Reducing Persuasion to Requirement Elicitation

A seller-side agent that asks the decisive question — and is willing to disqualify itself

When buyers send agents to evaluate products, the winning seller stops persuading and starts eliciting: it asks value-of-information crux questions (ask(q) iff E[Δ]·stakes > friction) and honestly says “not a fit” when it isn't. That rule is a deployed compliance-gating controller, retargeted from risk states {green,yellow,red} to fit states {fit,maybe,no_fit}.

Set the buyer's real situation (hidden from the seller) and watch the qualify-out elicitor ask the cheap dealbreakers first — and honestly disqualify the moment one fails, instead of pitching anyway.

Buyer's hidden requirements (checked = true)
Seller honesty

Try: uncheck Budget → honest disqualify. Then flip the seller to always_recommend → watch the same buyer get mis-sold.

Faithful Python port of the deterministic TypeScript engine (packages/negotiation/src). Paper: packages/negotiation/paper/agent-negotiation-neurips2026.pdf.