This article is written by Nomentia
Artificial Intelligence (AI) and Machine Learning (ML) are two technologies that have been widely discussed in recent years. They offer a range of tools that are gradually being integrated into our personal and professional lives. These technologies are particularly useful in situations where there are many time-consuming and manual tasks. Or where there is a large amount of data to analyze. One such application is payment fraud detection. Before diving into the use of AI and ML in fraud detection, it is important to address them separately.
AI systems are designed to integrate vast amounts of data with intelligent algorithms. They mimic or simulate human-like actions and decision-making processes. AI can be utilized for various techniques like problem-solving, natural language processing, image recognition, reasoning, learning, and more. The technology functions in a way that collects data, processes it, and selects an algorithm. It trains a specific model, modifies the model’s parameters, and evaluates the results. This can then potentially be deployed in the real world. As it operates, AI typically uses feedback loops from its users to improve over time.
Machine learning is a form of AI that teaches a system to think in similar ways to humans, as it learns and improves based on previous experiences. Machine learning algorithms typically need little supervision because they are learning by themselves.
We usually distinguish between three techniques of machine learning:
Supervised ML algorithms are taught to make predictions or decisions primarily based on previous data patterns. Here, the ML learns based on already labeled data. So, input data needs to be categorized beforehand. So that the algorithm will use it as a benchmark to come to conclusions when analyzing new data.
In unsupervised ML, there is no need for labeled or categorized data. Instead, the algorithm tries to find patterns or relationships in the data without explicit guidance. Typically, the goal of unsupervised ML is to explore data and find clusters or relationships.
With reinforcement learning, you train the algorithms to make sequential decisions. In an environment with a reward or penalty feedback mechanism that it learns from to improve its subsequent decisions. Usually, no labeled data is available. The algorithm needs to learn from its own experiences and the feedback it receives.
AI refers to the simulation of human intelligence in machines that are programmed to mimic human-like actions and decision-making processes. AI encompasses as a wide range of techniques, including problem-solving, natural language processing, image recognition, and expert systems, among others. It aims to create systems that can perform tasks that typically require human intelligence. Tasks such as reasoning, learning, perception, and problem-solving.
Machine Learning is a subset of AI that focuses on developing algorithms. This enables computers to learn from and make predictions or decisions based on data. ML algorithms allow machines to improve their performance on a task through experience without being explicitly programmed for that task. It does so by using statistical techniques to enable computers to learn patterns and relationships from data and make decisions or predictions.
Recent research by PwC showed that 51% of organizations have experienced fraud in the past two years. A 20-year high compared to previous survey results. Fraud mainly impacted organizations in terms of financial losses. Respondents highlighted that many of them will require new, advanced technologies to tackle the issue. Some respondents also mentioned that fraud had more considerable consequences. Like operational disruptions or damage to the brand or customer loyalty. Ultimately, AI has the potential to assist businesses in maintaining a secure payment environment. Thus safeguarding a company’s customers, revenue, and reputation.
RESEARCH BY PWC SHOWED THAT 51% OF ORGANIZATIONS HAVE EXPERIENCED FRAUD IN THE PAST TWO YEARS — A 20-YEAR HIGH COMPARED TO PREVIOUS SURVEY RESULTS
PWC’S GLOBAL ECONOMIC CRIME AND FRAUD SURVEY 2022
With the rise in cybercrime and the evolving sophistication of financial threats, we’ve come to an era where humans cannot keep up with processing an abundance of data efficiently and securely. We can by no means compete with the speed and thoroughness of data interrogation that AI and ML can deliver today. As a result, we need to embrace and team up with such technologies. To support this view, a recent study by the Association of Certified Fraud Examiners revealed that 17% of organizations already leverage AI and ML to detect and prevent fraud, and 26% of organizations are actively planning to adopt fraud detection AI or ML in the next two years. On top of that, technology providers are now heavily investing in developing practical AI-driven solutions to tackle payment fraud.
These days, ML and AI can help you with fraud detection in various ways. Let’s focus on some of the main ways how organizations currently leverage the technologies and what the future may bring:
For example, AI and ML can streamline payment processes and enable faster risk identification in payables, receivables, and reporting. They can help manage exceptions or spot anomalies in large data sets based on previous patterns it has studied.
AI can analyze large datasets much faster than human beings, and it provides good insights and points to pay attention to. Faster analysis will also help speed up decision-making. Especially with more data than ever and little time to analyze, it will become essential to save time while deriving insights by leveraging tools like AI and ML.
