Decisive AI · Vol. 5 · design before the model
Decision Design Brief — human design work before any AI query
The model is only as good as the question, and the question is our job. This guided brief from Decisive AI (Chapter 7) walks you through CAN vs. SHOULD labeling, the Five Whats, weighted criteria with the Critical Failure Rule, constraints, and a copy-ready Architect's Opening you paste into your own AI tool.
Open the Decision Design Brief
Decisive AI · Vol. 5 · Chapter 7
Decision Design Brief
The model is only as good as the question, and the question is our job.
Zero-Retention Policy:Your data is processed and purged. We don't train on your decisions.
1. The Decision
Label it
- CAN — capability and probability; answers live in data.
- SHOULD — values, accountability, and consequences; returns to a named human.
Follow Decisive Leader on LinkedIn
About this tool
Label before you ship
CAN questions — capability and probability — run freely; answers live in data. SHOULD questions — values, accountability, consequences — can use the model for analysis underneath, but they return to a named human with the label still attached.
Criteria set before output
Weight what matters 1–10. Mark each criterion Primary or Secondary. Any Primary scoring 3 or lower in the matrix is an automatic No-Go — no total can override it. Set weights before you see model output, not after.
Two artifacts from one pass
Export a branded PDF brief for the room. Copy the Architect's Opening plus Dissent Pass, Blind-Spot Probe, and Sensitivity Test follow-ups into your own AI stack. Continue into the EDGE Decision Matrix when options exist.