Corporate treasurers expect programmatic foreign exchange execution to account for 42% of their trading volume by 2031, up from 18% today, according to a new Integral survey of 143 corporate treasuries, banks and other FX market participants. The forecast points to a change in where corporate FX trades begin: inside enterprise resource planning and treasury management systems rather than on a dealer screen, multi-dealer platform or telephone call.
The shift is substantial, but it is not a forecast that manual execution will disappear. Voice and multi-dealer platforms each account for 34% of current corporate trading volume in Integral’s results, giving the two channels a combined 68% share. Even if respondents’ five-year expectations are met, 58% of volume would remain outside the two programmatic categories. The result therefore describes a redistribution of routine flow, while complex trades, exceptions and relationship-based execution can continue through established channels.
Integral’s study divides the expected programmatic share into user-triggered API execution, rising from 6% to 16%, and fully embedded execution within ERP and TMS software, rising from 12% to 26%. That second category matters most for banks. It places the corporate system at the front of the transaction, but banks can still supply the price, credit line, liquidity, execution, controls and settlement process behind it.
Bank APIs Could Become the New Corporate FX Front End
More than half of the banks in the survey expect corporate API trading to become the most important FX execution channel within five years. More than 80% said prices delivered directly into client systems, including order and treasury management platforms, will play a significantly greater role in their distribution strategies.
For a treasurer, embedded execution can connect an exposure identified in an ERP, a hedge required under treasury policy, an executable bank price and the resulting accounting record. The objective is fewer handoffs between exposure data, a spreadsheet, a trading venue and a confirmation system. A bank does not need to own the interface to remain central to that chain. It needs APIs that can quote reliably, enforce permissions and limits, return execution status and feed downstream records without creating another manual reconciliation task.
There are already working versions of this model. A Corpay and SKsoft integration embeds cross-border payments and FX exposure management in Microsoft Dynamics 365. In banking, Bank of Cyprus uses Integral APIs for pricing and risk services within its online banking platform. These are narrower deployments than an autonomous treasury operation, but they show that the execution interface can move into software already used by a client while a regulated financial institution remains the service provider.
The commercial risk for banks is less about being removed from the trade than becoming interchangeable behind it. When a TMS can request prices from several providers through standardized connections, screen familiarity and manual sales coverage carry less weight for repeatable flow. Price quality, uptime, credit availability, response time, product coverage and post-trade reliability become easier to compare. Banks that cannot expose those capabilities through mature APIs may lose distribution even if their traditional electronic trading service remains competitive.
Integration Is a Bigger Immediate Barrier Than AI
The survey’s adoption barriers temper its five-year forecast. Some 83% of corporate respondents identified at least one obstacle that could slow embedded FX. Internal system integration was the leading hurdle for large companies, cited by 56%, while 43% of mid-sized and growth companies identified bank API maturity as their biggest challenge.
Those figures describe two ends of the same connection. Large companies often have several ERP instances, regional treasury processes, acquired subsidiaries, multiple chart-of-account structures and inconsistent exposure data. A bank may offer an API, but the corporate still has to determine which exposures are eligible for automation, how they are netted, when a hedge is generated and which entity has authority to trade. Smaller companies may have a cleaner internal setup but fewer banking providers able to support the required product, documentation and integration work at an economic cost.
Migration risk also appears elsewhere in institutional FX technology. A recent Trading Technologies and Acuiti survey found that 69% of buy-side firms wanted a unified view of FX risk, yet almost half cited migration risk as the main reason they might hesitate to consolidate systems. The users in that research were asset managers, hedge funds and proprietary firms rather than corporate treasuries, but the operational lesson is similar. Connecting systems can remove fragmentation after implementation while creating testing, control and data-mapping risks during the transition.
Citi’s 2025 FX Vendor Review provides another indication that the demand is broader than a single technology-provider survey. Citi said 85% of surveyed clients still requested execution or workflow enhancements despite 90% reporting satisfaction with their primary vendors. Vendor switching fell from 51% in 2021 to 22% in 2025, suggesting that integration can deepen an existing relationship, but also that replacing infrastructure is costly enough to slow change.