AI can help automate essential but manual tasks such as data entry, reconciling payments, or generating reports. Minimizing manual processes, in turn, reduces the room for errors and fraud.
Reconciliation of payments is essentially comparing two data sets with each other and finding matches, which AI and ML are incredibly good at. Even when anomalies arise, you can train AI to handle them in pre-set ways.
Machine learning tools are great at uncovering complex relationships from data humans may overlook. Hence, it can provide quick and valuable insights that are potentially strategically important or can impact cash flow.
Any payment data can be screened based on benchmarking it with historical payment patterns to pinpoint out-of-the-ordinary payments faster.
In some cases, AI has been used to score payments by risk based on factors like locations, banks, sums, recipients, countries, previous behavior, and much more. This way, you can easily identify the payments with higher risk levels and make sure they’re safe.
The advantage of the learning capabilities of AI is that it can, over time, identify risks increasingly well. It allows new types of risks to be spotted that you previously may not have caught.
In a few cases, AI has been applied for account takeover detection. For example, if unusual payment behavior or account usage patterns occur, machine learning has the capability to recognize it and notify the right person or put a stop to it, limiting any further financial damage.
When integrated effectively, machine learning can analyze incoming invoices and identify out-of-the-ordinary, duplicate, mismatched amounts, or other indicators of fraud. With a large number of incoming invoices, this can help payable departments save time.
Master data is one of the biggest assets of companies, and AI can help maintain it to be in good health. For example, it can organize it by categorization, create links or connections between values, and clean up master data into more usable formats, to name a few. Better structured master data provides better insights and makes it easier to spot fraud.
You can let AI check future payments against historical payment data and establish certain tolerance levels and sensitivity for false positive management. Since dealing with false positives can be a lot of work for accounting, treasury, or finance, it can help speed up the process.
Suppose AI is connected to the internet or apps in real-time; it can deduce trends, market data, signals, and other external factors, providing insight into potential disruptions or risks at an early stage. There are various use cases for this. For example, it can help predict fraudulent attacks based on cybercrime trends, and it could even assist with considering the impact of fraud on cash flow in forecasts and help develop what-if scenario analyses.
AI is evolving rapidly, becoming more integrated with many apps and systems. In the future, it could be leveraged as a step-in-between systems for data checks, restructuring, reformatting, and other purposes to help speed up payment processes and minimize payment fraud.
AI is incredibly good at comparing data sets, which is exactly how sanctions screening works. Some payment technologies already offer sanctions screening, but AI has the potential to do so as well. This can be particularly helpful for teams that do not yet have a payment hub in place.
Perhaps in the future, payment hubs and TMS technologies will offer instructional functionalities like you’re already used to with ChatGPT, Google Bard, or Microsoft Copilot. Imagine all the questions you could ask based on all the data. Why did X payment not process? What is my exposure? How much do I have available on Y account? What caused the discrepancy between our budget and forecast?
While AI offers unprecedented opportunities for innovation and efficiency, it also raises various considerations that demand careful attention. Some of the things that we suggest you should at least consider:
Companies, banks, and financial technologies will likely incorporate new ways to leverage AI and ML technologies for fraud detection in the future. The true use of AI and ML was limited but is expected to develop rapidly. Recently, Euro finance hosted a webinar where some of the biggest companies explained how they leverage AI in treasury, finance, and accounting. It is definitely worth a watch if you are interested in how AI will shape the future beyond payment fraud.
This article is written by our partner, FIS
Receivables finance (RF) has long been a cornerstone of corporate finance, enabling businesses to convert outstanding invoices into immediate liquidity. By selling their receivables to a funder, companies may access cost-effective funding and reduce risk, while investors may gain exposure to short-duration, diversified assets at a competitive return.
However, as recent events surrounding First Brands Group illustrate, this structure remains vulnerable to risks beyond credit, such as fraud. These vulnerabilities highlight the need for greater scrutiny and the adoption of new technologies.
Fraud in RF transactions typically manifests through misrepresentation of receivables quality, double pledging of assets and fabrication of invoices. These risks arise because RF relies heavily on the integrity of the originator’s reporting and servicing processes. If invoices are falsified or pledged to multiple financiers, the asset base becomes compromised, exposing investors and lenders to significant losses.
The First Brands case underscores these vulnerabilities. The U.S. auto parts supplier, which filed for Chapter 11 reorganization in September 2025, allegedly engaged in widespread financial misconduct, including doctoring invoices and double-counting receivables to secure billions of dollars in financing.
