I've been pondering how "vibe coding" is becoming the new face of Grey IT - shadow IT's evolution in the low-code/AI era. Sharing thoughts on LinkedIn about these "vibe coders." They're building, deploying, and shouting from the rooftops about making money without "real" development knowledge. By "vibe coders," we're talking about folks rapidly assembling applications, often with no-code or low-code tools. Speed is the mantra, immediate results the goal, but what about the bigger picture?

This is off-topic from my usual content, but recent geopolitical shifts have fundamentally altered how I engage with technology. It's no longer just about convenience versus principles; It feels like we're reaching an inflection point in digital sovereignty.

We've all seen charts like this where models like DeepSeek challenge OpenAI's latest offerings. These comparisons rely on standardized tests to demonstrate capabilities.

After months of coding with LLMs, I've finally gathered enough experience to share my thoughts on how these tools are changing our field.

The initial excitement of AI-assisted development is remarkable. Code scaffolding happens at lightning speed, and applications seem to materialize effortlessly.

Yet, as your codebase expands, you'll hit what developers call the "pit of death" - that critical point where AI-generated code becomes unreliable and troubleshooting grows increasingly complex. This transition from rapid development to maintenance challenges represents a fundamental shift in how we must approach AI-assisted coding.