Pismo Highlights AI-Driven Limits of Banking Infrastructure

PYMNTS published a video installment of its "What's Next in Payments" series featuring Pismo's Leonardo J. Collado, who said legacy payment systems remain reliable but are slowing product development. Collado argued that AI is exposing the limits of infrastructure built for batch-era banking operations, pushing modernization discussions from IT teams to boardroom-level strategy. The interview highlights how payments firms are reassessing whether long-lived core systems can support the product velocity required by AI-enabled services.

Pismo Highlights AI-Driven Limits of Banking Infrastructure PYMNTS reports that, in its "What's Next in Payments" series, Pismo's Leonardo J. Collado said legacy payment systems still anchor reliability but are increasingly slowing product development. According to PYMNTS, the coverage frames AI as revealing the limits of infrastructure built for batch-era banking operations and says modernization has moved beyond IT teams into boardroom discussions. The piece centers on a video interview with Pismo leadership and highlights how payments firms are reassessing whether long-lived cores can support AI-enabled product velocity. What happened PYMNTS published a video installment in its "What's Next in Payments" series in which Pismo's Leonardo J. Collado discussed legacy payment systems. PYMNTS reports that the coverage emphasizes that legacy payment systems "still anchor reliability" while also slowing product development, and that AI is exposing the limits of infrastructure built for batch-era banking operations. PYMNTS states that the conversation framed modernization as a boardroom-level issue rather than solely an IT project. Editorial analysis - technical context legacy payment cores were designed for throughput, reconciliation, and batch processing, not for the low-latency, data-hungry workloads that many AI use cases demand. Companies modernizing payment stacks commonly adopt event-driven architectures, streaming platforms, and real-time data pipelines to support continuous inference, observability, and feature stores. These architectural patterns reduce end-to-end latency and make it easier to deploy iterative ML models for fraud, personalization, and routing. Context and significance public reporting from payments vendors and banks has repeatedly flagged integration friction between modern ML tooling and decades-old transaction processing systems. For practitioners, that pattern means integration work and data engineering often dominate the calendar when pilots move toward production. The PYMNTS piece highlights a broader industry conversation about whether governance and board-level resource allocation need to align with modernization timelines. What to watch observers and practitioners should monitor three indicators that will show whether modernization accelerates across the sector: - •increased adoption of event-streaming platforms and real-time change-data-capture CDC deployments; - •more production ML deployments tied to payments outcomes such as realtime fraud scoring and dynamic routing; and - •vendor announcements bundling modern orchestration and observability with core processing. For practitioners: the immediate implication is that integration, data quality, and latency measurement will remain central technical tasks as teams attempt to leverage AI across payments flows. PYMNTS has not published verbatim quotes in the scraped excerpt, and Pismo has not issued a separate public statement in the scraped material on the company rationale for the comments. Scoring Rationale The story highlights a notable, practitioner-relevant friction point where AI use cases meet decades-old payment infrastructure. It is important for data and ML engineers planning production ML in financial services but not a frontier technical breakthrough. Practice with real Payments data 90 SQL & Python problems · 15 industry datasets 250 free problems · No credit card See all Payments problems /problems/datasets/payments