Abstract
The present disclosure provides systems and methods for providing payment transaction information to a large language model (LLM)-based agent. A payment transaction history of a customer is obtained, the payment transaction history including a plurality of transaction records. A graded information-access ladder having a plurality of levels is defined, each level representing a different degree of data processing and abstraction of the payment transaction history. Customer behavioral information derived from the payment transaction history is served to the LLM-based agent through a selected level of the graded information-access ladder. The levels may include a no-signal level, a raw level, a query level providing a retrieval tool for filtered transaction rows, an aggregate level providing a provider-computed aggregate bundle, and a signal level providing pre-inferred prediction outputs as callable tools.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Liang, Yueqing; FAN, XIRAN; Zheng, Yan; FAN, Yujie; Wang, Liang; and Cai, Yiwei, "GRADED PAYMENT TRANSACTION INFORMATION INTERFACES FOR AI AGENTS", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/12471