I’ve posted a new guide and companion field note on Ask Good Questions that may ring true for anyone doing serious AI-assisted development.
Guide:
The AI Junk Drawer Problem
https://askgoodquestions.dev/guides/the-ai-junk-drawer-problem/
Field note:
When our AI workflow became a junk drawer
https://askgoodquestions.dev/field-notes/the-day-our-ai-workflow-became-a-junk-drawer/
The central idea is simple: more context is not always better context. Rules, notes, reminders, and examples can help, but after a certain point they can also start competing with each other and making the AI less precise.
In practical terms, the AI is no longer just doing the task. It is also sorting through the pile around the task.
That seems especially relevant for Clarion developers, where existing constraints, working patterns, version realities, and compatibility concerns often matter more than shiny rewrites or modern-looking substitutions.
I’ll be exploring this idea further in the upcoming book, Real Programmers Use AI.