Written by Nick Botha, VP Payments & Retail Banking at AutoRek
The payments industry is racing towards increased innovation across AI, instant settlement, open finance, and digital wallets, while simultaneously facing pressures of tighter regulations and safeguarding expectations.
However, payments companies still face operational burdens. While adoption rates rapidly rise, consuming more technology innovation, the pressure points are limiting the full potential of these developments.
Take AI for example. AutoRek’s 2026 Payments Survey revealed that 96% of payments firms have adopted AI in some capacity. But high adoption rates only tell part of the story.
Most payments firms have implemented AI somewhere in their operations, yet the research also reveals that only 30% use it widely across financial operations. The gap between experimentation and enterprise-grade deployment is where value is being lost.
Many firms are treating AI as a tactical solution to isolated problems rather than as a strategic capability requiring robust foundations. They’re deploying machine learning models for fraud detection, implementing chatbots for customer service, or experimenting with predictive analytics. But these initiatives often remain siloed, disconnected from core operational systems, and lacking the clean data foundations needed to deliver meaningful ROI.
The result is a misalignment between strategic ambition and operational readiness.
Is our data clean enough?
Before payments firms invest in and implement AI, they should be asking: Is our data consolidated, clear and clean enough to make our AI investment valuable?
AI technology is based on pattern recognition. It learns from the data it’s fed and amplifies whatever signals exist within that data – whether accurate or flawed. When data is fragmented, inconsistent, or ungoverned, AI doesn’t just underperform, it amplifies existing inefficiencies.
Consider the impact this may have on exception management. An AI model trained on fragmented transaction data – inconsistent formatting, missing fields, reconciliation gaps – will generate false positives that burden operations teams and erode customer trust. In these instances, the AI model isn’t failing, it’s doing exactly what it’s designed to do: learning from the data it’s given. If that data reflects operational dysfunction, the AI will automate that dysfunction at scale.
This presents a significant risk to the industry. Our survey reveals that 80% of payments organisations experience moderate to significant operational disruption from fragmented data. Yet many are still deploying AI on top of these foundations, expecting technology to produce results when fundamental operational problems still exist.
AI doesn’t solve data problems: it exposes them. And in payments, where data flows across multiple systems, geographies, and legacy platforms, those problems can cause significant operational disruption.
Simply put: Clean data in. Clean reporting out.
Payments ambition solved
The UK payments sector is showing considerable ambition in adopting real-time payments, AI, and emerging settlement rails, reinforcing its position as a leading global payments hub. However, sustained competitiveness won’t just depend on product and market expansion, but on ensuring that operational capabilities, data infrastructure, and governance frameworks evolve with front-end innovation.
The firms seeing the strongest returns from AI aren’t necessarily those with the most sophisticated algorithms, it’s the ones that invested in data foundations first.
This includes automated reconciliation processes that ensure transaction data is accurate and complete across systems. It requires standardised data formats that enable integration and analysis. Most importantly, it demands unified data foundations with proper governance frameworks that ensure AI models are trained on reliable, compliant data.
The regulatory imperative
These data foundations don’t just enable AI to work, they provide the infrastructure that won’t buckle under regulatory scrutiny. As the CASS 15 May 7th deadline approaches, implementing compliant processes and technologies is vital.
CASS 15 introduces stricter requirements around the speed and accuracy of client money reconciliations, demanding that firms account for safeguarded funds with far greater precision than before. For those still relying on fragmented data, this is a direct compliance risk.
Despite this, only 33% of firms in the survey say they are fully prepared for upcoming compliance requirements. As regulatory expectations increase alongside operational complexity, the gap between front-end innovation and back-office capability becomes a risk to reputation and competitiveness.
Building for scale
As payments infrastructure evolves toward instant settlement and 24/7 availability, operational margins for error are shrinking. Manual processes that were manageable in batch-processing environments become constraints in real-time systems.
The question for payments firms isn’t whether to adopt AI – it’s whether their data infrastructure can support AI that delivers value. Operational foundations matter more than the pace of adoption. Build the infrastructure first, and AI works. Skip it, and adoption statistics won’t convert into business value.