Cross-border payments are undergoing a structural shift, as AI is rapidly reshaping the slow, opaque, and cost-heavy process.
Sudipto Das, Vice President of Engineering at Convera, is leading the company’s efforts to implement AI infrastructure, enabling smarter, faster cross-border payments while preserving trust. It’s a process happening in parallel to AI’s transformation of payment security and its ability to embed trust from the start.
Why cross-border payments need infrastructure innovation
The rise of real-time payment (RTP) rails is a major force driving innovation in cross-border payment infrastructure. “RTPs are already live or soon will be in more than 80 countries, and that’s accelerating everyone’s expectations,” Das says. “Instant domestic payments are becoming the norm, and so will cross-border payments once certain infrastructure issues are solved.”
Legacy tech stacks weren’t built for real-time data exchange, intelligent risk scoring, or automated decision-making at scale. More importantly, they weren’t built for a world where AI is a core operational layer underlying the entire lifecycle of a payment.
“Moving money from point A to point B hasn’t fundamentally changed, but there are evolving aspects regarding risk, compliance, forecasting, personalization, and customer experience that can be enriched only with an AI infrastructure,” Das explains.
What infrastructure leaders are building with AI
Ironically, much of the innovation in AI is driving toward customer experiences that feel as “human” as ever.
Generative AI, for example, must handle and prioritize multiple intents while also recognizing voice inflection, dialect, and other human nuances that fundamentally transform the customer experience. “This AI experience has to be emotional and natural, connecting on a human level,” Das says.
Beyond improving customer experiences, financial institutions and payment providers that adopt AI tend to follow a predictable maturity curve with three phases.
1. Internal productivity and workflow automation
The financial sector is heavily regulated, so most organizations begin with low-risk, internal use cases such as automating repetitive operational tasks, supporting engineering teams, streamlining customer support workflows, and reducing manual review touchpoints.
“This is when companies build confidence in AI, establish guardrails and playbooks, and ensure the technology behaves reliably before touching customer-facing systems,” Das explains.
2. Real-time risk management and fraud mitigation
Features such as instant risk scoring of transactions and dynamic customer risk profiling have become central to advanced cross-border payments, fueled by generative AI.
“The faster a system can evaluate risk, the smoother a payment becomes,” Das says. AI infrastructure directly reduces friction, improves approval rates, and minimizes false positives. “Those are the critical advantages in cross-border flows where variability and complexity are higher,” he adds.
3. New value-added services
Once organizations are comfortable with AI’s internal and risk functions, they can expand outward, creating entirely new services such as intelligent treasury tools, automated reconciliation workflows, and cashflow forecasting.
“These services add value and can become strategic offerings,” Das says.
The role of generative AI in trust and speed
Traditional AI and ML have supported banking for more than a decade, but generative AI introduces new capabilities because it can reason, summarize, contextualize, and “agentize” tasks that once required human judgment.
When layered into payment infrastructure, genAI can:
- Explain why a payment was flagged
- Summarize compliance obligations
- Automate document checks
- Handle unstructured data
- Guide teams through exception handling
- Provide transparency to customers
As Das puts it, we’ve reached a moment where “things have become real.” GenAI agents can meaningfully assist with real operational tasks, so long as the appropriate controls, validations, and oversight structures are in place.
Trust is no longer just about meeting regulatory requirements. “Trust is also about intelligibility, explainability, and confidence, which GenAI can enhance when properly deployed,” Das explains.
Keeping humans in the loop
Even with rapid advances in agentic AI, the financial sector cannot operate on automation alone.
According to Das, human oversight remains, and will remain, essential in several key areas:
1. Governance and guardrails: Before AI touches a single customer interaction, teams must define what behaviors will be allowed and set escalation thresholds. Additionally, people need to be hands-on with their AI agents to validate model performance and approve data sources.
2. High-risk or high-value decisions: Large transfers, unusual behaviors, sanctions complexity, and regulatory interpretations are moments where human judgment still outperforms even the best models.
3. Continuous feedback to train systems: AI infrastructure is constantly evolving, and human experts are needed to tune parameters, refine classifications, and help systems adapt to new fraud patterns, new regulatory rules, and new market behaviors.
While AI accelerates decision-making, it’s up to humans to anchor those systems to ensure a positive customer experience, organizational compliance, and trust.
Balancing governance and innovation at scale
“The biggest challenge for payment providers isn’t choosing whether to adopt AI, it’s figuring out how to adopt AI responsibly,” Das says. “That’s where the question of trust comes in, once again.”
For Das, there are a few key issues to keep in mind:
- How do we innovate at the speed of genAI without compromising regulatory obligations?
- How do we ensure AI model outputs are explainable?
- How do we protect data across borders?
- How do we maintain trust when using systems that learn and evolve?
The answers usually come down to AI infrastructure offering centralized governance, standardized guardrails, and unified data management.
“With the right foundation, AI can become a reliable, compliant, and scalable layer powering the future of global money movement,” Das says.
“The cross-border payments companies that win in the next decade will be those that treat AI as a core operating system — one that enhances trust, speeds transactions, reduces friction, and unlocks entirely new value for businesses and consumers worldwide.”
Listen to the latest episode of the Converge podcast to learn more about building AI-driven infrastructure.
