A2A payments overview | the impact of Agentic AI and automated purchasing

A2A payments are projected to reach USD 1.4 trillion by 2029 from consumer transactions alone, and Agentic AI could be the catalyst that unlocks that growth beyond P2P.

This article is part of The Paypers’ four-part series celebrating the anniversary of our ‘Account-to-Account Payments Report’. We have reached out to experts to take the pulse of the industry and see how it is progressing in 2026, asking four main questions: 

1. Where is A2A payments adoption currently accelerating, and where is it still struggling? Why?
2. How does the rise of agentic AI and automated purchasing change the opportunities and risks around A2A?
3. Which regulatory changes are expected to further influence the real-time payments sector? 
4. What is the biggest challenge preventing A2A from displacing card payments at scale, and how can it be solved? 

Discover what they have to say about the impact of Agentic AI and autonomous purchasing on A2A payments below, and keep an eye out for The Paypers’ Key Themes page on A2A Payments this week to see what they share about the other topics.

How does the rise of agentic AI and automated purchasing change the opportunities and risks around A2A?

Duygu Inanc Koyunpınar, Head of Product, DIMOCO: Agentic AI increases the relevance of A2A by favouring payment methods that are fast, deterministic, and cost-efficient. Real-time confirmation and instant settlement make A2A well-suited for automated transactions where certainty is critical. However, the effectiveness of these flows will depend heavily on underlying banking experiences. Innovative banks are better positioned to support API-driven interactions required for automation, whereas traditional banking infrastructures may introduce friction or inconsistency. This creates both an opportunity to scale A2A in more advanced ecosystems and a risk of fragmentation where banking capabilities lag behind automation needs.

Mark Beresford, Director, Edgar, Dunn & Company: There is no doubt that agentic commerce is the exciting next wave of shopping, where AI agents increasingly discover, compare, negotiate, and buy on behalf of consumers and businesses. This is already raising expectations for a far more seamless, personalised, and automated commerce experience within the next few years. Agentic commerce makes A2A more attractive in certain consumer journeys, increasing the case for A2A in automated, low-friction, high-frequency purchases. There are huge questions around consent, fraud, disputes, liability, protocol compatibility, and merchant readiness. If an AI agent is choosing the payment method, merchant and consumer logic may favour fast, cheap bank-to-bank (A2A) settlement over traditional payment cards, especially where A2A is instant and confirmable.

Dylan Massey, Co-Founder & CEO, Interchecks: Agentic AI introduces a genuinely new challenge for A2A: trust. Consumers must feel comfortable sharing payment credentials with an agent acting on their behalf, and merchants accepting payments initiated by one. Neither is a given right now, and the hesitation is heightened with A2A's current limitations: bank payment acceptance for merchants is not universal, and A2A rails are largely regional, meaning cross-border transactions have routing complexities that even human-initiated payments struggle with today. Before agentic AI can meaningfully impact A2A, the industry first needs to address merchant adoption and infrastructure interoperability. 

Keith Olson, VP of ACH & Open Banking, Nuvei: As purchasing shifts from human-led decisions to automated execution, payment requirements change. Predictability, confirmation speed, and cost efficiency become more important than brand or form factor. This plays to A2A’s strengths, particularly in high-frequency transactions and B2B environments where margins are tight and timing matters. While removing friction in settlement has a direct impact on operational efficiency, the challenge is governance. Automated payments require clear controls, visibility, and dispute handling that can operate without human intervention. The infrastructure needs to support both execution and trust at scale. 

Magnus Bergaplass, Head of Mobile Payments, Merchant Acquiring/PF&I, Worldline:  Agentic AI can automate payment decisions only within strict consent and authentication controls. It relies on user-granted permissions and explicit payer consent flows, integrated with strong customer authentication (SCA) at initiation and during any sensitive steps. AI can handle routing, risk checks, and timing while transparently logging decisions and requiring re-authentication for high-risk actions. Governance, explainability, and auditable AI decisions ensure compliance with Open Banking and PSD2/SCA rules, so automation enhances efficiency without bypassing required consent or security. Without strong governance, risks include automation bias, fragmented decision trails, and potential security gaps. 

