Data and AI: Reshaping treasury in the age of volatility

Treasury used to be judged on how well it managed cash and closed the books. Now it's judged on how fast it can read the future. 

Geopolitical fragmentation, expanding mandates and a technology cycle started moving faster than most planning horizons. The corporate treasurer sits at the intersection of all three pressures, and key financial data access paired with AI has become the connective tissue that decides whether that role scales or strains.

Why this moment is different

For years, treasury transformation was slow and tactical. Systems were replaced on multi-year cycles; some went decades without review. Forecasting cadences were monthly, sometimes quarterly. But geopolitical volatility has reintroduced a level of unpredictability that treasurers can't schedule around. 

Sanctions exposure, trade policy shifts, inflationary pressures and regional instability impacting supply chains now move faster than a quarterly review can absorb. Meanwhile, the scope of treasury is broader. Treasurers are expected to act as a strategic advisor to the C-suite, weighing in on funding strategy, counterparty risk, working capital (particularly free cash flow) and even technology investment. 

Simultaneously, AI is advancing so quickly that the cost of waiting has changed. Capabilities that once seemed experimental are now actively shaping how organizations forecast, reconcile, report and detect anomalies. This creates a genuine tension for a discipline that’s always prioritized caution: move too fast and you inherit risk you don't understand, move too slow and you fall behind teams that have already adapted.

The common thread across all three pressures is data. Volatility demands better visibility. An expanded role demands better analysis. And AI is only as useful as the data it's given to work with.

The challenge underneath the challenge

Most treasurers already know where the friction lives: cash sitting across dozens of banking relationships in multiple currencies and jurisdictions; forecasting data that gets collected and consolidated with manual intervention; legacy systems that don't talk to each other, so the same figure has to be reconciled multiple times before anyone trusts it.

This is the obstacle to everything else. You can't manage geopolitical risk in real time if your true cash position is days or weeks old. You can't play a strategic role if your team spends its week firefighting reconciliations. And you can't deploy AI meaningfully on top of fragmented, manually assembled data, because the model inherits every inconsistency and won’t scale.

So, the real problem is that fragmented data and manual processes limit how far treasury can stretch, no matter how capable the team. And market volatility exposes that ceiling fast.

The challenge affects lean treasury organizations the most. Expectations expanded significantly while team sizes didn’t. Treasurers are being asked to provide deeper insight, respond faster to market events and support broader business decisions without a similar increase in resources. That gap between expectation and capacity is becoming a defining issue.

What data and AI actually change

Used well, AI pushes treasury from tactical reporting toward actionable intelligence. In practical terms, AI can:

• Improve forecasting by learning from behavioral patterns across historical flows, rather than relying on static assumptions or payment terms that clients aren’t adhering to in practice.
• Automate recurring work like cash reconciliation and standard reporting, freeing up a small team to spend time on scenario planning instead of commoditized processes and data cleanup. 
• Flag unusual patterns in transaction data that a set rules-based system can miss, helping detect fraud and catch errors – coupled with additional checks and balances from a specialized treasury solution.

AI-driven forecasting doesn't replace human insight; it's designed to help you make faster, better-informed decisions by surfacing relevant data points and correlations that would otherwise stay buried in spreadsheets. The value isn't autonomy, it's leverage – letting the same team see more, sooner, with greater confidence in the underlying numbers.

That’s exactly what treasurers need in a volatile market. Real-time visibility lets you respond to an FX or rate movement as it happens, not after the fact. Predictive analysis lets you model the effect of a trade policy shift before it hits the balance sheet, so contingency plans are in place. None of that is possible without a foundation of clean, connected, accessible data.

How to move forward without betting everything

It’s a natural instinct to wait for a perfect, all-encompassing rollout – but that’s also the slowest path to progress, if it gets off the ground at all. The more pragmatic approach starts small and compounds.

Fix the data foundation first. AI applied to fragmented data amplifies the fragmentation. Before layering intelligence on top, work toward consolidated, connected data across your banking relationships and systems. It’s not glamorous, but it determines whether everything after it performs to the level you need.

Start with one high-value use case. Cash flow forecasting is a natural entry point: it often carries a lot of manual effort, provides the highest payoff from accuracy, and has the C-suite’s attention. Prove the value in one region, measure it against a clear benchmark like forecast to actuals to gain confidence and trust in both the data and models, then expand.

Treat a small team as a design constraint. The goal of automation isn't headcount reduction. It's freeing up the bandwidth for stretched employees to take on the strategic work they were trained for while promoting job satisfaction – critical in today’s competitive skills market.

Keep governance in the room. In regulated finance, an AI capability has to be explainable, trusted and fit for purpose. Build oversight in from the start rather than retrofitting it after deployment, and choose a vendor partner with the global structure in place to maintain this process into the future.
​
The treasurers who succeed in the AI era won't be those who chase every new capability – or who wait for certainty. They'll be the ones who build trust in their data, target high-value outcomes and scale adoption through proven results.