Data-Led Decision-Making With AI and ML in Treasury Management

Of the 5.5 million small businesses in the UK, only 20% use artificial intelligence (AI) in their work. Within the proper framework, small businesses can embrace AI and machine learning (ML) at pace due to their agility and ability to adapt the way they operate, quickly.

In the treasury function of these businesses, for example, AI and ML can transform the tools used to manage cash, liquidity, and risk. The change is the equivalent of Industrial Revolution 4.0 for treasurers. Those who ignore this technology risk falling behind.

The problem is that many treasury teams – both small and large – still use spreadsheets. This is worrying because an estimated 9 out of 10 spreadsheets contain errors. Such manual systems are slow and error-prone. They also prevent treasurers from keeping up with new industry standards, and accessing AI’s business benefits.

About 55% of small firms understand AI and ML’s benefits. However, 46% of these companies think they lack the knowledge to adopt AI. Instead, they assume a data scientist or ML expert will need to be employed to implement the technology. This isn’t true.

ML algorithms are made for use by companies of all types and sizes. The model adapts to a business’ specific characteristics. By embracing these technologies, smaller businesses can enhance their financial stability, streamline operations, and position themselves for growth.

The fuel for any ML tool is the data it is trained with and operates on. ML models bring together large amounts of the data underpinning cash and liquidity management practices within seconds. As a result, leaner and smaller treasury teams have more time and scope to convert their data into valuable insights.

Automation made better

AI can automate repetitive and time-consuming tasks. This allows treasury professionals to focus on more strategic parts of their jobs. For small businesses, AI and ML can help in two distinct ways: making better predictions of future liquidity and optimising actions based on those predictions.

Historically, liquidity planning and cash forecasting processes were manual and time-consuming. Successful forecasting helps small businesses to minimise excess liquidity and make the best use of the surplus in various ways. This is why it is so important to process those predictions accurately.

By analysing past cash flow data from the business’ Treasury Management System (TMS) and Enterprise Resource Planning (ERP) system, AI can train the algorithm, and measure the confidence level of the output predictions. This also prevents mistakes which could affect the decision-making process.

AI can analyse huge amounts of data far quicker than ever possible with a human brain.

Cash forecasting

AI and ML can spot deviations from standard events, or recurring patterns, making them a great tool for cash forecasting. This ability allows small businesses to see the bigger picture – potentially for the first time – as ML can detect wider trends and anomalies. Therefore, treasurers can make better predictions.

Businesses can enrich the data on which the ML models are trained, produce forecasts for the different categories of cash they use every day, and produce a six or twelve-month forecast. Using ML techniques, a business can generate forecasts 3,000 times faster than manual processing. For small businesses, this financial planning provides a blueprint for future growth and resilience. Running a small business can be unpredictable, but accurate long-term forecasts signal the best growth prospects.

Bank reconciliation

While bank account reconciliation may already be automated, it remains a time-consuming process. This is because manual intervention is still needed to match statements with pre-booked transactions, or correct cash flows.

AI can help here as well. By training these models on millions of SWIFT/BAI2 messages, AI can efficiently classify transactions into different tags. The same applies to matching statement items to pre-booked transactions. This includes cases where there are value date or amount differences, or where there are many-to-many matches.

Treasurers can experience impressive efficiency increases by using these tools. Previously repetitive daily tasks are now just quick checks to make sure everything is in order.

Payment anomaly detection

No matter the business size, security is currently top priority, especially with the rise of cybercrime. Treasury teams are ideal targets for attack, as they oversee areas such as cash management, connectivity and payment management and processing.

AI tools can help prevent fraud before it happens. As it analyses the flow of payments, the AI begins to learn what the treasury function’s ‘normal’ behavior. All new payments are then screened, and any deviation from that learned behavior – whether fraud or user errors – can be detected in real time. It allows small businesses to make their operations safer, without any extra legwork.

This technology is for everyone, no matter the size

AI isn’t just for big companies. Small businesses can benefit too.

Any businesses hesitant to adopt AI can start by running their existing cash forecasting processes in parallel with ML and compare the results. This will show the real benefits offered by ML.

Small businesses that adopt AI and ML into their treasury operations will benefit from enhanced forecasting and decision-making, better pattern recognition, and predictive analytics.

AI also offers significant benefits in detecting payment outliers by quickly identifying anomalies in transactions that could indicate errors or fraud. It also streamlines the bank reconciliation process by automatically matching transactions and reducing manual errors.

Through AI, a business will become more proactive and capable of navigating complex financial landscapes in real time.

Ultimately, AI is a journey. There is no expectation that smaller businesses will suddenly shift all their cash forecasting and apply only ML techniques. But it’s important for companies to understand their data, trust in AI, and recognise the benefits it can offer.

About the author

Viola Hechl-Schmied is Product Owner for Machine Learning at ION Treasury - a leading provider of treasury and risk management solutions for companies of any size, complexity, geographic reach, budget, or IT capability. ION’s treasury expertise enables customers to select the right solution for their unique needs and automation journey.

AI and ML in Treasury Management

Learn how AI and machine learning can improve cash management, forecasting, and fraud detection.