Writing / positions in public
The thinking behind the practice.
Five current essays on what lasts when AI systems move from demos to production.
Article01
What Survives the Model Swap?
One question separates AI investments from AI theater. Ask it of every initiative on your roadmap.
ReadArticle02
Loop & Graph Engineering: Work That Runs Itself
A loop is the unit of agentic work. A signal is the contract between loops. A graph of loops is how whole jobs get done — with the architect still in the room.
ReadArticle03
The Technology Dissonance
Three years of agent hype, told from inside — and the one domain where the loop actually earns its keep.
ReadArticle04
From Value Stream Management to Value Stream Engineering
VSM measured the white space. Value stream engineering builds in it.
ReadArticle05
The Verifiability Axis
When agents write the code, cognitive load stops being architecture's primary optimization target. Verifiability takes its place — and it changes the shape of good systems.
Read