While companies of all shapes and sizes are exploring the opportunities presented by AI, what this looks like in practice is different for every business. No two AI journeys are alike – but nevertheless, there is much that corporate treasurers can do to prepare for AI adoption and make their projects a success.

Harnessing AI for process automation and cash forecasting

For many treasurers seeking to adopt AI, one of the most significant areas of interest is cash flow forecasting. Projecting future cash flows is of critical importance, as Jarrett Bruhn, Managing Director, Head of Data & Artificial Intelligence, Global Payments Solutions at Bank of America observes.

“If you have too little cash on your balance sheet, you may need to access liquidity lines,” he says. “And if you have too much cash, you’re not making the most efficient use of those funds.”

Nevertheless, forecasting is also one of the hardest problems for treasurers to solve. The forecasting process tends to involve gathering data from disparate sources around the business – and all too often, the resulting forecasts are not sufficiently accurate. As Bruhn explains, “The variability that companies face in their cash flows, seasonality, unexpected capital expenditure and unexpected operating expenses, mean that accurate forecasting is a significant challenge.”

With data-driven analysis and patterns, companies can predict more accurately when cash will come into and out of the business. As such, there is much that AI can do to enhance the forecasting process.

But while AI can provide very real benefits – both in the forecasting process and in other areas of treasury, such as process automation – it also needs to be adopted in a way that delivers measurable outcomes. For treasurers embarking on their AI journeys, this means staying focused on the desired results, and seeking advice from partners with experience of AI implementation.

Making the impossible feel inevitable

During his career in banking, Bruhn has led teams in several data-intensive businesses, including structured credit and commodities. These experiences have shaped the way he thinks about data and AI in his current role, which involves leading the global data and AI strategy for Bank of America’s Global Payments Solutions (GPS) business.

“My career has always been about helping clients make better decisions in complex, fast-moving markets,” says Bruhn. “Commodities, structured credit and payments may all seem very different – but they share a common theme, which is the value of data, analytics, judgement, speed and trust.”

With treasurers looking for ever more sophisticated technology, Bruhn says the adoption of AI is a particularly significant shift. “It’s not about replacing judgement,” he adds. “It’s about helping our teams move faster and get better information, so that they can make better decisions.”

At the same time, rapid growth in this space means that the scope of what is achievable is changing considerably. As Bruhn explains, “I was recently asked to write down one thing I would want others to know about the work I do. The description I came up with was, ‘The trick to our success is to take the impossible, and make it feel inevitable.’

“So many of the problems we deal with are highly complex – but when you create that inevitability, you turn that impossible thing into something that you can and will achieve.”

Adopting AI at scale

Adopting AI at scale does not happen by accident, and Bank of America has been investing in AI for a number of years. “A great example is Erica®, the virtual AI financial assistant that our Consumer Banking clients-use to help them manage their personal finances,” says Bruhn. Following its launch in 2018, Erica now helps more than 24.6 million clients manage their financial lives.

In 2025, the bank launched AskGPS, an internal GenAI assistant which allows employees to access answers to questions posed by clients. Enquiries that previously took an hour to resolve can now be addressed within seconds, enabling the bank to provide faster, comprehensive responses to clients. This, in turn, is freeing up staff to focus on higher-value advisory services.

As Bruhn explains, “AskGPS helps our teams access trusted information faster, so that we can answer complex questions quickly and consistently. It helps make our salespeople smarter for our engagements with clients and product recommendations.”

The bank’s AI-powered solutions for GPS also include CashPro Forecasting, which uses predictive analytics to forecast cash positions. Meanwhile, Bank of America Intelligent Receivables uses AI and advanced data capture technology to improve straight-through reconciliation rates by combining payment information and remittance detail from different payment channels.

Reflecting on the bank’s AI journey so far, Bruhn says that AI adoption requires strong data, strong infrastructure, clear testing, governance controls and human accountability. “That’s what makes the journey powerful,” he says. “It’s not about deploying AI – it’s about doing it responsibly, in ways that enhance service execution and client trust.”

Getting started

With AI evolving at a very rapid rate, Bruhn emphasises that clients both large and small can benefit from harnessing this technology. First, however, they need to get the foundation right.

“One lesson we’ve learnt is that AI cannot be about shiny toys and fancy dashboards – it has to be about identifying and solving problems that really matter,” he says. “This means knowing your business well. It also means figuring out which meaningful process or meaningful insight you want that only AI can bring you – such as better cash forecasting, or effective workflow tools.”

For clients looking to get started with AI, Bruhn also advises focusing on a project that’s a good fit for the company’s resources. “It’s important to start with a reasonably sized project for the problem you’re solving,” he says. “It’s better to get that small win that shows the incremental viability of AI, and then build on that over time. Frequently we see clients picking a project that’s too big.”

Other important considerations include building the right talent and culture within the organisation, having access to clean data, and putting appropriate governance in place.

• Talent. Where talent is concerned, Bruhn explains that this doesn’t necessarily mean hiring a data scientist, but “changing the way that your company thinks about data analytics and AI and upskilling your talent so that everyone thinks about these things in their daily workflows.” He adds that this includes building a culture that prioritises a willingness to learn.

• Clean data. When it comes to deploying AI, Bruhn says it is essential to focus on data from the outset. “If you don’t have clean, well-governed data, your results will be skewed, and your outcomes won’t be trusted,” he warns. “The quality of the outcome is determined by the quality of the inputs.”

• Governance and oversight. Another prerequisite for AI adoption is having appropriate governance and oversight in place. As Bruhn explains, AI models require effective governance, not least because inputs and outputs can change along the way.

• Adopt AI at scale. With all the other steps complete, the final step is to work out how to build on the initial success of the project and adopt AI on a larger scale.

A marathon, not a sprint

In summary, Bruhn says companies seeking to adopt AI need to dedicate the necessary resources to the project, analyse the results and then scale up accordingly.

“Wherever your company is on its AI journey, you need to get your data right, get your talent focused, pick the right use case and make sure you have the right governance in place,” he says.

“Understand that AI is a marathon and not a sprint – AI at scale is hard! And remember that the future will belong to companies that turn AI into real, measurable outcomes, not just ideas.”