Improving cash management with AI

It is no longer breaking news that AI and automation tools are increasingly finding their way into finance departments. In earlier articles, such as "How automation and KI will propel Treasury forward" (Edition 136), we have already given a detailed definition and an overview of the use of various AI and automation tools. This time, we will focus on specific system-based AI and automation options that help optimize cash management.

Besides boosting efficiency, these technologies also minimize errors and ensure consistent and transparent processing. But at the same time, security risks must be considered and mitigated.

In this article, you will learn how your company can optimize its cash management by making targeted use of these technologies and identifying the security measures needed to safely take advantage of AI technology.

Automation based on workflows offers considerable optimization potential in cash management by streamlining manual and time-consuming processes. A range of tools make it possible to automate and standardize complex workflows, leading to a significant improvement in efficiency and accuracy.

Take the process of manual payment transactions, for example, which traditionally involves several manual steps: First one person creates the payment, another checks it technically, and a third and fourth person approve the payment. There are usually several system breaks and sometimes paper-based intermediate steps behind these process steps. There may be a need for manual payment due to special tax payments (without a payment format), HR payments or other payments that are not mapped in the standard process. In many cases, paper signatures are also required to ensure the process is “compliant”. This results in a high use of resources and prevents optimal mapping in the treasury management system.

By using workflow-based automation, all these steps can be carried out seamlessly and efficiently. Using special tools, for example, a payment request is created and automatically forwarded to the next processor. Each processor is notified when a task (e.g. approval) is due and the system tracks the progress of each task. This reduces processing time and minimizes the risk of human error. By using a workflow tool, end-to-end processes can become automated. Sophisticated and integrated treasury management systems (TMS) are excellent at performing these tasks. This automates the entire payment process from creation to release and ensures that all necessary steps are followed. The result is a standardized process and therefore considerable time savings.

By implementing workflow-based automation in cash management, a company will optimize its processes and increase efficiency. Automated processes reduce the need for manual intervention, minimize errors and risks and ensure transparent processing and compliance-secure documentation.

Process mining is an innovative technology that can be used to analyze company processes in detail. It runs in the background and analyzes, for example, how long various tasks take, which steps are necessary and how stringent the workflows are. It records and visualizes the actual activities by extracting data from the IT systems and converting it into clear process models. In this way, the number of different activities and their sequence can be identified and evaluated.

A specific example in cash management is analyzing the payment transaction process. A dedicated process mining system can reveal how long it takes for payments to be approved and executed, which steps are involved and where possible bottlenecks or delays occur. Such a detailed analysis yields valuable insights into how inefficient process steps can be optimized.

These tools also make it possible to continuously monitor and improve processes. As a result, by integrating process mining into cash management, not only can companies increase their efficiency, but also improve the accuracy and reliability of their financial processes.

There are also optimization opportunities for cash management thanks to the latest developments in robotic process automation (RPA). By using RPA tools, repetitive and time-consuming tasks can be automated, which significantly increases efficiency and reduces errors.

RPA tools are easy to use as processes are replicated with the help of recordings. Simple and repetitive workflows (even with various predefined decisions) as well as time-critical or lengthy processes can be performed by a bot. A further advantage over conventional macros is the simpler handling and the ability to work in different systems. This is because RPAs can be used across systems, as they work using the front end and can automate work sequences as long as they are sufficiently rigid. The processing can take place on a schedule or as a function of events. An example of the use of RPA in cash management would be the time-consuming transfer of exchange rates at national central banks to the ERP system.

One benefit of RPA is that it minimizes human error and standardizes the process. When creating the bot, it is often possible to observe process optimization as a side effect. Automating processes not only increases accuracy but also saves valuable time.

Using an RPA tool, such automation can be implemented easily and efficiently. It makes it possible to optimize cross-system processes. In summary, this results in improved data quality and faster processing, which ultimately contributes to optimized decision-making in cash management.

