Code suddenly feels cheaper.

That does not make software easy. It makes judgment more important.

For most of my career, I worked around enterprise IT systems from the outside: governance, vendors, contracts, service management, controls, and the messy reality of systems that have to survive inside organizations.

I was close enough to understand the weight of software, but not close enough to build it myself.

Agentic coding changed that.

This blog is my working journal from that edge: where domain knowledge, IT experience, product judgment, UX, architecture, implementation, maintenance, and AI agents start to meet.

I am not interested in AI as a productivity slogan.

I am interested in what happens when people who understand real domains can finally build, test, refactor, review, and document software with the help of many agents — without pretending that code generation is the hard part.

Because the hard part was never only writing code.

The hard part is knowing what should exist, how it should behave, where it can break, and how to keep it understandable over time.