How CFOs Are Rewiring Corporate Cash Management for 2026

For years, the treasury department was managed on spreadsheets, gut instinct, and countless hours of overtime. A finance team would spend days pulling numbers from a dozen bank portals, feeding them into a forecast, and hoping nothing changed before Monday’s board meeting. That world is fading fast, and most CFOs have barely had time to notice.

 Corporate treasurers aren’t just experimenting with artificial intelligence. According to research conducted by the Association for Financial Professionals, 52% of US corporate treasurers are now piloting or have incorporated AI tools for cash forecasting, nearly double the figure from just two years ago.

 What began as a handful of curious finance teams testing chatbots has turned into a genuine shift in how companies manage liquidity.

From Spreadsheets to Systems That Watch Themselves

The old approach to forecasting was dependent on static models that were made once a quarter and quietly ignored the moment they went stale. Agentic AI works differently. Rather than producing a single forecast and waiting for someone to update it, these systems pull together:

●      Live bank feeds

●      ERP data

●      Accounts receivable and payable ageing

 The result is a rolling, constantly updating picture of the company's cash position rather than a snapshot that's already out of date by the time anyone reads it. Ivo Bozukov puts it plainly. "The treasurers who are getting real value out of this aren't the ones chasing the flashiest tool," he says. "They're the ones who insisted on governance and data quality before anything else went live."

 That distinction matters because the results genuinely vary. In production environments observed through 2026, agentic 13-week cash forecasts have reached accuracy rates of between 88% and 92%, but only when it’s based on three things:

●      Solid bank data

●      Proper ERP context

●      Deterministic controls, rather than a model left to guess on its own

The Bank-Led Push

Some of the clearest evidence of this shift has come from the banks themselves. JPMorgan’s Cash Flow Intelligence tool, now used by roughly 2,500 corporate clients, has reportedly cut manual forecasting work by close to 90% for some of its users. What used to take a treasury analyst the better part of a week can now be produced, checked, and adjusted in a fraction of that time.

 Machine learning models that are trained on years of historical payment behaviour are proving to be much better at spotting seasonal quirks and customer-specific payment habits than the rule-based systems treasury teams relied on previously. Meanwhile, large language models are increasingly being used to:

●      Generate plain-English summaries of foreign exchange exposure

●      Suggest hedging options based on that exposure

 That turns what used to be a dense analysis report into something a CFO can actually read between meetings. Ivaylo Bozoukov notes that this isn't simply about speed. "Cutting the manual work is nice, but the real change is that finance leaders are finally seeing their cash position as it actually is, not as it was three days ago," he explains.

Why the Governance Question Now Matters Most

According to the recent findings from PwC, 2026 is shaping up to be the year AI-driven treasury automation moves out of pilot programmes and into daily operations. That shift changes the conversation CFOs need to be having.

 It’s no longer a question of whether it’s good to integrate AI in treasury operations. Most organisations already have, in some form.

 The harder question, as Bozukov puts it, is "whether the system was actually built with proper controls from the start, or whether governance was bolted on afterwards because someone in compliance finally asked about it." Automated payment routing and intelligent liquidity positioning only deliver real value when the underlying architecture can be trusted, audited and explained.

Watching Closely From Outside the US

While much of this transformation has played out in developed markets, CFOs across the GCC, East Africa and Latin America are paying close attention. Organisations managing multi-currency operations and fragmented banking infrastructure stand to gain even more from real-time visibility than companies operating within a single, well-established banking system.

 The spreadsheet-driven treasury desk isn't disappearing overnight. But for CFOs still treating AI adoption as a future project rather than a current one, the gap between them and their better-prepared peers is growing every quarter.