Field Notes

Observations on how AI systems behave, where they break, and what to build instead.

What Breaks

• Why Your AI Works in Dev and Breaks in Production
  Why systems that seem stable during development fail under real-world conditions
  
  

• When AI Gets It Wrong—and Sounds Right
  How confident errors create invisible risk
  

↓

What That Means

• What Do You Mean by “Done”?
  Why unclear completion criteria lead to unreliable outcomes
  
  

• How AI Speeds the Spread of Zombie Ideas and Orphan Decisions
  How incorrect or unsupported ideas persist and propagate

↓

Why It Happens

• The Invisible Work Holding AI Together
  The hidden human effort required to keep AI outputs usable
  
  

• You Don’t Have a Prompt Problem—You Have a Memory Problem
  Why missing continuity—not prompting—is the real constraint
  

What to Build

Not Smarter — More Stable
Why stability—not intelligence—is the real goal for AI systems

↓

What to Build—Applied

How to Audit Reasoning in AI Responses (A Practical Checklist)
What to look for when AI reasoning is not externally stored

→