Why Disconnected Finance Data Is Holding Back AI in Payments

New research from National Technology News and Bottomline shows how connected finance data, cash visibility and reconciliation underpin reliable AI-driven improvements in B2B payments operations.

The heat is on to justify AI investments with measurable results, yet there’s also a distinct FOMO feeling around AI adoption. New research suggests that many organisations are moving ahead on AI without first addressing the fragmented data, systems, and processes that lift or limit its potential to transform B2B payments.

Taking this matter to the marketplace, National Technology News and Bottomline surveyed over 400 finance, treasury and accounting professionals in the UK and U.S. for The Connected Finance Function: Visibility, Control and AI Readiness. The findings reveal an office of the CFO function caught between rising AI hopes and hard operational truths.

It’s not that finance is lacking in AI adoption, and the report finds 72% face at least some executive pressure to introduce AI-enabled finance capabilities within 12–24 months, and 41% face a mandate or strong expectation for AI-enabled finance capabilities within 12–24 months. It’s that finance should favor connectedness and cleaner data before AI can produce its best results at scale.

 

When Forecasting Collides with AI’s Limited Visibility

Increasing confidence in short and medium-term cash forecasts is the leading priority for 48% of responding finance leaders over the next 12 to 18 months. It’s a finance function where an AI running clean, reliable data can substantially improve accuracy.

That priority reflects how the CFO’s role is changing. Payments and finance leaders are now expected to provide forward-looking insights that guide decisions about liquidity, investment, and risk with a degree of precision, and not just report historical performance.

However, just 40% of respondents said their four-week cash forecasts are highly reliable (22%), or even reliable enough (18%) to support most operational and liquidity decisions.  Leaving 59% challenged with a collection of accuracy, inconsistent data, disconnected processes or requiring manual adjustment.

A major underlying problem is visibility. Nearly two-thirds of respondents lack consistent, organisation-wide visibility into cash positions across their banks and entities. Only 19% have achieved near real-time, consolidated visibility.

“Reliable forecasts cannot exist without reliable operational data,” the report notes.

Put another way, forecasting can’t be improved in isolation. If cash positions, receivables, payments, and treasury information remain dispersed across multiple systems, even sophisticated forecasting models will be working with incomplete or delayed inputs.

As expressed in the report, Bottomline’s view is that cash visibility and forecasting should be treated as parts of the same transformation. A reliable opening cash position, combined with connected data on expected inflows and outflows, provides the basis for forecasts that decisionmakers can trust.

"In many of the conversations I have with customers, the automation of bank account reconciliation is their entry point to AI adoption, i.e., automating what the human does, and applying rules pre-configured by the human" said Kevin Grant, Managing Director, Treasury Management at Bottomline.  

"This is a logical, observable first step on the journey to wider large language machine models and goal orientated agentic AI. To enable it, connectivity to banks and back-end systems are a pre-requisite," Grant added. "Detailed payroll, accounts receivable, accounts payable, and all other forecast actual movements in or out of the bank account are required. Automated import of bank statements is the final piece of the jigsaw to facilitate the AI." 

 

Finance Teams are Acting as the Integration Layer

The operational cost of fragmentation is as significant as the data problem. More than two-thirds of respondents said that over a quarter of their operations time is spent on manual reconciliations, exception handling, and workaround processes. For roughly four in six respondents, those activities consume more than 40% of available time.

Finance professionals are effectively moving information between systems, validating transactions, and resolving discrepancies that connected workflows can address earlier in the process. This reduces the time available for liquidity planning, risk management, and strategic analysis. It also creates wider consequences, the study states.

Respondents link disconnected finance processes to low forecasting confidence, increased audit effort, control gaps, slow investigations and delayed cash visibility.

“Fragmented systems create fragmented decision-making,” per the report. Automating individual tasks won’t solve that problem if information must still be passed manually between AR, AP, treasury, banks, and ERP platforms. Sustainable improvement requires a connected operating model built around the entire cash lifecycle.

It’s an important distinction. Finance transformation should not be measured only by the number of processes digitised. The more meaningful question is whether those processes now share trusted data and support a consistent view of cash.

 

AI Expectations are Outpacing Readiness

More than 70% of respondents report at least moderate leadership expectations for AI implementation over the next 12 to 24 months. Compare that to the 22% who say AI is embedded in finance operations today, regularly supporting decisions and process execution. Nearly 60% said they’re stuck in pilot, evaluation, or non-adoption stages.

There’s also a striking investment gap. Some 45% of organisations have no significant technology changes planned, while none expect to make a major decision related to cash visibility, finance controls or automation within the next six months.

Many finance leaders are therefore being asked to demonstrate AI value without the investment, deep integration, or data foundation required to quantify value.

The survey explains that finance professionals understand the risk. Fragmented, incomplete or poor-quality data is their leading concern about AI, cited by 61% of respondents. The risk of inaccurate or misleading outputs follows at 60%.

“Better models and more sophisticated analysis cannot compensate for poor-quality operational data,” the report states.

Encouragingly, respondents are focused on practical use cases. Improving matching and reconciliation accuracy tops the list at 64%, followed by forecasting and scenario analysis, anomaly detection, and exception prioritization, at 49% each.

There’s a signal in the order of those priorities. Better reconciliation produces cleaner data. Cleaner data improves cash visibility. Better visibility strengthens forecasting and makes further automation safer. The strongest AI strategy may therefore begin with solving an operational problem, rather than a sprawling, ambitious technology programme.

"Fragmented technology landscapes create islands of data," Grant said. "Integration capabilities and connectivity-as-a-service can remove this barrier and oblige a trusted provider to maintain the integrity of data flow."

He noted that "Focusing on AI, the longstanding computer science tenet of the output quality of any system or computer model is correlated to the quality and timeliness of the input data holds true."

 

Build the Foundation, Scale the Value

The Connected Finance Function: Visibility, Control and AI Readiness presents finance leaders with a choice. They can layer AI onto disjointed systems and risk producing unreliable outcomes faster, or they can connect finance operations and create the trusted data foundation AI requires to become more trusted.

Bottomline’s approach is practical and incremental. Start with measurable outcomes like reducing manual reconciliation, improving cash visibility, strengthening forecast confidence, and directing human attention to the exceptions that require investigation.

The full report examines these findings in precise detail, including differences between UK and U.S. respondents, the leading barriers to adoption, and the technology decisions likely to shape the next stage of finance transformation.

Download the full report here