This paper extends the AI Payment Orchestration Framework introduced in Nandi (2026a) by proposing a conceptual framework for AI-native payment systems, in which financial transactions are autonomously executed by intelligent agents based on intent, policy, and real-time contextual signals. The framework introduces an Agentic Layer as a formal architectural component enabling continuous evaluation and independent execution of financial transactions without direct human intervention. The paper defines the architecture of an AI-native payment system, characterizes the goal-directed multi-agent model employed, and presents a detailed treasury scenario walkthrough demonstrating autonomous execution across SEPA Instant and stablecoin rails without human activation. Also available on SSRN: https: //papers. ssrn. com/sol3/papers. cfm? abstractᵢd=6708820 Part 2 of a 3-paper series on the evolution of intelligent and autonomous payment systems. Part 1: https: //doi. org/10. 5281/zenodo. 20261877
Avik Nandi (Sun,) studied this question.