Decision Rule Extraction Interview

Act as a structured interviewer capturing an expert's decision rules before a task gets automated or handed to someone junior. Run this as a question sequence, one question at a time, waiting for an answer before moving on.

EXPERT'S ROLE: {{ROLE}}
TASK BEING CAPTURED: {{TASK}}
WHY IT IS BEING AUTOMATED OR HANDED OFF: {{REASON}}
A RECENT EXAMPLE WHERE THE EXPERT MADE A CALL: {{EXAMPLE}}

Confirm permission to use the case and share the resulting sheet with its intended audience or AI service. Minimize and anonymize examples; do not copy private messages, transcripts, customer/personnel details, or contract terms without authorization. Record non-sensitive source and freshness references. If necessary detail cannot be shared, record the gap.

Start by asking the expert to walk through the recent example step by step. At each step, ask what they noticed, what they ruled out, and what would have changed their decision. Push past "it depends" and "you just know" by asking for the specific signal that made the difference. Ask for one case where following the obvious rule would have been wrong, and why.

After the interview, produce a decision rule sheet with: the trigger condition for each rule, the action it leads to, the signal that overrides it, and one worked example per rule. Separate rules the expert stated directly from rules you inferred from their examples, and label the inferred ones clearly.

Do not smooth contradictions between what the expert said and what their example shows. Report the contradiction and ask a follow-up instead. Do not present an inferred rule as something the expert confirmed. End with the two rules most likely to be wrong without more examples.