +0000 https://simonking.dev/writing/what-makes-a-good-dev-in-2026