Confidence is cheap. Comparable evidence is slower. How do we compare candidates without letting the first ten minutes write the rest of the interview?
Open the guide →Situation-first library
What decision are you making?
You do not need to know the name of a cognitive bias. Start from the work or life situation, then use a few evidence-aware questions to test the decision.
Deep decision guides
Five decisions where a bias list is not enough.
Each guide connects the situation to a review protocol, realistic practice, nearby explanations and explicit evidence limits.
A target date is not a forecast wearing a tie. How do we estimate a project when the business already has a date in mind?
Open the guide →The first number is information, not gravity. How do we use the other side's offer without letting it define our range?
Open the guide →Three fluent answers are still one model conversation, not three independent sources. How do I use AI to move faster without letting the model choose the frame and verify itself?
Open the guide →A green dashboard can still be describing a bad outcome. Is the metric measuring the goal, or has the metric quietly become the goal?
Open the guide →Review candidate judgments before confidence, first impressions or outcome stories become stronger than the evidence.
Open →Estimate time and effort with an outside view instead of relying only on the current plan.
Open →Separate what was knowable during an incident from what became obvious only after the outcome.
Open →Review roadmap choices without letting the loudest request, recent event or prior investment define value.
Open →Compare tools using precommitted criteria rather than demos, authority or the first price shown.
Open →Make a forecast that can later be checked, updated and learned from.
Open →Judge performance without letting one recent result or known outcome rewrite the quality of the underlying work.
Open →Protect a negotiation from the first number and from pressure to justify a position already taken.
Open →Use AI to expand research without treating fluent text, repeated output or model confidence as verification.
Open →Test whether repetition represents independent evidence or only makes a claim feel familiar.
Open →Compare staying, moving or reskilling without letting salary anchors, sunk time or today's mood decide the future.
Open →Check whether a score is still a useful proxy for the real goal and whether the comparison is being interpreted consistently.
Open →Boundary
A lens is a question, not a diagnosis.
These pages show reasoning risks that may be worth checking. They do not prove that a person or team has a named bias.