AI: An Opportunity for Treasurers in Search of Performance

Despite the buzz, AI has in fact been around a relatively long time in its ML (Machine Learning) and RPA (Robotic Process Automation) guises. Both have contributed towards recommendation engines and the deeper automation of cash management and reconciliation.

But the data deluge as commerce digitises further in the 21^st century, as well as the steady rise of Generative AI (GenAI), means the technology now has far more information to crunch through when running applications, and access has also become easier.

Accurate and more frequent cash flow forecasting (CFF) via a model that examines actual historical data analysis to enhance the quality of the data is now possible with contemporary AI.

Previously, unreliable projections hampered effective decision-making within corporate treasuries while leveraging on much more data, through deeper AI analysis performance and speed have both improved. Consequently, physical and financial supply chains can be aligned more effectively, enhancing client relationships and cash efficiency. Liquidity can be optimised, as longer-term investments become possible and less "float" money is needed in current accounts to grease the end-to-end value chain. Additionally, risk-based activities can be prioritised over everyday cash management, which is increasingly becoming automated.

AI can enhance productivity levels by leading to better automation and, ultimately, free up time for treasury department and corporate banking to focus on more added value tasks and services to be extended to clients. The technology can also be used to fight fraud or payment failures while improving financial compliance with: 
•    KYC (Know Your Customer)
•    AML (Anti-Money Laundering)
•    S&E (Sanctions & Embargos)
•    ESG stipulations, releasing better green loan terms as net zero approaches

AI has a multitude of use cases alongside its core automation. These include enhanced forecasting and improved liquidity management applications.

How to join the AI party  

The rise of GenAI applications such as ChatGPT, from the OpenAI and Microsoft collaboration, has democratised access and affordability for more users. This has furthered the technology’s advance in customer services and the ability of treasurers to develop AI tools for themselves at scale in the value chain. These include processing claims management, refunds, inventory or other mid to back-office procedures much more efficiently than was previously possible. 

“Treasurers and banks have similar stakes to deploy reliable and responsible AI models in their business environments. Internal skills are scarce, vendors’ panoramas are wide and extremely fast-changing. Corporate bankers can act as partners in the corporates’ treasury department transformation, by providing innovative data-driven solutions and services, but also informed insights and advice on AI transformation of financial processes” says Ellezam-Danielo. “AI will have broad applications in future, including coding itself. But we will still need watchful humans to check it before roll-outs and to ensure IT software development, engineering and indeed any AI projects, have appropriate governance and on-going control mechanisms. This is vital for scaling AI in a responsible manner.”   

Despite the evolution of AI, Anfaoui states: “Corporate banks and their clients are not using anywhere near enough data at the moment.”  

However, she continues: “AI will change this, especially as it overlaps with the trend towards increased usage of open APIs (Application Programming Interfaces) as an easier means of connectivity and cross-border data exchange between banks, businesses, and organisations such as Swift. Data will flow more freely in future. APIs will help to embed AI functionality into ERP and TMS systems and avoid digital islands, where its number-crunching and automating capabilities haven’t previously been able to go.”

“Allied to the vast new pools of data available thanks to increasing digitalisation, AI will become more and more effective,” predicts Anfaoui. “It will make us all data-centric organisations – if we have the skills to participate.”

Ellezam-Danielo, agrees but stresses: “Corporates have to recruit and retain AI and data experts to make this happen and, just as crucially, upskill their existing business and risk specialists. They are the people who can identify specific AI end use cases and design the right implementation conditions.”

“Banks and vendors can help with this,” she adds. “Collaboration with an experienced partner is advisable. Societe Generale today has almost 400 AI use cases in portfolio, half of which are already in production – and at scale.”

Better all round

The key benefits of AI, or the killer applications of it, that banks and their clients should already be aiming for are:

• Better cash management and forecasting: this provides automated efficiency when handling payments in and out. But AI’s ability to prevent payment failures, help with prevalidation checks and align with regulations, while enhancing the accuracy of forecasting is also a boon for allocating capital more effectively. AI brings enterprise-wide benefits. Being able to see the future more clearly, with humans interpreting the data, also delivers risk management advantages in an increasingly geopolitically and economically unstable world.

• Better risk and anti-fraud capabilities: these flow directly from more accurate data. AI’s ability to detect and prevent suspicious fraudulent activity is an immediate benefit that should be pursued. As scammers increasingly employ AI, the need to counter them with the same technology only increases. AI can also scan hundreds of pages of legal documents. This ensures compliance with the escalation and fortification of regulations regarding sanctions, which is prevalent due to the war in Ukraine. Other economies are similarly turbulent currently. The data interrogation powers of AI can aid risk management by spotting investment patterns to mitigate FX and other volatility. It can also better align services and obligations across an organisation.

• Better intraday liquidity management: this is already occurring in the banking sector. AI is atop of this specific data. The benefits are myriad: they include enabling the swifter overview of payment and settlement obligations and analysis of trade finance parameters in a timelier fashion, all of which impact pricing. This functionality can easily be rolled out to corporate clients. Normal and stress-test scenario planning and easier adherence to capital buffers are processes that are also simplified with more powerful AI-enabled data tools.

Anticipation is key to control risks

The desire to cut operational costs, enhance client satisfaction, boost compliance, and mitigate risks, while spurring growth and business development nuances are clear. But there are also dangers that come with AI, not least of which is the lack of control if implementation planning is poor. 
On-going governance is essential if an AI model is released into the wild. There are obvious dangers ahead if appropriate planning and oversight isn’t baked in before a roll-out.

Many examples of dysfunctional bots illustrate how AI can go rogue if its programming, governance, and oversight limitations, detailing the parameters it must operate within, are deficient. Putting such a system into the cash management process is dangerous if the groundwork has not been laid properly. A test sandbox and scenario role-playing activities are both advisable and wise before any roll-out.

No time like the present

Appropriate governance, and a clear strategy as to how AI should align with your data, people, processes, and aims, is key to achieving the best results. Otherwise, it will be a case of garbage in and garbage out – in the same way that a badly designed Excel spreadsheet can be disastrous if not configured properly.

“Keeping a 'human-in-the-loop', yes/no option or an ‘exit door’ for any contentious decisions or in the event of anything going wrong, should also be part of any implementation procedure,” advises Ellezam-Danielo, while stressing that good governance is vital to allay unwarranted fears about the "machines taking over"

“AI can potentially help in all areas across a business. Prioritisation is key to avoid investment dispersion. There are some ‘low-hanging fruits’, of course, that everyone is working on, such as using generative AI to draft meeting minutes, or as an assistant of sorts. But there is no one-size-fits-all ‘killer’ use case that is emerging as a must-have. The real value is much more in leveraging digital, automation, and AI as a whole package of technologies to redesign core end-to-end processes, which are instrumental to a firm’s efficiency or client satisfaction”, she adds.

“Getting quick wins spurs confidence and further adoption, so focusing your investment is a good idea,” Ellezam-Danielo continues. “This approach stops AI being just a technology topic and incubates an AI roll-out methodology within your organisation.”

“The rise of AI is an opportunity for treasurers to modernise their overall IT systems, improve digitalisation and better align services in order to enhance the agility, speed and performance of a model,” concludes Ellezam-Danielo. “AI will enhance the automation and efficiency of systems and the accuracy of forecasting and compliance, which is why roll-outs are already happening across financial services. As people see the benefits AI can bring, it only triggers further adoption. The time to start using this technology is now.”

This article is part of the whitepaper "Shaping the Future of Treasury", published by Treasury Management International (TMI) - September 2024. Click here to download the whitepaper.