AFP Treasury in Practice Guide
AI-Ready Treasury
Powered by advances in artificial intelligence, corporate treasury is evolving as organizations seek more strategic ways to manage cash, risk and liquidity. As AI becomes embedded in decision making, treasury teams are becoming strategic partners within their organizations, moving beyond day-to-day execution to play a more influential role in shaping business outcomes.
No longer a future concept, AI is actively changing how work is prioritized and value is delivered, enabling teams to move past routine transaction processing and backward looking reporting toward deeper insight and strategic engagement. For treasury leaders, the question is not whether AI matters, but how to adopt it thoughtfully. As expectations continue to rise, leaders are being challenged to modernize not only their tools, but their operating mindset.
This guide offers a practical perspective on what it means to become “AI-ready” in treasury today. It focuses on preparation: understanding where AI can genuinely help, where human expertise remains essential and how the two work best together. The result is a guide designed to help treasury professionals navigate change with confidence, make more informed decisions and position their teams to deliver greater strategic impact in an increasingly dynamic environment.
Wells Fargo, Global Payments & Liquidity
Introduction
Artificial intelligence (AI) is reshaping nearly every area of business — and treasury is no exception. AI creates opportunities to automate key tasks and decision-making processes, freeing up management’s time to focus on more value-added activities. Yet, as with any technological change, realizing these efficiencies requires preparation — in this case, recognition of the pain points in treasury processes and the development of a clear data strategy. Only then is it possible to implement AI in a way that will complement treasury’s existing skills and help the function to become a stronger partner to the business.
This guide is presented in four sections. The first section outlines the different types of AI and illustrates how the technology is being used in treasury departments. The second section describes some of the key risks to manage and hurdles to overcome. The third section examines how AI can be used to bridge different skills gaps within treasury departments. Finally, the guide concludes with a discussion of the specific skills needed within treasury to implement AI and outlines different ways to acquire them.
Understanding AI: Capabilities and use cases
AI is a broad term used to describe a range of technologies, which can make some of the language surrounding its use confusing. For example, what’s labeled “AI” could be anything from cash application solutions to the fraud protection tools used by banks. From the perspective of today’s treasury department, there are three core AI concepts: machine learning, generative AI and agentic AI.
Machine learning
Machine learning is the most established form of AI. Broadly speaking, machine learning works by recognizing patterns and trends in data and identifying data anomalies. This functionality can be used to develop more accurate cash forecasts and drive cash application solutions, both of which lead to a more efficient use of working capital. The rapid detection of anomalies helps protect against the significant risk of payment and other types of fraud.
Generative AI
Generative AI (GenAI), which works through natural language, supports a wide range of use cases. First, it can be used to draft first iterations of content such as reports, RFPs and contracts. For example, GenAI can compare the payment terms recorded in an ERP system with those in a contract and feed the results into cash forecasts.
Second, users can ask questions of the data directly, without the need for specialized reporting tools. For example, treasurers can ask free-form questions to identify data points and trends, such as “What is our cash position?” and “What is our expected cash position at the end of the month?” This functionality can be extended to provide executive leadership with more immediate access to information.
There is a learning process too. “With generative AI, we have to learn how to write the queries, so we get the information we need in the right format,” said Tara Herrera, Senior Vice President and Treasurer, Related.
Agentic AI
Agentic AI represents an extension of AI’s scope and functionality. AI can calculate positions and identify trends, and agentic AI empowers the technology to make decisions based on reasoned outcomes within preset parameters. So, for example, agentic AI would be able to automate both the decision to hedge a foreign exchange exposure and its implementation. While agentic AI is largely the future for now, there are examples of this functionality being used today.
“Agentic AI represents the next phase of treasury automation, where systems can act within defined parameters. While adoption will be gradual, disciplined experimentation today will help organizations prepare for more autonomous models without compromising governance or control.”
— Ather Williams III, Executive Vice President Head of Global Payments & Liquidity and Wholesale Digital Wells Fargo