Artificial intelligence agents have proven to be helpful at times, but they can also cause a lot of trouble at other times, or sometimes both. Payments are considered a very promising area; if these agents can be trusted, they could handle various financial transactions on behalf of people.
This is also why JPMorgan Chase and Hyundai Card of Korea are testing how AI agents can handle financial transactions without direct human intervention, including whether they will be able to negotiate with each other in the future and complete payments.
The two companies have completed a preliminary proof of concept involving a simulated foreign exchange transaction, and plan to move on to a more complex phase in the first half of 2027, which will involve multiple AI agents.
The Wee Kee Toh of Kinexys Insights, the global director, stated to CNBC that the next step is to test what will happen when autonomous agents go beyond the execution level and begin to interact with each other within an ecosystem.
"I would say that this is indeed the largest part of what we hope to expand upon from the scope of the first phase," he said.
In this initial trial, an agent of Modern Card with the identifier AI initiated a foreign exchange transaction request to maintain the liquidity of euros in its blockchain deposit account with the identifier Kinexys. JPMorgan indicates that this is the first autonomous test between a bank agent and a customer agent based on a private blockchain, aimed at corporate financial management.
Beyond Pilot Projects
Although the initial trials have been completed, JPMorgan has not yet set a timeline for the commercialization of this technology.
Toh indicates that the construction of the technology itself is relatively straightforward, but establishing appropriate governance and internal controls is much more difficult.
"From a technical perspective, this is not the biggest obstacle," he said, pointing out that the real issues lie in the attribution of responsibility and the risks associated with allowing AI to execute financial transactions without direct human supervision.
To address these concerns, Kinexys has developed a framework that limits AI proxies within pre-set transaction parameters. Requests that exceed the authorized range will be automatically rejected.
“You need to draw a frame around the agent. The agent can never do anything outside of that frame,” said Toh.
For example, he stated that if a certain AI agent is authorized with a maximum trading amount of 5 million US dollars, then any transactions that exceed this limit will be prevented from being executed.
Toh said that the company is still exploring whether to expand the cooperation to other enterprises under Hyundai Motor Group, and Hyundai Motor Group's broader range of businesses is also one of the reasons why JPMorgan chose Hyundai Card as a partner. No specific projects have been announced yet.
According to the website of the card-issuing institution, Hyundai Motor Group holds a total of 78.04% of the shares in Hyundai Card through Hyundai Motor, Kia, and Hyundai Commercial.
Modern Card has previously attempted blockchain-based cross-border payments. In July of this year, the company stated that it had completed a proof of concept, using the USDT stablecoin to transfer $20,000 between Modern Auto’s subsidiaries in the United States and Mexico, with the transfer and verification process taking an average of 7 minutes.
Kinexys has also been expanding its blockchain-based payment services. In August, the department under JPMorgan stated that since its establishment, it has processed over $4 trillion in transactions, with an average of about $7 billion per day.
In addition, JPMorgan launched the proof of concept for its dollar-denominated deposit token, JPMD, in June 2025, and made it available to institutional clients in November of the same year.











