Tags
Share
At Sibos 2026 in Miami, discussions of agentic AI, payment data, and security raised a practical question: what will it take to make AI dependable in financial operations?
That question was central to this year's Sibos, held in Miami from September 28 to October 1 under the theme Digital finance for AI-driven economies.
Exadel's Cesar Cotait, Global Head of Financial Services, and Ilan Pillai, Head of Business Development, attended the event alongside financial institutions, technology providers, and industry leaders exploring the next phase of digital finance.
Three themes stood out across the event's discussions: the growing role of AI agents in financial operations, the importance of reliable data, and the challenge of managing risk as automation expands.
Agentic AI in Financial Operations: How Much Can We Automate?
Agentic payments and operations were part of the broader conversation about the future of financial services at Sibos. Payment reconciliation offers a practical example: an AI agent could gather transaction records from multiple systems, identify discrepancies, and recommend how to resolve an exception, reducing the manual effort involved.
But this raises a more difficult question: when should an AI agent be allowed to act on its recommendation?
In financial services, that decision depends on clear permissions, approval processes, and records of what the system has done. The challenge is to introduce greater automation without weakening operational control.
Exadel brings AI Engineering and financial services expertise to this challenge, helping institutions integrate AI into operational workflows with the necessary controls in place.
ISO 20022 and AI-Ready Data
Before AI agents can make reliable recommendations, they need reliable information. That makes ISO 20022, another topic discussed at Sibos, particularly relevant.
ISO 20022 provides richer, more structured payment information, creating opportunities to improve reconciliation, monitoring, and exception management.
Yet adopting a common messaging standard doesn't automatically make information usable by AI. Payment data still needs to be accessible, consistent, and reliable across systems.
The broader challenge extends to customer-facing AI. Financial institutions are exploring how AI assistants and personalized digital experiences could improve client service, but those applications also depend on connected data and a clear understanding of customer needs. Before investing in new capabilities, institutions need to assess whether their data, systems, and processes are ready to support them.
This makes data engineering a practical part of the AI conversation. Exadel's Data Engineering & Analytics expertise helps institutions address the integration and data quality challenges that stand between better information and better operational decisions.
Trust, Governance, and Security: Who Remains Accountable?
Giving AI systems greater access to financial data and operational workflows creates new security and governance questions. Trust and cybersecurity were important themes at Sibos, particularly as institutions consider how much authority these systems should have.
Real-time payments make that balance particularly difficult. Institutions need to detect suspicious activity quickly without creating unnecessary delays for legitimate transactions. Explainable AI could help fraud and compliance teams identify suspicious activity while making it easier to understand why a transaction was flagged.
An AI system that summarizes information carries different risks from one authorized to make changes in a live financial system. Institutions need to know what each system is permitted to do, how its actions are monitored, and when a person must intervene.
That thinking is central to Exadel's Human Intelligence, Amplified by AI approach: applying technology to strengthen human expertise while keeping critical decisions under appropriate oversight.
What Sibos 2026 Means for Financial Institutions
The discussions at Sibos showed how closely connected these challenges are. AI agents could help financial institutions streamline operations, improve customer experiences, and respond to fraud faster. But each opportunity comes with questions about data quality, system integration, and accountability.
The starting point is a specific business problem, followed by a realistic assessment of what it will take to solve it.
Thank you to everyone who took the time to connect with our team in Miami and share their perspectives. We enjoyed being part of the conversation and look forward to continuing it beyond Sibos.
Your AI Partner
Looking to modernize your financial platforms or introduce AI into your operations?
