Fiserv co-president Dhivya Suryadevara, in an interview with PYMNTS this week, made a case that is going to be either widely adopted or widely contested over the next twelve months and is unlikely to settle in between. Her argument: AI agents deployed on top of legacy core banking systems may collapse what has historically been a multi-year, multimillion-dollar core-modernization slog into something materially faster. The implication for community banks and credit unions, which run on Fiserv”s infrastructure and have been told for a decade that AI requires a core replatform before deployment, is that the wait may be over.
The structural framing matters because Fiserv processes payments and core banking for thousands of US community banks and credit unions. Suryadevara”s position effectively gives those institutions permission to deploy AI on top of legacy cores rather than wait for a generational replatform. That is the opposite of what the industry consensus has been telling community-bank CIOs since 2018. It is also, if the framing holds up in practice, a meaningful expansion of the addressable market for the vendor category that sits on top of Fiserv”s rails — including Saris, nCino, Alkami, and the broader cohort of agentic-banking and lending-automation vendors.
The PYMNTS coverage lands the same day as a separate piece on Huntington Bank”s payments-resilience infrastructure, suggesting a coordinated narrative push from the payments-and-banking-infrastructure community around two themes: AI as accelerator, and resilience as the underweighted operating priority. Whether the coordination is intentional or coincidental, the pattern reads as the major core-providers actively making the case to their installed bases that they can deploy modern capabilities without the disruption of a core replacement.
The substantive question, for community-bank CIOs and CFOs reading the framing, is whether the math actually holds up in production. Agentic AI on top of a legacy core works for some categories of workflow — back-office automation, customer-correspondence routing, lending documentation, compliance evidence — and works substantially less well for others. Anything that requires deep core-data restructuring, real-time event-driven integration, or modern API patterns that the legacy core does not natively support will struggle under the AI-as-shortcut framing. The vendors making the most aggressive claims are working hard to obscure that distinction in their pitches.
For community-bank executives, three questions are worth pressuring the vendor on. First, which specific workflows does the AI overlay actually automate without core changes, and which would require core modernization before they could function reliably? Second, what is the operating model and accountability structure when the AI overlay produces an error — does the responsibility sit with the bank, the vendor, or the core provider? Third, what is the regulatory posture on the deployment, and has the OCC or state regulator with jurisdiction over the institution had visibility into the architecture?
Watch for: Jack Henry”s and FIS”s competitive responses, regulatory guidance specifically addressing AI-overlay-on-legacy-core deployment, and whether the framing produces a measurable shift in community-bank AI procurement in the second half of 2026.
Reporting based on PYMNTS interview with Fiserv co-president Dhivya Suryadevara.