Several market features of RF contribute to fraud risk:
These factors can make RF particularly vulnerable when governance fails or liquidity pressures incentivize aggressive accounting.
The First Brands saga has accelerated calls for digital transformation in RF oversight. Advanced technology and reporting platforms can reduce fraud risk through:
The collapse of First Brands is a cautionary tale for all stakeholders in the RF ecosystem. While RF remains a powerful liquidity tool, its resilience depends on effective governance and technological safeguards. Platforms that deliver real-time transparency, automated controls and immutable records are no longer optional: They are essential to maintaining trust and confidence in this asset.
In essence, resilient frameworks are often built on digital efficiency and the irreplaceable insight of experienced practitioners.
As institutional investors continue to seek exposure to trade finance assets and corporates aim to unlock working capital, the combination of advanced technology and human expertise within complex RF structures will help to shape the market’s future.
While digitalization may stand as the frontline defense against fraud, it’s equally vital to maintain effective human oversight by having seasoned professionals conduct independent reviews and engage in regular dialog with originators and funders.
This interplay between technological innovation and expert judgment helps ensure that not only are anomalies flagged automatically, but also the nuances of complex transactions are properly understood and addressed. In essence, resilient frameworks are often built on digital efficiency and the irreplaceable insight of experienced practitioners.
Treasury Masterminds is a community of professionals working in treasury management or those interested in learning more about various topics related to treasury management, including cash management, foreign exchange management, and payments. To register and connect with Treasury professionals, click the button below.
This article is written by Nomentia
Why are manual treasury processes expensive?
Manual treasury processes become expensive because they require recurring effort to collect balances, prepare payment files, update forecasts, check approvals, and reconcile data. Even when each task seems manageable, the combined impact can reduce efficiency, slow down decision-making, and increase operational risk.
Treasury teams are used to making imperfect systems work.
A spreadsheet here. A bank portal there. A local ERP export from one entity, a payment file from another, and a cash forecast that still depends on email updates from the business. None of these workarounds may look dramatic on their own. In many organisations, they are even seen as normal.
The problem is that “normal” can become expensive.
Manual treasury operations rarely create one large, visible cost line. Instead, they create a pattern of hidden costs: time spent collecting data, delays in decision-making, duplicated effort, payment exceptions, outdated forecasts, missed visibility, and control gaps that only become urgent when something goes wrong.
That is why treasury automation ROI should not only be discussed as a technology question. It is also an operating model question. How much time does treasury spend managing the process instead of managing cash, liquidity, payments, and risk?
The cost of manual work is often underestimated because it is distributed across people, entities, systems, and routines.
A treasury analyst may spend hours preparing a daily cash position. A regional finance team may manually upload payment files. Another person may validate bank data, check approvals, update forecasts, or investigate why one bank statement does not match the expected format.
Each task may be manageable. Combined, they create a significant operational burden.
This is also why many teams struggle to build a cash forecasting business case. The value of better forecasting is not limited to “faster reporting”. It is the value of better decisions: knowing earlier where liquidity is needed, reducing dependency on outdated data, improving confidence in funding decisions, and giving leadership a clearer view of what may happen next.
External research points in the same direction. PwC’s 2025 Global Treasury Survey notes that treasury teams are under pressure to improve cash visibility, cost efficiency, and risk management, while leading organisations increasingly adopt real-time liquidity tools, AI-enhanced forecasting, and centralised payment models. HSBC also highlights that cash flow forecasting has remained a key treasury priority, reflecting the need for precise and timely forecasts in a volatile environment.
In other words, manual treasury processes are not only inefficient. They can slow down the organisation’s ability to respond.
Cash visibility is one of the clearest examples of hidden treasury cost.
When balances are collected manually across banks, accounts, currencies, and entities, treasury may technically have the data, but not necessarily in time to act on it. The team may know yesterday’s position, but not today’s. It may have a consolidated view, but only after several people have updated files, checked bank portals, and reconciled different formats.
That delay matters.
Without timely visibility, companies may keep too much cash idle in one place while borrowing elsewhere. They may struggle to identify trapped cash. They may make liquidity decisions based on incomplete information. They may also spend valuable time explaining numbers instead of improving them.
Nomentia positions its Smart Treasury Suite around visibility, control, and predictability across payments, cash, liquidity, and risk, integrating with ERPs, banks, and other systems. For companies operating across multiple banks and entities, that integration layer is not just technical infrastructure. It is the foundation for turning fragmented data into usable treasury insight.
