This blog is a preview of our session at Sirus Connects on October 9, 2026, where we tackle the price of over-engineering and show what simple, AI-ready code looks like in practice.
Does this sound familiar? You open a solution to do a simple bugfix – like passing one extra field to an external API – and before you know it, you’re clicking through twenty different projects. You work your way through five abstraction layers, pass three design patterns nobody remembers the name of, and get completely lost in a forest of interfaces. By the time you’ve found the actual database call, two hours have gone by. Unfortunately, I’ve been seeing it happen constantly in recent years: we build software as if we need to scale Netflix to a billion users tomorrow, while today we’re simply delivering a solid web app for a few hundred users.
Why do we do this to ourselves?
When I take a critical look around the market, I see three main culprits. The first is Resume-Driven Development: we all want the latest complex patterns on our CV. Then there’s the blind idolisation of Big Tech (“Google does it this way, so we do too”). And finally, the experience to foresee the pain of long-term maintenance is often missing. A complex pattern feels like an intellectual triumph on day one, but six months later it’s a millstone around your team’s neck.
Not an architecture, but a maze
The most painful symptom isn’t even slow code, but the impact on your colleagues. When the juniors and mediors on your team are constantly at your desk with questions because they simply no longer know where or how to make a change, you haven’t built a “robust architecture”. You’ve created a maze.
AI gets lost just as easily
And here’s the modern reality: exactly the same applies to AI. We now massively use AI assistants to code faster. But if a human developer already gets lost in your twenty projects and endless abstractions, an LLM is just as much in the dark. AI thrives on clear, predictable and simple code. An over-engineered codebase is simply not AI-ready.
Discipline and ego-less development
The solution requires a hefty dose of discipline and ego-less software development. With every line of code, we should ask ourselves three simple questions:
- Will the people (and the AI tools) who come after me still understand this?
- Can my solution be simpler?
- Is this extra step really necessary?
The rise of AI forces us to completely reconsider old architectural dogmas. Thanks to AI, we may finally be able to throw overboard certain abstractions we’ve considered “sacred” for years. Because let’s be honest: do we really still need those heavyweight ORMs today? Or can it be much more direct and lightweight?
What I'll show at Sirus Connects
During my session at Sirus Connects, I’m going to name this elephant in the room loud and clear. I’ll bring a number of juicy, true stories from complex organisations that completely went off the rails, and show you step by step what the simpler, AI-friendly alternative looks like.
Ready to rediscover how liberating simple code can be? I’d love to meet you live at Sirus Connects to talk it through together!
Want to follow this live?
Join this session at Sirus Connects on October 9, 2026, from 13h30 to 14h15. Register and secure your spot.
Can’t make it? Reach out to us, we are happy to share our experience and discuss how to give citizen development direction in your organisation.
