The AI Builder Principles
Twelve principles for building production software with AI when you cannot read the code.
Sep 08, 2026
← Back to blog
Principles
Every post tagged Principles on Sam Henry's blog.
Autonomy is earned
Run the same steps every time, and watch less only after the agent has proven it.
Sep 07, 2026
Classify then route
Match the task to the model that can do it.
Sep 07, 2026
Ground wide, work narrow
Make the agent trace what a change touches, then carry forward only what bears on it.
Sep 07, 2026
Guidance is no guarantee
Write the rules down, then spot-check the work.
Sep 07, 2026
One codebase
One place the software lives, with every change committed so you can always get back to working.
Sep 07, 2026
Own the decisions
Know which decisions are yours, and delegate the rest on purpose.
Sep 07, 2026
Set and enforce boundaries
Declare what a change may reach, and enforce it where the machine can.
Sep 07, 2026
Shared language
Use words that mean one thing to the builder, the model, and the engineer.
Sep 07, 2026
Subtract first
Less is more. Ask what to remove before you add anything.
Sep 07, 2026
Think big, start small
Decompose until the smallest piece still delivers value on its own.
Sep 07, 2026
Trust and verify
Own the definition of done, and write the acceptance criteria yourself.
Sep 07, 2026
Write contracts
Define what each piece promises before anything builds it.
Sep 07, 2026