Abstract

This paper discloses a method by which an AI agent recommends a voluntary payment amount, a suggested tip in a pay-what-you-want sale, and shows it with stated reasons, without tracking the customer and without revealing the merchant's costs. Six components are specified. (1) The merchant confidentially discloses a cost breakdown by category (materials, labor, a skill premium, overhead, regulatory cost and margin) to the recommender; the customer never sees it. (2) The recommended amount is a non-binding anchor: paying it, more, less or nothing completes the sale, and whether a customer accepted or overrode past suggestions is not recorded. (3) Each recommendation carries plain-language reasons drawn from the cost categories through a controlled vocabulary, never itemized amounts. (4) Suggestions are made consistent across merchants by calibrating against aggregated, anonymized cost distributions per product category and region, which are published, while each merchant's breakdown stays confidential. (5) Context comes from the transaction and the goods (time, product category, the merchant's region), not from a profile of the customer. (6) Regional calibration uses public parameters for local tipping norms and cost of living, keyed to the merchant's location. In the design as now specified, the recommender receives no input identifying the customer in any pricing mode, and the suggested amount appears only after the customer holds the goods, shown but never pre-filled. The methodology is dedicated to the public domain under CC0 1.0 as a defensive publication, and its authors will neither seek nor assert patents on it.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Download

Share

COinS