Abstract
A system can generate context-aware payment recommendations to assist users in selecting a suitable payment method, which may be beneficial given the complexity of financial product benefits. The system may employ a multi-stage processing pipeline where, for example, an intent classifier can first infer a user's need from a transactional context. A request may then be processed by a handler that can query a retrieval-augmented generation system to obtain relevant data from a knowledge base. A large language model may synthesize this data to generate a structured output, such as a JSON payload containing both human-readable text and machine-readable user interface directives. The structured JSON output is designed to bridge the gap between unstructured natural language and platform-specific interactive UI components. This process can enable a client application, such as one on a smartphone or wearable device, to dynamically present payment guidance to a user, for example, within a checkout experience.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Luo, Zeyu; Dayanand, Dinoop; Chiu, Adam; Xia, Summer; Katz, Aimee; and Patel, Kushagra, "System for Context-Aware Payment Recommendations Using Retrieval-Augmented Generation", Technical Disclosure Commons, (January 12, 2026)
https://www.tdcommons.org/dpubs_series/9174