Srinivasan Seshadri, Chief Growth Officer and Global Head of Financial Services at HCLTech, opened the session by describing autonomous financial systems “like self-driving cars,” exhilarating in potential, but only if they “don’t hit the walls of compliance.” AI, he argued, is now “in the flow of money,” routing transactions, resolving disputes and shaping customer journeys.
Consumers already expect “instant and embedded interactions at checkout and in-app,” yet “trust, transparency and resilience haven’t quite caught up with the same speed.”
HCLTech’s latest research, The Future of Payments: AI everywhere. Trust nowhere? underlines the tension: almost all payment leaders use AI somewhere in their operations and over half expect to operate as autonomous financial services organizations within two years, but a minority have the modern data backbone to make autonomy safe, trustworthy and scalable.
What makes autonomy different from automation?
Seshadri’s first question pressed for a definition: how do today’s autonomous payments differ from earlier waves of automation?
For ING’s Roel Huisman, innovation “should never be about innovation,” but purpose; better experiences and efficiency together. He described autonomy as a staged evolution: “going all the way back to scheduled payments…direct debits…event-driven or trigger-based payments,” and now “the decision logic of Agentic AI,” not revolution “falling from the sky,” but maturity across layers. The “stars are…aligned,” he said, as instant payments infrastructure, open banking, IoT and AI converge to make payments “more frictionless...and better for our customers.”
HCLTech’s data matches that arc: executives see the defining forces behind autonomous finance as AI decision-making (57%), reduced human intervention (56%) and real-time data orchestration (55%). Yet 47% still lack formal AI governance, amplifying the execution gap as autonomy scales.
The backbone: Instant payments and why they matter
Real-time is the backbone, explained Huisman. Always-on, high-volume/low-value instant payments reduce marginal costs and unlock “new propositions,” including autonomous flows. ING seized its own “lifecycle moment” to build a central instant payment platform across Euro entities to create scale and economies, exactly the plumbing autonomy needed.
The industry knows the stakes. A strong majority of leaders worry they’ll lose customers if they don’t support instant capabilities, even while 82% acknowledge the risks of running real-time, 24/7 systems.
Embedded by nature, or by design?
Turning to merchant solutions, Kilian Thalhammer of Deutsche Bank explained how embedded finance is evolving. “Payments…are embedded by nature,” he said, because they’re inseparable from the underlying process. The goal is to make them “as invisible as possible,” but not at the cost of clarity. Regulation and market drivers sometimes require visibility “to ensure the consumer or the corporate…is aware of what he or she is doing.”
For SMEs, jargon is irrelevant: “make it easier for them to use this technology.” And on the perennial question, when will payments become fully invisible, Thalhammer was clear: “It will never be invisible completely.” When problems surface, including insufficient funds, fraud and dispute, payments must become obvious again.
HCLTech’s research captures that duality: 41% of leaders worry about evolving expectations for “embedded, seamless payments,” and 40% cite the impact of AI agents on customer trust and experience.
Cross-border: Closing the last mile
Is frictionless cross-border a tech, trust or regulatory problem? Dr. Roland Nehl of Commerzbank pointed to the ISO 20022 transition: with adoption around 70% and a push toward full coverage, the industry is building a common language; “standardization [and] interoperability” enables better Straight-Through Processing (STP) and transparency. But once that foundation is in place, the “hard work” begins, including optimizing networks, speeding reconciliation and knitting the “last mile” to domestic instant schemes so payees get funds fast, and know they have them.
It’s an internet-era analogy, Nehl argued that there will be a period of gradual change, before compounding acceleration. Leaders in the report agree that disruption is multifaceted: over half (51%) cite system-level changes, such as instant payments, CBDCs and decentralized/embedded systems as the bigger lift, nearly matching those who fear rapid customer-behavior shifts (49%) like digital wallet adoption and embedded journeys.
