As treasurers explore the use of agentic AI in financial transactions, one global bank urges them to envision the flow of funds five-to-six years from now.
J.P. Morgan’s payments team has issued a case study demonstrating accountability issues and solutions arising when agents are permitted to operate, even within pre-approved parameters.
“The treasury of 2032 is not a room full of people moving money,” writes Zack Anderson, Chief Data & Analytics Officer for Payments and Global Banking, J.P. Morgan. “It is a nervous system – sensing, adapting, acting and constantly asking itself: ‘Am I still within bounds?’”
J.P. Morgan’s example is outlined in a June whitepaper entitled ‘Proof of Movement: AI, Autonomy and the New Architecture of Corporate Cash’. The scenario involves overnight actions on a payment exceeding US$1m.
The supplier unexpectedly switches the invoicing currency. It is a “a small change buried in a payment file that, left unaddressed, would have created a US$2.3m foreign exchange mismatch over the quarter,” the J.P. Morgan document explains. “The system reclassified the exposure, proposed a 90-day rolling hedge, priced it against three counterparty quotes sourced through an Application Programming Interface (API), and queued the trade for human approval.”
The treasurer wakes up, opens a work laptop and “the exception is waiting: a two-sentence summary, a confidence interval, an audit trail.” It takes just nine seconds to approve the action.
However, the screen then indicates an “invoice anomaly”. It tracks evidence of the agent’s actions and reasoning including “the purchase order, a matched receipt, prior counterparty behaviour, and an explanation panel showing why the model flagged this as unusual,” J.P. Morgan writes.
Simultaneously, the Chief Financial Officer’s screen is flashing a cautionary alert: a 12% probability of a debt covenant breach.
“The AI recommends drawing down a revolving credit facility. The CFO pauses. The model’s working-capital assumptions look stale. They flag the scenario for the weekly risk committee,” J.P. Morgan notes.
“The treasurer’s nine-second approval was possible because the hedge fell within pre-authorized policy bands,” the report explains. “The CFO’s hesitation was warranted because the twin’s recommendation sat outside them. The line between the two – between what the machine can do alone and what it must not – is the defining design problem of the next decade in corporate finance.”
Fiat and digital liquidity
That transaction implied only traditional currencies. The New York-based bank mentions that fiat and digital currencies can now be managed within one platform.
In April, San Francisco-based fintech Ripple launched Digital Asset Accounts and Unified Treasury, embedding crypto directly into a treasury management system.
“CFOs and their treasury teams can now view, hold, receive and manage fiat and digital liquidity held within their bank and custody providers in a single system,” Ripple promised in a press release.
“The design principle behind both capabilities is that digital assets should behave exactly like cash within the platform,” added Mark Johnson, Vice President of Global Product at Ripple Treasury.
With that, J.P. Morgan says “CFOs can now view, hold and manage both fiat and digital liquidity – including stablecoins – within a single platform … a preview of what ‘seeing the whole board’ means when the board includes tokenised assets.”
However, there is a risk that deployment of the fintech solutions “hits a wall” in terms of breadth. The constraint is “that risk, liquidity and exposure data is spread across ERPs, TMS platforms, bank portals and business units – limiting what agentic AI can do responsibly without deep data unification,” the J.P. Morgan report warns.
Competitive advantage
Reflecting on the examples outlined earlier – the treasurer’s nine-second approval and the CFO’s hesitation – J.P. Morgan describes treasury “not as a team in a room but as a cyber-physical system for money. APIs are the sensors [that] ingest balances, transactions and market signals continuously. Payment rails and FX engines are the actuators [that] route, schedule, net and execute.”
The invoice scenario’s nine-second approval was enabled partially by the agentic summary. However, “the future of treasury is not a chatbot. It is not a dashboard. It is a control system – and the competitive advantage belongs to organisations that integrate all five stages (sense, predict, decide, execute, audit) into a single, governed, auditable system before their competitors do,” J.P. Morgan explains.
“The winners won’t be the firms with the most AI. They’ll be the ones with the most governable AI.”
