Maximising Efficiency

Published: November 26, 2024

Download this articles as a PDF
Maximising Efficiency
Jane Cooper picture
Jane Cooper
Editorial Contributor, Treasury Management International (TMI)
Narendra Kumar Parhi picture
Narendra Kumar Parhi
Head of Digital Channels APAC and Head of Treasury Products Hong Kong, Global Transaction Services, Bank of America
Serina Hourican picture
Serina Hourican
Head of Commercial Sales, Asia Pacific, Global Payments Solutions, Bank of America

Optimising Treasury and Transactions with AI

AI’s evolution into the cognitive realm means that treasurers can make better, faster, and smarter decisions in a number of areas, including transactions, hedging and cash forecasting. Not only does this free up time to deal with the complexity of treasury operations, AI is also making an impact on the bottom line of businesses.

AI is becoming even more intelligent. The technology has moved through the stages of automation and predictive data analytics and is now firmly in the arena of smart decision-making. And now AI can simulate human cognition, inform people’s judgement, and is helping people make informed choices, says Parhi.

This evolution has coincided with a treasury shift away from solely being an operations function. “The KPI for treasurers is no longer just to execute; it is also to contribute towards the P&L of the company – and that is where the evolution of AI becomes relevant for treasurers. It helps treasurers significantly because the technology has evolved from dealing with the repetition of automated tasks to creating more predictive analysis and is now increasingly focusing on some of the cognitive aspects to help with decision making – that’s how the role of treasury has kept pace with the evolution of AI,” Parhi explains.

Too much data, too little time

As technology has evolved, so have the demands of the treasurers’ role. Global companies are operating across multiple jurisdictions, in different time zones and currencies, and managing transactions is complex. It’s easy to become overwhelmed. Hourican acknowledges one of the major challenges facing treasurers: “There’s just too much data”, she stresses.

This is where AI can step in, crunching the data and offering predictions to help treasurers execute their duties. “Now treasurers are looking at P&L opportunities. They’re examining ways to fund other transactions in the company, such as M&As, and they’re working much closer with the different stakeholders they have within their company,” she continues.

The role of the treasurer has grown, and so has the amount of data confronting them. “Treasury teams are spending considerable time getting that data together, putting it in a format that makes sense, and then working out the trends to help them make a decision,” Hourican explains. “We still see a number of treasury teams manually using spreadsheets instead of leveraging the technology they have.” Here AI can step in to collect, organise, and make sense of the data to give treasurers clearer insight – and oversight.

APIs pave the way for AI’s efficacy

AI can optimise transactions, but are APIs necessary to do this effectively? Parhi thinks they certainly are. Previously, he explains, companies and banks struggled with the numerous formats that exist in the financial infrastructure. “Creating standardised communication across different systems is key to deriving the most benefit out of AI,” Parhi declares.

He points to an example of a non-bank FI that was processing large volumes of transactions in a short period of time. “They had more than 100 file formats that they were managing across 30 banking partners,” recalls Parhi. “They had an army of people just to manage the version controls and the format upgrades.”

This is where APIs came in, and through Bank of America’s global API gateway, the formats were streamlined into just four API templates, which cut the processing time from two to three hours to less than three minutes. This is an example of how treasury can improve its processes and also have a real impact on the business. With quicker turnaround times, the company’s customers were able to receive payouts much faster, which in turn improved the overall customer service experience.

Hourican points out that the API is a two-way channel; it’s not just the bank pushing information to the treasury team. “It’s like an open tunnel where cars go both ways – that’s where you can reduce the time spent on processes and increase efficiency,” she explains. With Bank of America’s Guaranteed FX Rates product, for example, the FI can guarantee the FX rate for a period as short as 30 minutes and as long as one year (for some currencies). The API connection speeds up arranging such a currency conversion as the tool can transmit information to the client about the rates, and they can push information back about the volumes they require. With APIs, Hourican explains, “it doesn’t have to be a manual push one way and then a manual push the other way, automation is the key”.

AI receivables and ‘banana dramas’

Reconciliation is a key area where AI can optimise transactions, especially for large organisations that are working with multiple formats. Parhi explains that when corporates receive payments, multiple data elements need to be compared and in global companies there could be tens of thousands of transactions at any one time. And if their customers are sending confirmation of payments, these could also be in different formats and via different channels. For some clients, “there were so many ways in which they received the confirmation of payment together with the underlying invoice”, he says. This information also needs to be reconciled with the funds after they reach the account, as well as the need to keep track of outstanding invoices.

