I build agentic software systems that have to survive production.

I have spent 25 years building software. The tools have changed often, but the responsibility has not. Someone still has to decide what is worth building, understand the constraints, and own what reaches production.

That is the perspective I brought to agentic engineering. Agents are a powerful new way to make software, but speed only matters when the surrounding judgment is strong enough to use it well.

appointmed is where those decisions became impossible to treat as theory. We built a profitable, bootstrapped healthcare SaaS that now serves roughly 3,500 customers and generates about €2M ARR.

Real people depend on the product in their daily work. The code carries years of decisions, integrations, customer expectations, and release risk. Every new workflow has to survive that reality.

Today, AI agents write 100% of our production code. That did not make engineering disappear. It moved the scarce work toward context, architecture, verification, review depth, and release judgment.

Agents do the implementation work. Humans still own product direction, business impact, and what happens when a change is wrong. That distinction is the foundation of our operating model.

AgenticPrime is the public build log for that operating shift. I share the developer tools, workflows, prompts, checks, and mistakes that survive contact with a real production codebase.

The point is not to make agents look impressive. It is to help software teams use them responsibly, with enough context and verification to ship work they can stand behind.