One way AI is being adopted in cash management is via Robotic Process Automation. Corporates can use the technology to improve daily cash positioning and support forecasting, Amy Beninato, Executive Director, explained. She said how on the banking side, RPA also supports payment analytics and fraud detection plus a suite of efficiencies and productivity gains that will lower costs and lead to greater client acquisitions and retention.
Elsewhere, AI will support liquidity management by allowing companies with established domestic and regional pools to use predictive analysis to pull excess funds from overseas or support tax payments in regulated markets.
Karen Hom, Managing Director, added that in Indonesia, where many banks are not SWIFT enabled and automatic sweeping is challenging, AI is adding value to Fintech solutions. In trade finance, AI is used to pre-fill standard templates and review documents where language is customised to highlight outliers.
Getting started
Corporates can begin integrating AI by developing a flow map of their treasury operations core systems. It is very difficult to begin replacing manual processes until companies have the full scope of all their existing processes which highlight the repetitive, predictable and non-strategic tasks that AI does best.
This map of existing workflows should go beyond account payables and receivables, advises Beninato. For example, detailed process flow documents can include sub workflows (often with strict sequencing) around, say, maturing time deposits, outstanding investments, previous day cash flows, or investment limit calculations. On top of this, the map can add in granular intra-day payments, or collections forecasts and other data management.
However, she warns that constructing detailed road maps can also lead down rabbit holes and that these platforms can end up as multipage documents. “The devil is in the detail and anything unexpected can have a negative impact on the AI programme.”
Both Beninato and Hom agreed that treasury is playing a central role as corporates begin to craft their data strategies and gather the raw material needed to feed AI. Treasury functions that need a data strategy include investment, cash flow forecasting, exposure identification and management, execution and settlement and payments.
Applying AI to customer service and vendor management, KYC, AML and risk management also requires a data strategy. Challenges include ensuring good quality data from multiple sources spanning ERP, TMS, bank portals and third-party providers. “It entails a close partnership between treasury, operations, the business function and the data architecture and tech teams,” says Beninato who lists the frequent involvement of procurement, risk management and even marketing in cross functional AI taskforce teams working under treasury’s remit in the initial phase.
Integrating AI comes with key risks. Top of the list is data security and preventing cyberattacks and data breaches. AI also brings operational risk that could include a system malfunction and regulatory risk. “It’s important to known how to address these risks and implement robust security measures,” says Hom who adds that monitoring AI systems and keeping abreast of risk, especially regulatory compliance, will always require a human element.
“CrowdStrike showed the risk of reliance on third-party providers,” she said. She described the relationship as one of partnership where fintechs provide a value add, and said that evolving agreements shape how fintechs and banks share data.
Rather than AI “taking over” treasury, Hom believes that treasury will be front and centre positioning the human element of the technology.
“The treasury function is shaped by human relationships and human decisions,” she said.
AI is best seen through the lens of value add around risk profiling or, for example, standard LC issuance when language is approved and standardised. “Treasury will always need a human element.”
Beninato concluded that AI will also have an impact on treasury’s ability to integrate with sustainability matters and metrics. For example, the technology will support data and trend analysis around the UN’s SDGs or disease patterns, enabling corporates to target intervention.
