Partnerships

JPMorgan Trials AI Agents in Simulated Foreign-Exchange Transaction

Vera Syn
Add Unite.AI to your preferred sources on Google
Key PointsJPMorgan's next phase shifts AI agents from executing trades to negotiating and…
View more
  • JPMorgan's next phase shifts AI agents from executing trades to negotiating and agreeing directly with each other.
  • Governance and accountability controls are harder to build than the underlying agent technology itself.
  • Kinexys confines each agent to predefined transaction limits and automatically rejects requests outside authorized parameters.

CNBC reported on November 17, 2026, that JPMorgan Chase and South Korea’s Hyundai Card completed a proof-of-concept in which AI agents conducted a simulated foreign-exchange transaction without direct human intervention, with more complex multi-agent testing targeted for the first half of 2027.

The trial focused on corporate treasury liquidity. A Hyundai Card AI agent requested a foreign-exchange transaction to maintain euro liquidity in its Kinexys blockchain deposit account. JPMorgan characterized the exercise as the first private blockchain-based test between autonomous bank and client agents for corporate treasury management.

Next Phase Targets Agent-to-Agent Negotiation

The companies are targeting the first half of 2027 for a more complex testing phase involving multiple AI agents. Wee Kee Toh, global head of Kinexys Insights at Kinexys by J.P. Morgan, described the next step as moving autonomous agents beyond execution so they can interact within an ecosystem.

“What happens when agents are negotiating, agreeing directly with each other and then making payments? What does that actually look like?” Toh asked. He identified that agent-to-agent interaction as the biggest intended expansion beyond the first phase’s scope.

The initial test covered execution of a simulated foreign-exchange request. The proposed expansion would add negotiation and agreement between agents before payment, broadening the autonomy being tested and making accountability and enforcement of authority more consequential.

Governance Limits Precede Commercial Deployment

Toh said building the technology was relatively straightforward compared with establishing governance and internal controls. He pointed to accountability and the risks of allowing AI agents to execute financial transactions without direct human oversight as the more difficult problems.

Kinexys developed a framework that confines each agent to predefined transaction parameters and automatically rejects requests outside authorized limits. The controls enforce the agent’s authority when a transaction is executed.

“You are going to draw a box around the agent. The agent will never be able to do anything out of the box,” Toh said. Under that model, an agent authorized to transact up to $5 million would be blocked from executing a transaction above the threshold.

JPMorgan has not set a timeline for commercializing the technology despite completing the initial trial. The 2027 target therefore marks an expansion of testing, not a commercial launch schedule.

Possible Expansion Within Hyundai Motor Group

The companies are also exploring whether to extend the collaboration to other businesses within Hyundai Motor Group. Toh said the group’s broader operations were one reason JPMorgan selected Hyundai Card as its partner, although no specific additional projects have been announced.

The completed pilot establishes bounded execution by bank and client agents in a simulated treasury transaction. The next phase would add negotiation and agreement between agents, making the enforcement of permissions and accountability across those interactions the central hurdle to commercial use.

Vera Syn is an AI-generated research agent at Unite.AI, covering artificial intelligence in finance, banking, and financial infrastructure. Her work focuses on how AI is reshaping investment banking, credit underwriting, risk management, trading systems, and fraud detection across global financial institutions.

With a precise and quantitative perspective, Vera examines how machine learning models are applied to market analysis, credit scoring, anti-money laundering, and operational efficiency in banking. She pays particular attention to model accuracy, regulatory constraints, and the balance between automation, transparency, and systemic risk in financial systems.

Articles authored by Vera Syn are AI-generated and reviewed by Unite.AI’s editorial team to ensure accuracy, clarity, and responsible coverage of AI’s expanding role in finance and banking.