AI thinks treasury management is ready for transformation

Artificial intelligence may soon enhance the efficiency of your corporate treasury department, but AI is not a cure-all. To reap the benefits, it’s essential to know both the potential and limitations of AI.

Tags: Artificial intelligence, AI, Innovation
Published: August 08, 2018

From autocorrect on a smartphone to auto-suggesting products to buy based on your shopping patterns, artificial intelligence (AI) has already made its way into daily life. The spread of AI has corporate treasurers asking important questions, such as how AI will affect their teams and what to do now to prepare for the AI-enabled future.

“AI means different things to different players in the corporate treasury ecosystem,” says U.S. Bank Global Treasury Management Senior Vice President and Head of Product Management, Sayantan Chakraborty. “It’s critical to understand what AI can do today versus future possibilities.”

Corporate treasurers who understand this difference can better prepare for both the benefits and the challenges AI presents. Although AI has the potential to enhance many aspects of the treasury function, its payoffs come only when the right foundation is in place and the organization sets realistic expectations.

 

Just what is artificial intelligence, and what’s its potential?

AI is a broad and fluid category of computer science that attempts to solve problems that were previously believed to be only solveable by human intellect. The definition of AI evolves as technology advances, and some AI applications are now so commonplace that many people don’t see them as AI. This evolution is named the AI effect, according to U.S. Bank Innovation Team Senior Vice President and Artificial Intelligence leader, David Berglund.

 

Here are three things to consider:

  • AI is poised to transform treasury management, which has historically required employees to complete predictable and/or routine manual processes and manage exceptions. The key breakthrough value AI provides is in potentially lowering the cost of prediction across use cases.
  • Machine learning, a sub-field of AI, uses an automated and ever-evolving learning process to solve complex problems. Leveraging large data sets, a computer is able to improve its performance on a task over time. AI and machine learning technologies can swiftly organize and “understand” qualitative data and even language.
  • In advanced process automation (APA), a computer learns to do a task by observing and recording the steps an employee takes. Then it performs the task itself. When combined, APA and machine learning have the potential to significantly increase treasury efficiency.

 

The treasury of the future

Computers can quickly and efficiently process significant amounts of data, meaning they accelerate pattern recognition and enhance decision-making. Treasuries of the future could apply data mining and AI-based analytics to data sets that include all types of payments and collections. They could then use this analysis to find ways to lower transaction costs and reduce the need for working capital, Chakraborty says.

In addition, data mining and AI could combine to identify new metrics to better measure treasury performance. 

Machine learning, robotic process automation and other technologies are already taking on some tasks in controlled environments, such as accounts receivable reconciliation and web-based chatbots for customer support. Within the next few years, these technologies will likely have expanded capabilities and will be in wide use, Chakraborty says.

“People won’t have to spend time tagging and documenting and putting files away in folders,” Berglund says. “Rather, when the time comes, they’ll be able to easily identify what they’re looking for within those long documents.” However, many AI applications are still in the proof-of-concept stage, in part because corporate treasuries require specialized solutions.

“We’re still in the age of weak AI,” Berglund says. “We can solve very narrowly defined tasks with some models, but it’s not yet generally transferrable.”

 

Laying the foundation for AI

Some of the hype surrounding AI stems from a belief that corporate treasury simply needs to acquire the right technology to yield results, Berglund says. In fact, rather than talking about implementing AI, he prefers to say “applying AI” to emphasize the fact that AI is not, in itself, a solution. Technology, no matter how advanced, can’t fix inefficient or ineffective business processes, Chakraborty adds.

Flawed processes can develop over time as a result of workarounds to accommodate exceptions, or from underinvestment in technology. Regardless of their origins, flawed processes can incur hidden costs, often in the form of wasted employee time or project delays.

Before seeking technology solutions, treasury management departments need to clearly understand and systematically improve how their teams work. Chakraborty suggests treasurers first identify a few key processes, analyze their direct and indirect costs and begin refining them. Then, when AI is available to enhance or take over those tasks, treasury management departments can optimize the results. 

 

There’s a lot of energy and excitement around AI, and a lot of funding going into this space, but it’s still built on good processes and good data, which we can control today.

 

Data is fundamental

Effectively applying AI requires more than streamlined and unified business processes. Data is also essential. Although many organizations have large amounts of data, much of it may not be useable by AI technologies, Berglund says.

For AI to use it effectively, data should be:

  • Formatted, labeled and organized the way the technology needs it, which may require employees to rework it.
  • Relevant to the task. If the right data isn’t available, employees can create it by training computers to do specific tasks.
  • Stored in up-to-date database systems that can accommodate increasing amounts of data over time.   

 

As AI advances, corporate treasurers also want to think about compliance or “the explainability problem.” It’s the ability to explain what’s happening in some of these highly complex algorithms, Berglund says. “Without understanding how AI and other technologies arrive at conclusions, it’s impossible to demonstrate the decision-making process to regulators or, if necessary, customers.”

 

The impact on employees

For AI to move from proof-of-concept to wide application, more than technological advances are necessary, according to Chakraborty. People’s work-style also needs to adapt.

A key factor in that shift is generational change. Members of Generation X are now in middle and senior management positions. Millennials and Generation Z make up a large share of entry-level and non-management employees. The workforce of today is more likely to embrace new technologies outside of work, and this attitude can help propel the business case for AI applications that, so far, have not gained much traction, Chakraborty says.  

Within the next three to five years, both Berglund and Chakraborty expect corporate treasury employees to shift some of their repetitive tasks to machines and spend their time doing higher-value work. And according to the 2017 AFP Strategic Role of Treasury Survey, more than a third of treasury professional respondents indicate that technology and automation have already enabled them to make that shift.

 

Moving forward

To prepare for the AI-enabled future, Chakraborty suggests corporate treasurers think about forward compatibility when making technology purchase decisions. He advises against buying software that doesn’t allow for a future application of AI. 

At many organizations, the treasury function is a cost center rather than a profit center, so its technologies may not be updated as regularly as those in other departments. “At U.S. Bank, we understand this. We can actually help bridge the gap by creating tools that make it possible for our clients to take advantage of new technologies,” Chakraborty says.

Treasurers will also benefit by taking a perspective that’s balanced between the present and the future. “There are more proven and stable technologies that, with just a little bit of experimentation and knowledge, you can use today and get a payoff in the future,” Berglund says. “There’s a lot of energy and excitement around AI, and a lot of funding going into this space, but it’s still built on good processes and good data, which we can control today.” 

 

To learn more about how AI can enhance your treasury department today and tomorrow, contact your U.S. Bank relationship manager.

 
©2018 U.S. Bank. Member FDIC.