Justin Fraser, Chief Revenue Officer, Yaspa: Agentic AI impacts purchasing from decision to delegation, changing everything for payments. When AI acts on behalf of a user, it needs fast, consent-native, programmable rails. A2A is built for this, as its consent architecture creates a trust framework that card rails can't replicate for non-human actors. In iGaming, we're seeing early signs, such as AI-optimised deposit management, session-aware funding, and predictive cashier routing. However, the risk landscape is new: liability when an agent authorises a transaction the user didn't explicitly intend, or applying SCA to non-human actors. A2A can be the trust layer for agentic commerce if built right. 

Tarik Zerkti, CEO, MyBank: Agentic AI modifies A2A payments from user-initiated flows to machine-executed transactions. The issue? PSD2 SCA is built around human authentication (biometrics, OTP), whereas agents require persistent, delegated authorisation without repeated user involvement. PSD3 and the PSR improve API reliability, standardisation, and consent management, yet still largely assume a human in the loop, and do not fully enable machine-native authorisation. Moreover, A2A’s instant and irrevocable nature amplifies the impact of agent errors or manipulation. As a result, A2A’s opportunity in autonomous commerce, while strong, is dependent on evolving SCA frameworks and liability models under PSD3/PSR. 

Tareq Shaheen, Product Development Director, Payment Solutions, Eastnets: A2A payments are well-positioned to become the preferred payment rail for AI agents, as they are inherently more programmable and potentially more cost-effective than traditional card networks. However, removing the human from the payment decision creates challenges around consent, authentication, malicious agents, fraud prevention, transaction monitoring, and liability. If autonomous agents are allowed to initiate payments, the market requires stronger guardrails around delegated authority, payee verification, spending limits, and exception handling. While the opportunity is substantial, scale depends on trust, control, and auditability, supported by clear regulatory upgrades around Open Banking and instant payments. 

Martín Azcue López, Business Development Director, Bizum: As with any other payment instrument, AI and agentic technologies will significantly accelerate developments, enabling fully customised and automated payments. Although A2A solutions are still relatively young compared to others (Bizum turns 10 in 2026) and still have basic needs to cover, they must keep up with market demands and explore embedding these technologies into their services. AI enables purchase automation, allowing users to set payment orders, like target prices for plane tickets, which are then paid automatically if found by AI agents. Additionally, AI catalyses fraud prevention, helping teams to identify undetected and rapidly evolving data trends. 

Roy Prayikulam, SVP Risk and Fraud Division, INFORM: Agentic AI will make payments increasingly frictionless by embedding A2A directly into automated purchasing flows, and enabling banks to capture more ecommerce volume. For consumers, the payment method becomes almost invisible. However, this also expands the fraud landscape. AI-driven scams, such as fake merchants or automated social engineering, can scale rapidly, and in A2A environments, funds are often irrecoverable. As banks expand further into ecommerce, they increasingly take on risks traditionally managed by card networks, making real-time ecommerce fraud detection and prevention essential. 

Gabriel Lucas, Director, Redbridge Debt and Treasury Advisory: If purchasing becomes increasingly automated, payments will need to be instant, secure, low-cost, and easily integrated into digital workflows. In that sense, A2A payments are naturally well-suited. At the same time, automation raises new questions around authorisation, fraud prevention, and accountability. When transactions are triggered by software agents rather than humans, ensuring strong authentication, clear consent frameworks, and robust risk management will be even more important to maintain trust in the system. Ultimately, payments have never been only about making a transaction work, but also about ensuring trust and reliability.

 

Don’t miss out on what’s coming tomorrow: which regulatory changes are expected to further influence the real-time payments sector? 

Be a part of this conversation

As you read this, we at The Paypers are already at work on the third edition of the A2A Payments Report, which will be launched this summer. If you’re a solution or technology provider willing to share your expertise with other professionals in this space, we’d love to hear from you! Reach out to sales@thepaypers.com