Manoj Dugar, head of Global Payments Solutions, Asia ex Greater China, HSBC. Photo: HSBC

Ask treasurers what they want to be known for, and rarely will you hear “my team reconciles faster than all others.” Rather, treasurers aspire to be the trusted advisor in the C-suite, helping to navigate market uncertainty, shape investment decisions, and guide the company towards growth.

Yet for many treasury teams, the daily reality is still dominated by operational work: collecting data from multiple sources, patching gaps with spreadsheets, and producing reports under time pressure. This struggle between strategic ambition and operational burden is why AI adoption in treasury is poised for its hockey-stick moment.

Treasury professionals certainly see the promise in AI: more than 7 in 10 believe that the technology can replace up to 25 per cent of treasury tasks over the next five years. They are also clear about where AI can bring the most value – cash-flow forecasting, followed by receivables reconciliation, and FX risk management, areas where better prediction, faster handling, and smarter detection can directly lead to improved liquidity and reduced risk.

However, there is a clear gap between promise and action: 1 in 2 companies have yet to deploy AI at all in treasury, and less than 1 in 10 have implemented AI into daily treasury workflows.

While strong conviction is currently met with limited execution, there is every reason to believe that AI adoption in treasury can, will, and must take off.

Start with “can”. We know from our recent treasury roadshows across Asia-Pacific that the top three hurdles to implementing AI are integration with existing systems (70 per cent), cost (44 per cent), and expertise (41 per cent).

Integration is the stand-out for a reason. Most large corporates rely on ERP and TMS software systems that were built for a pre-AI era. Connecting modern AI tools to this environment requires significant technical effort and careful change management. In addition, data are usually held across dozens of bank portals and software systems, with much of it inconsistent or incomplete. AI is only as good as the data it accesses – and treasury data have historically been fragmented.

The skills gap is also real. Treasury teams are rarely stocked with AI specialists, and many functions rely on self-learning. Add in the need for governance, and it is understandable why many organisations are still piloting rather than scaling.

That said, these hurdles are not arguments against adoption. They are merely familiar friction points of every major technology shift and can be overcome.

If history is a guide, AI adoption in treasury will take off. Treasury technology does not change overnight. It changes when connectivity becomes easier, when standards emerge, and when enough organisations demonstrate value that others can implement with confidence.

That is exactly what has happened with APIs, or application programming interfaces. APIs have transformed to become the practical, high-return plumbing that underpins automation, reconciliation, and real-time visibility. And the pace of change is measurable. We have seen at first hand an exponential increase in our API call volume over the past few years – and 34 per cent on-year. As at-scale API adoption has shown, better connectivity and greater automation can lead to reduced processing time, lower error rate, and higher forecasting accuracy.

Finally, the reason AI adoption in treasury must take off is that there is simply no alternative.

Treasury teams are leaner than ever: many have less than 10 full-time employees worldwide. At the same time, the expectations placed on treasury teams have only increased. Treasurers are being asked not only to keep the lights on, but also to help navigate market uncertainty, shape investment decisions, and guide the company towards growth – in short, to become a more strategic function.

While the workload has been increasing, manual spreadsheets are still deeply embedded in treasury workflows. They remain the primary technology in nearly 1 in 3 treasury functions, rising to 54 per cent for cash-flow forecasting and 45 per cent for reporting and analytics.

The tension between being strategic and being operational. Treasury teams cannot strategise their way out of their operation workload, but they can automate away repetitive tasks by using technology such as AI and APIs.

While APIs can remove friction by connecting systems and enabling straight-through processing, AI can sit on top of these flows to improve forecasting, highlight anomalies, accelerate analysis, and triage exceptions – thus freeing up humans to focus on judgement, stakeholder management, and decision-making.

True, AI adoption in treasury is still in the “promise exceeds action” phase, but the scene is set for its hockey-stick moment. As a treasurer, you would not want your claim to fame to be the “fastest reconciliation in the industry,” would you?