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,” exciting in potential, but only if they “don’t crash against compliance.” AI, he explained, is now “integrated with the flow of money,” directing transactions, resolving disputes and guiding customer journeys.

Customers already expect “immediate and embedded experiences at checkout and in-app,” but “trust, transparency and resilience haven’t quite kept pace.”

HCLTech’s recent study, The Future of Payments: AI everywhere. Trust nowhere? highlights the issue: nearly all payment executives use AI in their operations and over half expect to become autonomous financial services organizations within two years, but only a few have the modern data infrastructure to make autonomy safe, reliable and scalable.

How is autonomy different from automation?

Seshadri’s first question sought clarification: how are today’s autonomous payments distinct from past automation?

For ING’s Roel Huisman, innovation “should never be about innovation itself,” but about purpose; combining better experiences with efficiency. He described autonomy as a gradual progression: “beginning with scheduled payments…direct debits…event-driven or trigger-based payments,” and now “the decision-making of Agentic AI,” not a radical change “out of nowhere,” but advancement across different stages. The “stars are…aligned,” he said, as instant payments, open banking, IoT and AI combine to make payments “less friction...and better for our customers.”

HCLTech’s data aligns with that view: executives see the key factors behind autonomous finance as AI-based decisions (57%), reduced human involvement (56%) and real-time data handling (55%). Still, 47% do not have official AI management, widening the gap as autonomy increases.

The foundation: Instant payments and their importance

Real-time is the foundation, Huisman said. Always-on, high-volume/low-value instant payments lower costs and enable “new possibilities,” such as autonomous processes. ING took its own “turning point” to develop a central instant payment system across Euro organizations for scale and efficiency, exactly the infrastructure autonomy needs.

The industry recognizes the challenge. Most leaders worry about losing clients if they don’t offer instant options, even as 82% recognize the dangers of running real-time, 24/7 systems.

Embedded by default, or by intention?

Shifting to merchant solutions, Kilian Thalhammer of Deutsche Bank discussed the progression of embedded finance. “Payments…are embedded by default,” he stated, as they cannot be separated from the underlying process. The aim is to make them “as invisible as possible,” without sacrificing clarity. Regulations and market demand can require visibility “so the consumer or business…knows what they’re doing.”

For small and medium businesses, jargon doesn’t matter: “make it easier for them to use this technology.” Regarding the ongoing question of when payments will become completely invisible, Thalhammer was clear: “They will never be fully invisible.” When issues appear, such as insufficient funds, fraud, or disputes, payments need to become clear again.

HCLTech’s study reflects that conflict: 41% of leaders are concerned about new expectations for “embedded, seamless payments,” and 40% mention the effect of AI agents on customer trust and experience.

Cross-border: Solving the last hurdle

Is seamless cross-border mainly a tech, trust or regulatory issue? Dr. Roland Nehl of Commerzbank referred to the ISO 20022 transition: with about 70% adoption and movement toward complete coverage, the industry is creating a shared language; “standardization [and] interoperability” allows better Straight-Through Processing (STP) and transparency. But once that groundwork is finished, the “difficult part” starts, such as improving networks, accelerating reconciliation and connecting the “last mile” to domestic instant systems so recipients get money quickly, and know it is available.

It’s a modern metaphor, Nehl noted there will be a period of slow change, followed by rapid acceleration. Leaders in the report agree that disruption is complex: over half (51%) highlight system-wide changes, like instant payments, CBDCs and decentralized/embedded systems as the major shift, nearly as many as those who fear rapid customer-behavior changes (49%) such as digital wallet usage and embedded journeys.

Programmability and instant payments: A strong pairing

Regarding programmable money and instant payments, Nehl expects each major region to have an instant system “ready to use,” with layered functions, like request-to-pay in Europe, integrated in real business processes. Helena Forest of Mastercard agreed: “a perfect pair for the future,” speed from instant, independence from programmability. With most countries now using some form of instant payment and more connections forming, cross-border small-value trades will increasingly merge local instant, wallet links and, in some cases, stablecoin set-ups, depending on requirements.

Leaders notice these shifts, digital wallets (56%), instant payment and real-time settlement (55%) and AI-enhanced optimization (51%), as the most dramatic factors for the next three years.

Control, certainty and the customer

What needs solving for autonomy to succeed? “Control,” Huisman replied. If agentic actions trigger payments, customers must be able to set boundaries and grasp their responsibilities. The industry must avoid creating a sense of lost control.

Thalhammer urged subtlety: “immediate is always better” is a misconception. Sometimes, delayed or staggered flows are better for user comfort or process integrity. That’s where “programmable payments…take a role.” Merchants don’t always need instant settlement; many buyers already think payments are instant. Design must fit the reality outside the payments world.

HCLTech’s survey repeats those safety concerns: 42% note data-privacy/security risks from agentic assistants; 40% worry about old-system integration and 38% mention the risk of biased or wrong agent choices. Unsurprisingly, 91% are concerned about using AI in payments, even as 82% think AI is the only real way to offer seamless experiences with strong fraud controls.

Making autonomy work: Law, strength and real value

How do you release account-to-account products that feel simple while automated decisions happen in the background? Forest offered three key points:

1. Rules and responsibility: If your fridge buys groceries or your car pays to recharge, “who is really responsible if something fails?” Clear guidelines protect trust, because trust encourages use.
2. System resilience and data quality: Systems must stand up to bigger traffic, “speed [and] scale,” sharing clear data between all parties
3. Design for results: There is no universal solution. Products should be easy and safe without going out of control. Users need to opt in and take charge when needed.

On risk, Thalhammer thought autonomy will succeed 99.9% of the time, but “then you have those one or two exceptions” where things go wrong and banks “need to fix it.” The industry can’t design for every rare event, but must be ready for them.

The policy setting is crucial. So far, only 20% of firms have fully updated, modern real-time data systems; 47% still lack clear AI policy and 60% of leaders think current AI anti-fraud tools are less effective than they should be. The hope to operate as autonomous — 17% do now and 52% hope to within two years — will stall without better systems and rules.

Beyond jargon: Where AI helps and where it shouldn’t

Asked about fairness and personalization, Nehl drew a clear distinction: core payment functions have to be trustworthy and exact; “if I send you money, you want it…accurately,” so AI is most powerful around the core: screening, anti-fraud, enrichment and decision aid. Forest agreed, mentioning years of ML in fighting fraud and scams, trained on billions of records. As autonomy grows, platforms must also “comprehend” machine-initiated orders they used to treat as threats, another reason to invest in model management and accountability.

HCLTech’s results support this approach: leaders expect the greatest short-term results from autonomous tools in real-time fraud identification/resolution (51%), smart routing (47%) and automatic compliance/reporting (47%).

Hidden when possible, clear when necessary

Seshadri ended with a cautious verdict: AI is real, here and speeding up. For now, we will be nearly autonomous and almost invisible, on purpose. The job is to connect the systems he outlined, including open banking, blockchain, IoT and APIs on modern cloud with strong coordination, so that automatic decision making, voice-activated payments and predictive cash flows all build “embedded trust.”

HCLTech’s study gives both urgency and guidance. Leaders consistently prefer bold actions, but preparedness lags; 87% fear losing clients without instant payment, but 43% say unclear regulations are the risk they’re least equipped for. The winners will move from trial to action: investing in strong data, clear AI policy, and result-driven design that lets users stay in charge — even as transactions fade into the background. That’s how invisible payments create visible value.