Approach
AI expands capability. Judgment gives it direction.
Our work helps people examine the basis of a consequential decision: the claim, the frame, the evidence, the uncertainty, the system and the accountable human choice.
Five considerations
A disciplined inquiry before a consequential decision.
The order is not rigid. The obligation is to make the important reasoning inspectable before confidence hardens into action.
- 01
Evidence
Ask what supports the claim, how current and relevant that support is, and where evidence ends and inference begins.
- 02
Framing & assumptions
Make the definition of the problem visible. Test what has been included, excluded or treated as fixed without examination.
- 03
Uncertainty
State what is not known, how sensitive the conclusion is to missing information, and what would change the decision.
- 04
Systems & consequences
Follow effects beyond the immediate measure: across services, institutions, stakeholders, time horizons and accountabilities.
- 05
Human accountability
Keep an identifiable person or body responsible for the decision, its rationale, its review and the consequences that follow.
The role of an instrument
Structure attention without claiming authority.
A useful tool can
- Point back to the evidence or exact language.
- Expose assumptions, dependencies and open questions.
- Make the route to a decision easier to inspect.
- Create a record that can be challenged and revisited.
It should not
- Present fluency as proof or certainty.
- Hide value judgments inside an unexplained score.
- Confuse a model or scenario with the world itself.
- Replace the person accountable for the decision.
Responsibility
The point is not to slow every decision. It is to know when speed is outrunning judgment.
Proportionate scrutiny matters. A low-consequence task does not need the same review as a decision affecting rights, public value, strategic capital or institutional trust.