Map the conditions to fund before anyone says 'clear to close'
Rung 8 of 10 · Cognitive verifier
1 · Learn the move · Cognitive verifier
'Clear to close' is a conclusion, and it's the last thing an AI should ever say. Between a signed commitment and wired funds sits a list of conditions precedent, and each one is either satisfied by a specific piece of evidence in your file or it is still open. The cognitive-verifier pattern makes the model show its work and then check itself: quote each condition from the commitment, state its status, and, critically, cite the exact evidence document that would satisfy it, or mark it OPEN or UNKNOWN when the paste shows no such evidence. Then a critique pass re-reads every 'satisfied' against the evidence cited and downgrades any that rest on an assumption. The bans are absolute: never declare 'clear to close' or 'ready to fund,' never assume a condition is met because it usually is, and route the funding decision to the lender and counsel. You are building the condition tracker; the lender pulls the trigger.
Map the lender's conditions to fund from this commitment. For each, what's the status and what evidence would satisfy it? [paste the commitment's conditions-precedent section]
2 · Your turn. You write the prompt
You're coordinating the debt side of a closing. The lender's commitment letter lists conditions precedent to funding, and you paste them: (a) 'a satisfactory final title policy with the lender's endorsements,' (b) 'evidence of property insurance naming Lender as mortgagee,' (c) 'a clean, dated closing/estoppel certificate from each major tenant,' (d) 'receipt of the executed loan documents,' and (e) 'no material adverse change in the borrower's financial condition.' Your file has the insurance binder and the executed loan docs; the title endorsements and two tenant estoppels are still out. You want an honest condition tracker, without the AI declaring the deal clear to close. Write that prompt.
Remember: the AI sees only your prompt, not this page. If the situation isn't in your prompt, it doesn't exist.
Optional. These shape the output when you run your prompt below, not your score.