Compared to the technologies mentioned above, machine learning (ML) is classified as AI. It is used in cash management, especially in the processing of account statements. Using ML technologies helps companies to automate their processes and improve the efficiency and accuracy of their financial transactions.

One example of the application of machine learning in cash management is the automated processing of account statements and advice notes in ERP systems. There are a range of tools on the market that offer ML solutions to automatically process bank statements and payment advice notes. These technologies analyze the data, recognize patterns and continuously learn so as to improve the allocation of payments to open items.

The SAP S/4 HANA Cloud also offers functions that use machine learning to process bank statements. These do not require complex data transfers via interfaces between special solutions and the central ERP system. Such ML-supported solutions are capable of automatically analyzing bank statements and extracting the relevant information. They recognize and categorize transactions, match them with open items and post them accordingly. Doing so reduces manual effort and minimizes the risk of errors.

Companies can significantly speed up and improve the accuracy of their bank statement processing by using machine learning. ML algorithms learn from historical data and continuously improve their predictive accuracy. In turn, this leads to faster and more precise processing of financial transactions and improved follow-up processes such as liquidity planning or dunning.

To sum up, machine learning in cash management can significantly optimize processes. The automatic processing of account statements and payment advice notes reduces manual effort, minimizes errors and ensures consistent and transparent processing. Taken together, this improves decision-making and increases the reliability of cash management as a whole.

Although there are many benefits to using AI tools in the cash management department, there are also certain security risks that need to be considered and carefully managed. Below are some of the most important security risks:

1. Data integrity and security:
   AI tools process massive amounts of sensitive financial data. Failure to adequately protect this data can lead to data leaks and unauthorized access. This is why it is necessary to implement robust encryption and access control mechanisms to ensure the integrity and security of the data. An important decision to consider is whether to choose a cloud-based solution or to store the financial data on a company's own server. The fine print of these solutions really does matter as a result.
2. Cyberattacks:
   AI systems may be targeted by cyberattacks in which hackers attempt to gain access to sensitive financial information or manipulate the systems. To prevent such attacks, regular security checks and the implementation of advanced security protocols are necessary.
3. Lack of transparency and traceability:
   AI algorithms can be complex and difficult to understand. It can lead to a lack of transparency, which makes it difficult to identify and correct errors. Reviewing and validating AI models regularly ensures that they work correctly and reliably.
4. Manipulation and fraud:
   AI tools can be susceptible to manipulation, especially if they are based on erroneous or falsified data. Therefore, it is essential to implement strict data validation and monitoring mechanisms to ensure that the data fed into AI models is accurate and trustworthy.
5. Regulatory and compliance risks:
   The use of AI tools must be in compliance with the applicable regulatory and compliance requirements. This includes data protection laws, financial regulations and internal guidelines. It is essential that companies ensure that their AI systems meet these requirements and are regularly reviewed to avoid any compliance breaches.
6. Dependence on third-party providers:
   Many AI tools are made available by third-party providers. This can lead to dependencies and additional security risks, especially if the third-party providers' security standards do not meet internal requirements. For this reason, it is advisable to carry out careful due diligence and conclude clear contractual agreements in order to minimize the aforementioned risks.

Companies can effectively manage cash management security risks and reap the benefits of AI technology by implementing robust security measures and regularly reviewing AI tools.

Paying full attention to the risks on the one hand, but also the numerous use cases and benefits that AI brings on the other, it is worthwhile taking an assessment of current processes in order to be well prepared for upcoming developments. A technologically well-equipped treasury department will secure competitive advantages in the future. This will also change the job profile in cash management going forward, as the increased use of intelligent tools will shift the focus from manual entries to the control of AI tools and their decisions. For this reason, it will be imperative to continuously further train employees in the treasury team.

Source: KPMG Corporate Treasury News, Edition 152, March 2025
Authors:
Börries Többens, Partner, Finance and Treasury Management, Corporate Treasury Advisory,KPMG AG
Nadine Hauptmann, Managerin, Finance and Treasury Management, Corporate Treasury Advisory, KPMG AG