Programmability and instant payments: A powerful duo
On programmable money and instant payments, Nehl expects every major region to have an instant scheme “ready at their hand,” with layered services, like request-to-pay in Europe, integrated into real business flows. Helena Forest of Mastercard agreed: “a great duo for the future,” speed from instant, autonomy from programmability. With many countries now living with some form of instant payments and interlinking underway, cross-border low-value commerce will increasingly blend domestic instant, wallet connectivity and, in some corridors, stablecoin infrastructure, depending on the use case.
Leaders see those very trends, digital wallets (56%), instant payment and real-time settlement (55%) and AI-driven optimization (51%), as the most disruptive forces over the next three years.
Control, confidence and the customer
What needs to be solved for autonomy to succeed? “Control,” said Huisman. If agentic decisions trigger payments, customers must be able to configure boundaries and understand liability. The last thing the industry can allow is a sense of lost control.
Thalhammer urged nuance: “instant is always good” is a myth. Contexts exist where delayed or asynchronous flows are better for user comfort or process integrity. This is where “programmable payments…comes into [the] game.” Merchants often don’t need true instant settlement; many consumers perceive payments as instant already. Design should meet the reality outside the payments bubble.
HCLTech’s survey echoes those guardrail concerns: 42% cite data-privacy/security risks from agentic assistants; 40% worry about legacy-system integration and 38% highlight the chance of biased or inaccurate agent decisions. It’s no surprise that 91% are concerned about applying AI in payments operations, even as 82% believe AI is the only viable way to balance frictionless experiences with strong fraud prevention.
Commercializing autonomy: Regulation, resilience and real outcomes
How do you launch account-to-account products that feel intuitive while decisioning moves invisibly in the background? Forest offered three imperatives:
- Regulation and liability: If your fridge orders groceries or your car pays for charging, “who’s actually liable when something goes wrong?” Guardrails must protect trust, because trust drives adoption.
- Operational resilience and data quality: Systems must handle higher volumes, “speed [and] scale,” with machine-legible data across participants
- Outcome-driven design: There’s no one-size-fits-all. Products should be intuitive and safe without becoming uncontrollable. Users must be able to opt in and override.
On risk appetite, Thalhammer suggested that autonomy will get 99.9% of decisions right, but “then you have these one or two cases” that go wrong and banks “need to deal with the problem.” The industry can’t design everything for the 1-in-10,000 edge case, but it must be prepared for it.
The governance backdrop is pivotal. Today, only 20% of firms have fully modernized, cloud native real-time data systems; 47% lack formal AI policies and 60% of leaders rate current AI fraud tools as more ineffective than effective. The ambition to run autonomously — 17% already do and 52% expect to within 24 months — will stall without stronger infrastructure and policy.
Beyond the buzzwords: Where AI helps and where it shouldn’t
Asked about inclusion and personalization, Nehl drew a bright line: core payment execution must be reliable and precise; “if I send you money, you want to have it…exactly,” so AI’s role is strongest around the core: screening, fraud, enrichment and decision support. Forest concurred, pointing to decades of ML in fraud and scam prevention, trained on billions of data points. As autonomy arrives, systems must also “understand” machine-initiated instructions they once treated as adversarial, which is another reason for investment in model governance and explainability.
HCLTech’s findings reinforce the direction: leaders expect the biggest near-term impact from autonomous capabilities to land in real-time fraud detection/resolution (51%), intelligent routing (47%) and automated compliance/reporting (47%).
Invisible when it should be, obvious when it must be
Seshadri closed with a measured verdict: AI is real, here and accelerating. In the medium term, we will be almost autonomous and almost invisible, by design. The job now is to align the stack he sketched, including open banking, blockchain, IoT and APIs on modern cloud with robust orchestration, so that autonomous decisioning, voice-initiated payments and predictive cash flows are matched by “embedded trust.”
HCLTech’s research offers both urgency and a map. Leaders overwhelmingly prefer bold innovation, yet readiness lags; 87% fear losing customers without instant capability, but 43% say regulatory uncertainty is the disruption they’re least prepared for. The winners will move from experimentation to execution: investing in resilient data, formal AI governance and outcome-driven design that keeps customers comfortably in control even as the transaction fades into the background. That’s how invisible payments create visible value.
