The Ho System
Anyone can make an AI generate code. The gap is between “AI generated this” and “I built this”—and the bottleneck is judgment, not access.
Structure the collaboration so the human keeps architectural authority and genuine understanding—and encode that structure in artifacts the AI conforms to. A is one bounded, verified, recorded step; a sequence of them is a whole project.
See the full Ho System applied to a real project. Walk through sharibako’s chain and steps as they really are.
“What is this?”
The problem, the claim, and the discipline in readable form—plus the writing behind it.
“I want to try it on my project.”
A short on-ramp to the discipline, then the named parts of the framework and the skills that run it.
“I want it run against my code.”
The practice against your codebase or alongside your team—verified, recorded, reviewable steps.
Andrew T. Marcus—author of the Ho System; builder of Kanyō, a computer-vision system running at Harvard.