“We have seen, in general, that large complex organisations couldn’t automate more than 30% or 40% of their reconciliations for incoming receipts. The remainder had to be carried manually,” explains Parhi.

This is where Bank of America’s Intelligent Receivables steps into the spotlight. It is a solution that consumes incoming sources of data in such a way that they can be read by a machine. “Once the tool reads a sufficient amount of data, it is self-learning, which is where ML comes in. The more it learns – such as the format to expect from certain customers – the more it can automate tasks,” enthuses Parhi. With this tool in place, the 30 to 40% automation figure can rise to 90% and beyond, he adds, thereby significantly reducing the manual reconciliation efforts.

Hourican adds that clients have specifically hired people to do this process manually and now their time can be spent on other responsibilities, while also increasing their straight-through processing rates and reducing turn-around times. The tool, she adds, becomes more attuned to the customer over time. “It is also an opportunity to be more efficient with your working capital and reduce your Day Sales Outstanding she adds.

The tool is adaptable and through its Intelligent Deductions module, can accommodate situations that frequently occur. For example, if a company is shipping bananas from Ecuador to the UK and the quality of the fruit deteriorates during the journey, the buyer will probably complain. For the seller, this isn’t necessarily an issue because they can change the transaction to a partial refund immediately. “This deductions module is clever enough to deal with that kind of situation automatically,” Hourican explains. She adds that this solution is suited to companies with high volumes of transactions, such as e-commerce companies or those in the fast-moving consumer goods (FMCG) sector.

Driving efficiencies with predictive forecasting

Despite the evolution in AI, it is surprising how many treasurers rely on Excel to do their forecasting, says Parhi, echoing Hourican. And, as a result, they are struggling to predict their cash flows effectively.

Using AI can be simple yet effective, explains Parhi. “We can take the data from the accounts companies have with us, as well as their cross-border traffic over Swift, and analyse it for them.” At a basic level, this could be a simple cash position of what occurred during the previous two years on a monthly basis. From this, the tool can forecast for the next year based on certain trends.

Each month, as the data is refreshed, the tool will look at the variance between what it predicted and the actual data – and learn from it. “When we do the forecasting for a period of time, similar to the Intelligent Receivables product, the ML actually improves the accuracy of the forecast,” explains Parhi. The forecast, he says, can achieve around 90 to 95% accuracy over time and this can have significant benefits to cash positioning and optimisation by treasurers.

“We have seen customers in the industrial, consumer and technology sectors, because of the simple forecasting implementation, reduce their surplus operating balances globally by up to US$200m within a span of 12 to 18 months,” Parhi says. And in the previous higher-rate environment, that could save them around US$10m per annum, he adds.

Parhi points out that corporates can take advantage of this predictive analysis if they subscribe to Bank of America’s CashPro solution. And the tool allows for account and transactional data held at other institutions to be included into the analysis thereby enabling a global multi-bank forecasting implementation.

“If they don’t have a history with us, there is an option available for them to feed their historical account and balance data from other banks to just see how the forecasting would look for them – this helps them create the business case internally to subscribe to it.”

Embracing change for the better

Alongside the evolution of AI, the global financial infrastructure has also been migrating to using the same messaging standard: ISO 20022.

“I cannot stress enough how ISO 20022 is one of the most impactful trends in the current payment landscape,” Parhi says. It is a game-changer, he elaborates, and will lay the foundation for the next 50 or even 100 years. With the new messaging standard, more data will flow with the transactions, and the opportunities – for automation, predictive analysis and AI-driven decision-making – will only increase.

Parhi comments that, during the course of his 20-year career in treasury, he has seen a number of technological advancements and notes how innovations tend to have a cycle of five or six years, where new technology eventually needs to be upgraded. Because of the endless sequence of upgrades and implementations and the huge drag it has on cost and resources, corporates in general tend to be highly resistant to change. The move to ISO 20022, however, is different because it creates a new baseline for the industry on a global scale that should continue to work for decades irrespective of future technological advancements.

This kind of standardisation is just one of the elements that goes into future-proofing treasuries. This, combined with APIs and the ability to leverage AI to optimise transaction workflows and decision making at global scale, means that treasurers can continue to gain efficiencies, and contribute more strategically to overall growth and sustainability of their organisations.

Sign up for free to read the full article

Download this articles as a PDF
Article Last Updated: November 26, 2024

Related Content