Payments are another area where manual processes can appear cheaper than they really are.
At first glance, uploading files through bank portals or managing payments across local workflows may seem acceptable. The team knows the process. The banks are connected somehow. Payments are executed. Work continues.
But payment operations carry a high cost when they depend on scattered portals, inconsistent approvals, manual file handling, and local exceptions.
The hidden costs include time spent preparing and checking payment files, resolving format issues, validating approvals, tracking payment statuses, and answering questions from subsidiaries, AP teams, banks, and auditors. More importantly, weak payment control can increase exposure to duplicate payments, missed cut-offs, fraud attempts, and compliance issues.
This is where payment automation benefits become easier to explain. Automation is not only about faster payment execution. It is about standardising the process, improving traceability, reducing manual intervention, and making payment control easier to prove.
Forecasting is often where manual treasury processes become most visible to leadership.
The CFO does not necessarily see how many files were collected, how many emails were sent, or how many adjustments treasury made before the forecast was ready. But the CFO does see when the forecast is late, when confidence is low, or when the numbers change without a clear explanation.
A manual cash forecast can still be useful. Many experienced treasury teams are excellent at working around incomplete data. But as the business grows, expands into new markets, adds banks, or inherits systems through acquisitions, the limits become harder to ignore.
Forecasting depends on data quality, timing, ownership, and repeatability. If treasury spends too much time gathering inputs, it has less time to analyse drivers, challenge assumptions, and model scenarios. A forecast that takes days to prepare may already be outdated when it reaches decision-makers.
This is why the business case for treasury automation should include both time savings and decision quality. Faster data collection is valuable. But the larger value often comes from giving treasury more time to interpret what the numbers mean.
Manual controls are often built around expertise. The team knows which approvals are needed, which files need checking, which bank deadlines matter, and which exceptions require escalation.
That works until complexity increases.
As more entities, banks, users, and payment types are added, control becomes harder to manage consistently. Processes may differ across countries. Approval rules sit outside the system. Audit trails may require manual reconstruction. Exceptions depend on individual knowledge rather than embedded workflows.
In a stable environment, this may go unnoticed. During growth, restructuring, audit, staff changes, or periods of financial pressure, it becomes a risk.
The Nomentia Treasury Trends Report 2026 describes treasury teams facing pressure to deliver real-time insights, stronger controls, and more strategic input, often while dealing with fragmented systems and limited IT support. The report is based on 384 treasury and finance leaders across the Nordics, DACH, Benelux, and the UK.
That is the reality many treasury teams recognise: expectations are rising faster than operational capacity.
A strong treasury automation ROI discussion should not begin with software features. It should begin with operational impact.
From there, TMS cost savings become easier to frame. The value may come from fewer manual hours, lower operational risk, more efficient payment execution, improved cash visibility, reduced dependency on spreadsheets, or stronger audit readiness.
The most useful business case is not a generic promise that automation saves money. It is a structured estimate of where the organisation currently loses time and where better treasury processes could create measurable improvement.
Treasury Masterminds is a community of professionals working in treasury management or those interested in learning more about various topics related to treasury management, including cash management, foreign exchange management, and payments. To register and connect with Treasury professionals, click the button below.
This article is written by Cobase
For an industry built on numbers, banking has always struggled with something more basic: speaking the same language.
Ask any treasury team trying to connect to banks globally and you’ll hear a familiar frustration. The expectation is simple – money is digital, banks are global, so connectivity should be straightforward. In reality, it rarely is. What looks like a plumbing issue is something deeper: a system that was never designed to be unified in the first place.
Modern banking didn’t emerge as a coordinated network. It grew in fragments. National systems were built to serve domestic economies, shaped by local regulation, infrastructure, and political priorities. Payment schemes evolved independently. Messaging formats were defined in isolation. Even basic concepts like how to confirm a payment or report a balance took different forms depending on where you looked.
The result is not just variation, but incompatibility.
SWIFT is often held up as the closest thing to a global standard. And in one sense, it is. It created a common messaging layer that banks across the world could use. But it never standardised what happens after the message is sent. Two banks can receive the same SWIFT instruction and process it in entirely different ways – different cut-off times, different validations, different interpretations.
This is where the idea of “bank connectivity” begins to unravel. The challenge is not just reaching a bank, but dealing with how each bank behaves once you do.
