AI and Payments

AI is already embedded in many parts of the payments value chain. But today, most implementations focus on supporting transactions rather than driving them. According to the research:

• Nearly all banks now deploy AI powered chatbots for customer service.
• Machine learning models can analyse millions of transactions in real time, identifying unusual behaviour patterns and preventing fraudulent payments before they are completed.  Around 85% of banks expect to use AI for instant fraud and risk resolution
• 77% of retailers use AI to provide 24/7 payment support.  AI systems can support always-on payment assistance, guide customers through checkout processes and help resolve issues such as declined transactions or authentication challenges.
• AI allows organisations to analyse transaction patterns, understand customer behaviour and generate insights that improve payment journeys.  Around 69% of retailers use AI to analyse customer behaviour and transaction data

But the fastest growth is yet to come. Over the next three years, both sectors expect sharp increases in:

• Payments processed autonomously by AI agents
• AI‑driven transaction routing and orchestration
• Personalised, end‑to‑end payment journeys with minimal human intervention

This shift enables agentic AI, systems that do not just recommend actions, but execute them. And once AI agents can initiate and complete payments, commerce itself begins to change. And it is this transition that will begin to turn the dial for AI in payments.

For banks, AI in payments is more than a technology upgrade, it’s a strategic turning point. Payments are shifting from a utility service to a control point for digital ecosystems, and banks that modernise early are already pulling ahead.

The research shows that payment leaders are already pulling ahead by modernising their infrastructure:

• More than half of leading banks are building infrastructure that supports digital assets, compared with only 6% of organisations at an early stage of modernisation.
• Many payment leaders see tokenisation and digital assets to dramatically improve efficiency. According to our research, 64% of leading organisations believe tokenisation and digital asset infrastructure can significantly reduce payment processing costs and settlement times.
• Leaders are far more likely to have cloud native, API first platforms that support ecosystem integration.

This matters because agentic payments demand infrastructure that is, Realtime, Programmable, Data rich and secure by design.

Banks that fail to modernise risk becoming invisible utilities behind AI‑powered commerce journeys. Those that act decisively can embed themselves directly into agentic workflows, enabling AI agents to initiate, route and reconcile payments safely.

In practice, this means banks must:

• Embed their services directly into agentic workflows, enabling AI agents to initiate and reconcile payments safely
• Provide tokenised and programmable money, allowing conditional, rules-based execution
• Use AI to scale trust, enhancing fraud prevention, identity, consent and compliance in real time
• Monetise enriched payment data, enabled by ISO 20022, through value added services

It is no coincidence that 78% of banks expect to use AI to personalise services using behavioural and contextual data, and 71% plan to extract pricing and liquidity insights from payment data. In an agentic world, compliance and intelligence must sit inside the transaction, not around it.

What truly turns the dial for banks?

It’s the shift from being transaction processors to trusted orchestrators of intelligent payment ecosystems, where AI, data and compliance converge inside each transaction.

The implications for retailers are even more immediate. As commerce becomes more digital, automated and personalised, payments are no longer the last step of checkout, they are rapidly becoming a core part of the customer experience.

The research shows that 54% of retail executives believe payment modernisation is critical to the future of their business, and leaders report dramatically stronger outcomes than beginners, including higher conversion rates, lower cart abandonment and increased revenue.

But agentic commerce introduces a new risk: disintermediation at the point of purchase.

AI agents are already emerging that can:

• Compare products and prices
• Optimise for value or convenience
• Complete payments autonomously

If retailers are not embedded into these journeys, they risk losing:

• Visibility at the moment of decision
• Control over the customer experience
• Loyalty, as algorithms, not humans, choose where to transact

Leading retailers are responding by elevating payments from a back‑office function to a core element of customer experience:

• 73% plan to expand stored value wallets
• 84% plan to increase tokenised payments
• 67% plan to expand buy now pay later
• 60% are investing in embedded payments

They are also integrating AI more deeply:

• 64% use AI to detect and resolve fraud in real time
• 62% use AI to enable smart checkout and self-checkout experiences
• Over the next three years, the fastest growing use cases include autonomous payment processing, AI driven routing, and personalised payment journeys

In agentic commerce, payments are no longer the end of the journey, they are part of the experience itself.

Agentic commerce only works when payments are deeply embedded into digital journeys rather than added as a final step.

Yet the research highlights persistent structural challenges:

• 48% of retailers cite fragmented and outdated payment infrastructure as a major issue
• 54% of retailers say limited access to payment data and analytics is holding them back
• Only around half of banks and retailers consider themselves midway or advanced in building payment ecosystems

To support agentic AI at scale, payments must become:

• API‑first and composable, enabling real‑time initiation and reconciliation
• Programmable, supporting conditional and tokenised execution
• Secure by design, with AI‑enabled fraud and access controls
• Data‑rich, leveraging ISO 20022 to provide context, not just confirmation

This is why 51% of banks believe the future winners in payments will be those with the strongest ecosystems, and why leading retailers expect to work with multiple payment partners over the next five years.

KPMG’s research shows that leaders are not winning by chasing novelty. They are focused on four decisive moves:

1. Modern, agent ready payment foundations: Cloud native platforms, real time rails, ISO 20022native data and orchestration layers.
2. Ecosystem led strategies: Payment innovation increasingly depends on collaboration across banks, retailers, fintechs, AI providers and regulators to accelerate innovation and reduce operational complexity.
3. One integrated AI and payments agenda: Leaders treat AI, payments, data and security as a single operating model.
4. Trust by design: Security, consent, explainability and compliance embedded directly into AI driven payment flows. 

KPMG works with banks, retailers and payment system operators globally to help move beyond experimentation and into scaled, commercial reality in AI-enabled payments.

Drawing on deep expertise across payments, AI, regulation and transformation, KPMG helps organisations:

• Define agentic payments and commerce strategies that align business objectives, technology capabilities and ecosystem partnerships.
• Modernise payment architectures, including real time rails, tokenisation and ISO 20022
• Safely integrate payments into AI and agent platforms
• Design and govern ecosystems that balance speed, resilience and trust
• Monetise payment data and insights while embedding compliance and regulatory controls by design.