Building Better Software in the AI Era

Software development is changing faster than most people realize. AI tools promise unprecedented productivity, but they also introduce new risks: projects that look impressive in demos but crumble in production, systems built without fundamentals, and codebases that become unmaintainable within months.

I believe there’s a better way—one that harnesses AI’s power while maintaining engineering discipline.

My Philosophy

AI is a tool, not a replacement for thinking. The most effective AI-assisted development happens when you combine the speed and capability of large language models with structured processes, clear documentation, and human oversight. My work focuses on building systems that are not just functional, but reliable, secure, and maintainable over time.

Context matters more than code. Whether you’re working with Claude Code, GitHub Copilot, or any other AI assistant, the quality of your output depends entirely on the quality of your guidance. I advocate for treating documentation as a living control mechanism—not an afterthought, but the foundation that ensures coherent design across all AI interactions.

Engineering fundamentals never go out of style. Version control, testing, security, proper architecture—these aren’t optional luxuries. They’re the difference between a proof-of-concept and a production system. Between a demo that impresses and software that delivers value year after year.

What I Write About

My posts explore the intersection of AI capabilities and software engineering discipline:

  • Practical AI implementation patterns that work in production, not just demos
  • Structured workflows for AI-assisted development
  • Documentation strategies that guide AI tools toward better outcomes
  • Engineering fundamentals in an age where it’s tempting to skip them
  • When NOT to use AI—because the right tool matters

I draw on real experience implementing AI solutions for the Dutch public sector, including projects with Ministerie van Financiën, Tweede Kamer, Ministerie van Justitie en Veiligheid, and Ministerie van Volksgezondheid, Welzijn en Sport.

Why My Perspective Is Different

My path to software engineering wasn’t typical. I started as an intensive care nurse, studied political science, before transitioning fully into software development. This background gives me an unusual lens:

  • Healthcare experience taught me that systems must be reliable under pressure—failure isn’t just inconvenient, it’s dangerous
  • Political science training sharpened my understanding of organizational dynamics, communication, and how technology intersects with policy
  • Cross-domain thinking helps me bridge gaps between technical implementation and real-world impact

I currently work as Lead Developer at ePublic Solutions, building cloud-driven applications and AI solutions for the public sector and healthcare. I am also available for consultancy and advice work.

What I’m Working On Now

Right now, I’m particularly interested in:

  • Rationalized documentation practices for AI-assisted development (inspired by Parnas and Clements’ “Fake It” paper)
  • Production-ready AI agent systems with proper guardrails, observability, and cost management
  • Scaling AI adoption in organizations without falling into the 95% failure rate trap

Beyond Code

When I’m not building software, I think about the ethical implications of AI systems, particularly around potential sentience and how we should interact with increasingly capable models. I also enjoy music, science fiction, and running—activities that keep me grounded and inspired.

Let’s Connect

I write for people who want to build better software—developers adopting AI tools, engineering leaders scaling AI initiatives, and anyone who believes we can harness these powerful technologies without sacrificing quality.

If you’re interested in practical AI implementation, engineering discipline, or bridging the gap between impressive demos and production-ready systems, I’d love to connect on LinkedIn.

Albert Sikkema