Over the years, the industry has made repeated attempts to smooth this out. None have fully succeeded. Not because the technology wasn’t good enough, but because the incentives never aligned. Banks compete. Regulators don’t coordinate globally. And legacy systems – often decades old – continue to run critical infrastructure that no one is willing to replace lightly.
The expectation of a unified system persists. But it’s built on a false premise.
Banking isn’t fragmented because something went wrong. It’s fragmented because that’s how it was built.
Few ideas in banking have generated as much optimism in recent years as APIs.
They arrived with the promise of simplicity. Clean, modern interfaces. Real-time data. Standardised access. Compared to the heavy, file-based integrations of the past, APIs looked like a reset moment, a chance to finally make bank connectivity behave like the rest of the digital world.
And in some ways, they delivered.
Large banks began exposing endpoints for payments and reporting. Developers could interact with bank systems without navigating layers of legacy protocols. In controlled environments, things worked exactly as advertised.
But step outside those environments, and the picture changes.
APIs in banking are not a single standard. They are dozens, sometimes hundreds, of individual implementations. Each bank defines its own structure, its own authentication methods, its own limits. Even when two banks claim to follow the same framework, the differences show up quickly – in edge cases, in error handling, in performance under load.
The regulatory push behind open banking added momentum, but also confusion. PSD2 created a baseline, but it was never designed for corporate treasury. It focused on retail use cases, with limited scope for bulk payments, complex approval flows, or multi-entity structures. For large organisations, it solved a small part of a much bigger problem.
Meanwhile, neo-banks and aggregators entered the picture, offering simplified access and faster onboarding. They improved the experience at the edges, particularly for account opening and basic transactions. But they didn’t remove the need to engage with traditional banks. In many cases, they simply added another layer to manage.
The result is a familiar pattern in financial infrastructure. New technology doesn’t replace the old – it accumulates around it.
APIs didn’t eliminate fragmentation. They made it more dynamic.
From the outside, bank connectivity looks deceptively simple. Payments go out, balances come in, and everything appears to move through a single system.
What’s less visible is the machinery underneath.
For companies operating across multiple countries, connectivity is not one connection, it’s dozens. Each bank brings its own requirements. File formats differ. Security models vary. Some require certificates, others tokens. One bank processes payments in batches, another in real time. Cut-off times shift by region, sometimes by product.
Even within the same bank, behaviour can change depending on the channel used. An API might support one set of payment types, while host-to-host supports another. Documentation doesn’t always reflect reality. Test environments behave differently from production. Exceptions are handled inconsistently.
None of this is unusual. It’s the normal state of the system.
This is why, despite all the talk of innovation, older methods remain firmly in place. Host-to-host connectivity – direct, file-based integration – continues to handle a large share of corporate payments. It’s not elegant, but it’s predictable. It does what it’s supposed to do, at scale, without surprises.
In certain markets, local standards dominate. EBICS, for example, is deeply embedded in parts of Europe. It works not because it’s globally relevant, but because it reflects the specific needs of those markets. In those contexts, it often outperforms more “modern” approaches simply by being consistent.
And then there’s SWIFT, still acting as the global fallback. When no direct connection is available, SWIFT is usually there. Not perfect, not always efficient, but broadly accepted.
Put all of this together, and a pattern emerges. There is no single best way to connect to banks. There is only a set of trade-offs.
The real work is not choosing one method, but managing all of them at once, and making them behave as if they were one.
That work increasingly sits in a layer most corporates never set out to build, but inevitably do: an orchestration layer that absorbs differences between banks, channels, and formats, and presents something coherent on top.
This is where platforms like Cobase operate.
Rather than trying to standardise banks themselves, Cobase standardises the interaction with them. It connects across SWIFT, EBICS, APIs, and host-to-host channels, translating between formats, normalising data, and embedding bank-specific behaviour into a central system. A payment instruction created once can be converted automatically into whatever each bank requires. Data coming back – balances, statuses, confirmations – is aligned into a consistent structure.
The complexity doesn’t disappear. It is relocated.
Instead of sitting in day-to-day treasury operations spread across teams, spreadsheets, and manual fixes, it is contained within a controlled layer designed to handle it.
Because in the end, the hardest part of bank connectivity is not building connections.
It’s making them invisible.
Treasury Masterminds is a community of professionals working in treasury management or those interested in learning more about various topics related to treasury management, including cash management, foreign exchange management, and payments. To register and connect with Treasury professionals, click the button below.
