Hyundai Card and Kinexys test AI agent FX payments in proof of concept
For the head of treasury services at a regional bank
Hyundai Card said on Nov. 18 that it completed a proof of concept with Kinexys by J.P. Morgan in which AI agents handled a simulated foreign exchange conversion, according to Finextra's account of Hyundai Card's announcement.3
Hyundai Card initiated the test to explore emerging payment technologies, according to the Finextra report; J.P. Morgan said the phased proof of concept aims to set the trust framework needed for future agent-to-agent transactions.13 According to J.P. Morgan's newsroom post, the Hyundai Card AI agent monitored the balance of Hyundai Card's Kinexys Blockchain Deposit Account in Frankfurt, identified a Euro shortfall and determined that a Sterling Pound to Euro FX conversion was required, then requested FX quotes and engaged Kinexys for execution.1
J.P. Morgan said a Kinexys AI agent verified the Hyundai Card agent's identity and authority and checked the request against a digitally signed mandate and applicable controls; it could auto-execute, escalate to a human, or deny the request.1
The Kinexys agent auto-executed the request and a simulated FX transaction was booked, and J.P. Morgan said the complete reasoning and decision data was captured in an immutable audit trail.1
Phase one ran in a test environment with no live value movement, J.P. Morgan said.1
Hyundai Card designed the scenario and the authorization and control framework, while Kinexys built the agents for both organizations, according to Finextra; that report describes the scenario as transfers between hypothetical overseas subsidiaries, a framing J.P. Morgan's post does not use.13
The test shows how a bank can let an agent act on a client's behalf within signed limits, with escalation to a person as the exception path. Banks and treasury vendors weighing agent-driven payments should plan who staffs and answers those escalations, and how mandates and audit trails will be reviewed. The trade was simulated, so it does not show how this works with live funds.
Our take. When a corporate client asks whether its agent can instruct your rails, JPMorgan has written the answer sheet: verified identity, a signed mandate, a human for exceptions. Its own test moved no real money, so the answer is untested.
Action. Treasury services: Define the identity check, signed-mandate limits and human escalation path required before accepting payment instructions from client AI agents.
What our records say
JPMorgan runs LLM Suite, a firm-wide generative-AI platform, with almost 1,000 AI use cases and a technology budget of about $18 billion in 2025 and nearly $20 billion in 2026.12 JPMorgan Chase's proprietary generative-AI platform, LLM Suite, has been in the hands of most of its workforce since 2024, and its technology budget was about $18 billion in 2025 and nearly $20 billion in 2026.12
From our records
- JPMorgan Chasebank · latest Card-number block reported for outside AI agents (Nov 2026) · vendors Mitek, Kasisto, ChatGPT Enterprise · enforcement 5 actions since 2016; latest CFPB civil action (lawsuit, dismissed with prejudice) (Dec 2024)
Sources
- Primary sourceHyundai Card and Kinexys agentic payments proof of concept Newsroom Post JPMorgan
- Also reported byHyundai Card and Kinexys test AI agents for FX and fund transfers Thepaypers
- Also reported byHyundai Card validates agent-powered transaction model with Kinexys by JP Morgan Finextra