Skip to content
APAgenticPrimeby Bernhard Keprt
Build LogReadinessMy Story

The build log

Field notes from the agentic engineering floor.

What we shipped, what broke, and what production pressure taught us about using agents responsibly.

  1. October 30, 2026The agent says done. What does that actually mean?A browser test repair shows why agent completion needs a precise finish line, evidence tied to the tested version, and an explicit handoff to human QA and release approval.
  2. September 11, 2026How game development led me to synthetic customersBuilding small games led to a testing concept now in development: synthetic customers that exercise software through realistic scenarios and expose behavior over time.
  3. July 6, 2026The attention budgetFor a while my limit was tokens. Now it is my own attention: where I point it while agents work across several projects. Why attention became the real bottleneck, and Summon, the tool I am building to spend it on purpose.
  4. July 5, 2026How we run 100% agentic coding at a €2M ARR healthcare SaaSThe operating model behind appointmed's agentic coding workflow: how we shape work, supervise agents, review code, and keep release risk owned by humans.

Stories and field notes on software, technology, and life with AI.

© 2026 goosefraba GmbH

Build LogRSSContactPrivacyImprint