See It Work — Hard rules it cannot break — not guidelines it might ignore | Book 02 · AI Agents for Executive Decisions

Guidelines are 'should not' — and 'should not' fails under deadline and at scale. Agents need hard rules they cannot break: lines built structurally into the system so crossing them is impossible, not merely discouraged. This constitutional governance is the only approach that scales as the number of agents grows.

A guideline an agent can ignore is no protection at all once you have dozens of agents acting fast. Rules they structurally cannot break are the only thing that holds as you scale.

What this means for you

Agents are bounded by hard rules they structurally cannot break — the only governance that scales. What this means for you: you can let agents move fast without lying awake about a catastrophic action, because the dangerous lines aren't guidelines they might ignore under pressure — they're built into the system so crossing them is impossible.

Hard structural rules replace ignorable guidelines:

Governance & Guardrails

guidelines'should not' — fail under pressure

hard rulescannot be broken

constitutionalrules in the system

scalesthe only approach that does

Agents need hard rules they cannot break, not just guidelines. Constitutional governance is the only scalable approach.

For the technical reader — the command, and how to verify it yourself

# one line · you do not need to run this
see walkthrough

see walkthrough
# -> agents bounded by hard rules they structurally cannot break

Full step-by-step is in Appendix RX: Hands-On Demonstrations in the book.