How should treasurers be using AI?

Artificial intelligence has become a fixture of treasury conversations. Most teams agree it will change how they work, but many are still unsure exactly how. A wide gap remains between interest and practical use, and treasury is a function where mistakes cost more than in most other areas.

This guide sets out where AI can help treasurers today, including in cash investment decisions. It also covers where to be cautious, how to govern AI before using it more widely, and how to get started safely.

Where treasury stands on AI today

AI adoption in treasury is underway, but it remains at an early stage. PwC's 2025 Global Treasury Survey found that 74% of treasury teams are either expanding or actively using AI, yet only 26% rate their AI processes as mature. Coalition Greenwich research from early 2026 showed that fewer than one in 10 teams use AI for their core tasks.

US treasury teams are moving AI up their agenda quickly. NeuGroup's 2026 Treasury Outlook Survey ranked AI enablement as the second-highest treasury priority for 2026, behind only cash flow forecasting, after it failed to reach the top five the previous year.

Governance has not kept pace. The 2026 AFP Treasury Benchmarking Survey, which drew primarily on US-based treasury practitioners, gave AI and emerging technology policies an effectiveness score of 2.9 out of 5, the lowest of any policy area measured. Cash management policies, by contrast, scored 4.4. Our own AI in Treasury Report found a similar pattern: treasurers often trust AI least for the tasks they most want help with.

Where AI adds value in treasury

AI performs best on tasks that are data-heavy, frequent, and easy to review. Much of the treasury workload fits that description, so the biggest early gains come from day-to-day operations rather than strategy.

Cash flow forecasting

Forecasting is the most widely cited use of AI in treasury across the major surveys, and the 2026 AFP Treasury Benchmarking Survey named cash and liquidity forecasting as treasury's most challenging task. Machine learning can identify patterns in historical receipts and payments that spreadsheets miss, including seasonal trends and customer payment behavior. AI can also explain variances between forecast and actual cash, showing teams where their assumptions may be off.

Human judgment is still needed for one-off events. Acquisitions, quarterly estimated tax payments, or changes in supplier terms can confuse models trained on normal conditions. Many teams run AI forecasts alongside their existing methods until the models show consistent accuracy.

Reconciliation and exception handling

Matching bank statements to ledger entries is repetitive and rules-based, which makes it well suited to automation. AI can clear routine matches and route only the exceptions to a person for review. This frees up analyst time and stops errors from compounding.

Reporting and portfolio queries

Generative AI can turn raw holdings data into clear answers and draft reports. Treasurers can ask which investments mature this month or how exposure is split by counterparty without exporting data to a spreadsheet. Our article on using AI to query portfolio data explains how this works in practice using TreasurySpring’s MCP.

Fraud and payment anomaly detection

Payment fraud remains a significant risk for US companies, and AI can flag unusual transactions before they are released. The 2026 AFP Payments Fraud and Control Survey found that 76% of US organizations experienced attempted or actual payments fraud in 2025, and business email compromise affected 74%. Only 17% were using AI to combat it.

The same survey identified treasury as the department most likely to uncover attempted fraud. That places treasury teams in a strong position to lead adoption of AI-based detection across the wider finance function.

Using AI in cash investment decisions

Cash investment is another area where AI could add significant value., and also where treasurers need to be most careful. Decisions about surplus cash involve counterparty risk, liquidity needs, and policy limits, and errors here can have serious consequences.

Monitoring policy and concentration

AI is well suited to checking whether a portfolio complies with its rules. It can compare current holdings against the limits set out in the investment policy statement and warn when exposure to a single bank or sector approaches a limit, before it is breached.

This kind of monitoring depends on clear, granular data. Pooled products that combine many underlying exposures are harder for any tool to analyze, while single-issuer investments make each exposure easy to see. The more transparent the structure, the more reliable AI oversight becomes.

Keeping people in charge of the decision

AI can support higher-level work such as modeling liquidity scenarios and comparing investments on a risk-adjusted basis. The treasurer should still make the final decision on where cash goes, through an approval process that can be audited later. Our research found that only 4% of treasurers want AI to execute transactions, a clear sign that most prefer caution.

Most teams want AI to strengthen their judgment without giving up control. A model can suggest a counterparty, but it can't be held accountable for that choice. The treasurer who approves it can.

Where treasurers should hold back

Some tasks carry too much risk for AI to handle alone. Releasing payments, executing hedges, and placing investments all need close human oversight, because mistakes are expensive and difficult to reverse.

Generative AI can give different answers to the same question, which sits poorly with treasury's need for clear audit trails, particularly for companies whose controls are tested under Sarbanes-Oxley. Treasurers should not rely on AI to forecast interest rates or market movements, since a confident answer can mask real uncertainty. Never enter sensitive data such as bank balances and counterparty positions into public AI tools unless appropriate controls are in place.

Governance comes before scale

Most treasury teams already have well-established policies for cash, counterparties, and payments. AI should operate within those same rules, with clear guidelines on which tools are approved and what data they can access.

Building AI into existing policy

A treasury AI policy should set out who can use each tool, which decisions require human approval, and how outputs are recorded. It should also explain how models are tested before use and reviewed as conditions change. Involving IT, risk, internal audit, and legal early helps prevent informal, unapproved tool use.

Data comes first

AI is only as effective as the data it works from, with poor data quality being a major obstacle. PwC found that 76% of treasurers cite poor data quality as a barrier to better forecasting. Consistent cash flow categories, reliable bank connectivity, and a single source of truth for positions will improve AI results more than switching models.

Where to start with AI in treasury

Start small and build trust gradually. Teams can move from personal productivity gains to supporting larger decisions as their controls mature, treating each stage as a step on a ladder rather than a single leap.

Low-risk first steps include summarizing market commentary, drafting policy language, and querying existing data in plain language. From there, teams can automate reconciliation and benchmark AI forecasts against their current methods. Support for liquidity and investment decisions should come later, once the data, controls, and confidence are in place.

For teams starting with plain-language queries, our guide to the TreasurySpring MCP server shows how to connect investment data to an AI assistant with read-only access. Measure results at every stage. A baseline for forecast accuracy or time spent on reconciliation shows whether AI is making a measurable difference.

Putting AI to work in treasury

AI is now a permanent part of the treasury landscape. Fully autonomous systems remain some way off, but treasurers already have practical use cases to build on. Starting with simple, repetitive tasks allows a team to gather evidence of value before extending AI to higher-stakes work.

If you would like to see how AI can support your cash investment process, contact our team. We can show you how our platform is designed to connect your portfolio data to the tools your team already uses.

For anyone who can't tell the difference between their COP and their CCP, we've introduced our Treasury & Cash Investment Glossary.

*TreasurySpring’s blogs and commentaries are provided for general information purposes only, and do not constitute legal, investment or other advice.