Pay360 2024: How Payments Found Its AI Footing

The payments industry is not, historically, a sector that rushes at new technology. There are good reasons for this: money is involved, regulation is extensive, and the cost of getting it wrong is not a broken user journey but an actual regulatory action or a fraud event. The culture is cautious by design, and when an industry of cautious people gets enthusiastic about something, it means something different than when a Silicon Valley startup is enthusiastic about it.

Pay360 2024, The Payments Association's flagship conference at ExCeL London, had payments people enthusiastic about AI. That felt worth noting.

Not the evangelistic enthusiasm of tech-event keynotes, where everything is going to transform everything by Q4. More the sober, specific enthusiasm of people who've found that something is actually working in their domain and are comparing notes. The difference in atmosphere is noticeable if you've been to both kinds of event.

What AI Is Actually Doing in Payments

The sessions and conversations at Pay360 that covered AI were not primarily about consumer experience or the vision of conversational commerce. They were about the back-end work: fraud detection, compliance automation, anomaly detection, customer support triage. Less photogenic, considerably more production-ready.

Fraud prevention was the most-discussed application, and it's not hard to understand why. Financial crime teams have been using machine learning for pattern detection for years; it's one of the earliest production uses of what we'd now call AI in financial services. What's changed is the sophistication and the adaptability. Revolut grew its customer base 34% in 2024 while increasing its customer support team by only 5% — AI absorbing volume growth without a proportional headcount increase. Monzo, similarly, has been investing in ML to trigger fraud interventions and has reported meaningful improvements in its ability to prevent unauthorised fraud.

The compliance automation conversations were interesting because they're a less visible application of AI but a genuinely costly one for the industry. KYC, AML monitoring, transaction reporting: these generate enormous volumes of work that is rule-based enough to be automatable in significant part. The compliance burden on UK payments firms has grown considerably under successive regulatory iterations, and AI that can absorb a meaningful share of that work has real commercial value.

According to The Payments Association's own survey data, AI was already identified as the leading opportunity by 25% of Pay360 respondents in 2024, with fraud prevention, compliance automation, and operational efficiency as the dominant named applications. That's not speculative enthusiasm; that's people describing what they're actually building.

Open Banking: Still Not Quite There, But Getting Closer

Open banking was the second-biggest named opportunity at Pay360 2024, cited by 14% of survey respondents. The mood on it was somewhere between frustrated and optimistic: frustrated because the consumer proposition still hasn't achieved the kind of seamless everyday experience that would drive mass adoption, optimistic because the infrastructure is genuinely improving.

UK open banking reached approximately 12.1 million user connections by the end of 2024, real progress from the one or two million of a few years earlier, though still a small fraction of the addressable market for a technology that had been live for six years. The API quality and reliability issues that plagued the early years are mostly resolved. What's still missing is the consumer-facing proposition that would make open banking feel as natural as tapping a card.

The AI-open banking intersection is where some of the more interesting conversations were. Open banking data (real-time visibility into account balances, income patterns, spending behaviour) is genuinely useful material for AI systems doing affordability assessments, fraud risk scoring, and personalised financial product recommendations. The combination of API access and ML analysis is where the category starts to become something meaningfully different from traditional banking data work.

What UK Payments Told Us About AI in 2024

The read I took away from Pay360 was that the payments industry's relationship with AI is considerably more mature than the equivalent conversation in commerce or marketing. These teams have been building and deploying ML systems for years. They've had production failures, learned hard lessons about data quality and model drift, and developed the operational discipline that comes from deploying AI in contexts where the failure modes include regulatory action and customer harm.

That maturity is worth paying attention to. When the payments community's consensus is that AI is genuinely valuable for fraud, compliance, and operational efficiency, and cautious about consumer-facing applications until trust infrastructure is built, that's a useful calibration signal for how the wider agentic commerce conversation should be framed.

The payments industry was doing AI before it was called AI. That probably means they have better-calibrated expectations about both what it can do and what it takes to do it properly.

Pay360 2025 would go further still — the agentic payments conversation arrived in full.