AI’s transformative potential for corporate treasuries in Asia

Artificial intelligence has captured the imagination of businesses, investors, and individuals. For example, in 2022, the launch of ChatGPT highlighted the potential of the technology even though AI solutions had existed for decades beforehand.¹

In Hong Kong, 96% of payment instructions are done entirely digitally,² while India’s digital payments system processed 117.6 billion transactions in 2023.³ Corporate treasuries may be ripe for AI integration,⁴ with a range of tasks and processes that could be enhanced with new technology. In many Asian markets, the proliferation of digital payments technology provides a strong starting point.

A consensus is emerging that the successful businesses of the future may be those that move early to incorporate AI, including generative AI, into their operations.⁵^,⁶ In a recent survey by consultancy EY, 71% of Asia Pacific CEOs said they needed to act quickly on generative AI to avoid giving their competitors a strategic advantage.⁷

But the same CEO Outlook Pulse survey found a similar proportion said that uncertainties around AI were hindering their efforts to press ahead with relevant investments.

One of the key areas of uncertainty for businesses is how AI and generative AI will be regulated, as different countries consider a range of supervisory approaches to this new technology – even within Asia.⁸

So, what more do business leaders need to know to implement AI strategies with confidence?

Corporate treasurers are exploring a wide range of potential use cases for AI tools – from payments and operations through to reporting and risk management:

• AI algorithms can be used to predict customer behaviour and spot trends, helping corporate treasurers build more accurate and dynamic cash flow forecasts. E-commerce companies, with access to vast amounts of data on sales and searches, have been pioneers in this area⁹.
• Corporate treasury officers could deploy AI to identify where resources and cash are tied up within the business, improving efficiencies. In a recent HSBC survey¹⁰, improving return on capital and capital re-allocation are key concerns for treasurers, with AI poised to help improve business operations in these areas.
• AI could help reduce some manual work, for example by creating efficiencies in manual processes. In the same HSBC survey, more than half of those surveyed said that 70% of their time is spent on running processes whereas only 30% is spent on matters to drive change and grow their businesses,¹¹ highlighting the potential benefits that AI tools could bring.
• Integrating AI into risk management processes could also improve fraud detection capabilities.
• Finance managers may also benefit from other uses of AI, such as content summarisation and content generation capabilities or visualisation. For example, one study found that business professionals who used AI could write 59% more business documents per hour.¹²

However, AI’s enormous transformative potential comes with technical, social, ethical, regulatory, and security risks. The more immediate and relevant concerns for businesses focus on privacy, intellectual property, and possible job disruption.¹³^,¹⁴

Treasury managers could also consider whether interactions with AI applications are secure and that customers’ data privacy is protected. This includes assessing technology providers’ security credentials, where their data will be hosted and whether the technology provider will use that data to train their own large language models.

As the regulatory and industry guidance on governing AI continues to evolve, treasury managers ought to be aware of potential risks that any AI solutions might pose and consider the best ways to mitigate those with appropriate controls frameworks and human oversight.

Here are some important considerations for any treasury manager to keep in mind:

• AI tools need accurate and timely data. Real-time data is essential for accurate AI projections.
• A successful AI model depends on the digitisation of internal data. AI models are trained with data that allows them to perform specific functions, which might require businesses to have sufficiently matured data sets to reap benefits from AI applications.
• Continuous tracking and monitoring are needed to protect against inaccuracies and mitigate risks.
• Data security can be a big concern, especially for treasurers dealing with personal or sensitive information as well as data privacy.

The situation of every business is unique, but the following considerations may be useful when weighing up whether to invest in AI solutions.

1. Take your time. New systems take time to build. Before investing in AI technology, finance managers should think about the right approach for their business. This includes how much automation is appropriate, and where AI can best add value, to best understand the scope of the problems that AI solutions could support the business with.
2. Allocate resources carefully. While some off-the-shelf generative AI tools, such as ChatGPT, are available at low cost, developing and implementing AI for business processes requires a gradual approach. Tools such as ChatGPT would likely not be suitable to meet every business need. Large Language Models (the technology behind ChatGPT) tailored to the specific needs of the business, trained on internal data rather than on the internet, and any of the associated technology, monitoring, and talent required to develop these models and stress-test them might be costly.
3. Develop an AI framework. Businesses need their own comprehensive and relevant AI governance frameworks to ensure they meet their customers’ expectations and comply with evolving regulations that will likely differ considerably across markets. Common features of such frameworks include data integrity, accuracy, transparency, interpretability, and accountability. Rather than employees having to make ad-hoc decisions, it is important to set out clear guidelines on AI development, testing, implementation, and oversight.
4. Consider how digitisation may impact AI. AI models are only as good as the quality of the data they are trained on. Digitisation can be the first step towards unlocking the benefits of AI. E-commerce companies, for example, are looking to AI to forecast cash flow and to deploy funding. However, the high number of transactions on an e-commerce platform poses a challenge, as these need to be captured in real time across multiple payment channels.¹⁵
5. Consider the use cases. Any investment in technology needs to deliver a tangible return. In a corporate treasury setting, there are many ways AI could add value, such as by helping forecast demand, predict cash flows, and initiate and process payment instructions, or by automating analytics and reporting. However, businesses will need to consider whether this investment offers good value for money depending on their individual needs and AI use cases, and whether it is suitable given their technology capabilities and regulatory expectations.

These considerations have helped to inform HSBC’s own approach to AI. We continue to help our corporate clients unlock the benefits of digitalisation¹⁶, and have developed principles for the ethical use of data and AI, including protecting privacy, using AI with a clearly defined purpose, and addressing unfair bias and decision-making among other factors.¹⁷ Moreover, we are actively developing to AI-powered solutions to respond to customer needs. For example, HSBC AI Markets allows users to access bespoke financial market analysis and browse HSBC Research with the use of our proprietary natural language processing (NLP) engine (where local regulations permit).¹⁸

With these factors in mind, businesses may wish to consider leveraging AI across their corporate treasury functions, provided that the risks, benefits, and costs are thoroughly considered and accounted for.