Agentic AI Depends on Rules, Data and Exception Handling
Only 8% of the corporate respondents are currently piloting AI projects, while half expect AI-driven agents to manage 25% of their FX workflow within five years. The distance between those numbers makes the AI result an ambition rather than evidence of scaled adoption.
An agent can monitor exposures, compare them with policy limits, request prices and recommend or initiate a hedge. It still needs reliable source data, approved counterparties, instrument rules, credit checks, execution limits and a clear route for human escalation. A system that receives duplicated invoices or misclassifies a forecast exposure can automate the wrong hedge more quickly. The control design therefore has to cover the decision that precedes execution, not only the API message that sends an order.
Automation also changes the evidence a treasury team and its bank must retain. The December 2024 FX Global Code covers governance, execution, information sharing, risk management, compliance, confirmation and settlement. It is a voluntary set of good-practice principles rather than regulation, but its emphasis on resilient infrastructure and transparent conduct becomes more relevant when an automated workflow makes decisions at machine speed. Treasurers will need records showing the exposure, policy rule, price sources, authorization, execution result and any human override.
This is also why AI is unlikely to eliminate voice dealing. A routine hedge generated from a confirmed payable is a plausible candidate for rules-based execution. A large, illiquid or unusual transaction may require market color, staged execution or a discussion about liquidity and timing. Voice can become the exception channel while APIs absorb repetitive flow, leaving dealers and treasury staff to concentrate on decisions where judgment adds more value.
Embedded Execution Could Give Treasurers More Time for Risk
Nearly two-thirds of corporate respondents said largely automated and embedded FX execution would allow them to give greater priority to risk-management strategy. That is the strongest potential benefit in the study because execution is only one part of corporate currency management. Treasury teams also have to improve cash-flow forecasts, decide hedge ratios and tenors, assess natural offsets, manage counterparty exposure and explain results to management.
Removing rekeying and routine order handling can create time for those tasks, but the benefit depends on the quality of the workflow. An automated trade followed by a manual confirmation, allocation or accounting process simply moves the bottleneck. The higher-value model connects exposure capture, policy, execution and post-trade records so that exceptions receive attention without every transaction requiring intervention.
Integral’s own bank deployments show the buildout occurring in layers. Banco de Chile connected pricing and liquidity aggregation across dealer platforms, sales teams and client portals, while Banco Base replaced separate pricing processes with an integrated setup for FX derivatives and swaps. Integral has also expanded the underlying capacity, with its Singapore deployment processing more than one million FX tickets a day. These examples establish that the pricing and execution infrastructure exists, although they do not prove that corporate demand will reach the survey’s 42% projection.
The market is large enough for incremental changes in corporate behavior to matter. The Bank for International Settlements measured average global FX turnover at $9.6 trillion a day in April 2025, including $4 trillion in FX swaps. Corporates account for only part of that total, but swaps and forwards are core hedging instruments, and their workflows involve more than obtaining a spot price.
The 42% Figure Is a Directional Expectation, Not a Forecast
The public report page says the findings are based on 143 respondents globally across corporate treasuries, banks and other FX market participants. It does not publish the number in each respondent group, the regional mix or the survey field dates. That limits how precisely the percentages can be generalized, especially when answers from banks describe distribution priorities while answers from corporates describe trading behavior and operational barriers.
The time horizon also asks respondents to estimate technology adoption five years ahead. Budgets, regulation, market volatility and the performance of early projects can change those plans. The projected rise from 18% to 42% is best read as evidence of intent and investment direction, not as a measured market forecast.
Harpal Sandhu, Integral’s chief executive, said the transition depends on “strong technology foundations, including API first platforms.” The survey data support that sequence. Corporate users are asking for more execution inside their own systems, banks see APIs becoming a primary distribution route and both sides identify integration as the main near-term obstacle.
The next competitive contest in corporate FX may therefore occur behind the screen. ERP and TMS providers can own more of the user workflow, while banks compete to supply the pricing, balance sheet, controls and execution embedded within it. AI may eventually manage part of that chain, but the institutions that solve connectivity, data quality and governance first will determine how much of the projected 42% becomes real trading volume.
