Guide Coding is the practice of directing AI code generation through intentional human design. This is its manifesto — a draft, open to refinement through the same mechanism it advocates for.
We are uncovering better ways of developing information systems by guiding artificial intelligence rather than relinquishing control to it. Through this work, we have come to value:
That is, while there is value in the items on the right, we value the items on the left more.
The human owns the architecture; the AI generates the syntax.
Syntax is a commodity. Design is the craft.
Understand the problem before you prompt the solution.
Every system serves someone. Keep that someone in view.
Constrain the probabilistic to achieve the predictable.
If you cannot explain the system to the AI, the design is the problem — not the prompt.
Guardrails must push back. Code that escapes the design is rejected, not reshaped.
Readability counts — doubly for code you did not write.
The AI proposes. The Guide disposes.
Unverified code is not fast — only unfinished.
Show the work. Defend the result.
Accountability does not transfer to the generator.
The Guide decides the why and the what. The AI accelerates the how.