Scaling Regulated Data Workflows Without Lock‑In - with Juan Orlandini of Insight

In this episode, Juan Orlandini warns finance leaders that generative AI models are fundamentally statistical, not mathematical. When CFOs layer AI on top of legacy tax and reporting systems without proper constraints, the tools can hallucinate numbers that look plausible but fail audit checks. By anchoring AI outputs to existing balance‑sheet controls and reconciliation processes, organizations keep compliance intact while still leveraging AI’s speed for data extraction and preliminary analysis. This framing helps executives understand why AI must serve, not replace, the rigorous math that underpins financial reporting.

Orlandini stresses that the real bottleneck is data, not algorithms. Most enterprises sit on sprawling data lakes that quickly become data swamps after mergers or legacy system sprawl. Deploying data engineers to cleanse, normalize, and catalog these assets creates a reliable foundation for AI to generate actionable insights. Leveraging SaaS platforms with built‑in governance—such as SAP, Workday, or other ERP‑integrated AI modules—provides built‑in audit trails and reduces the risk of vendor lock‑in. He advises CFOs to evaluate vendors on long‑term roadmaps, leadership stability, and the ability to export or migrate models, ensuring that today’s AI investment remains flexible as technology evolves.

Finally, Orlandini recommends a phased rollout that pairs technology with people development. Start with low‑risk, clearly defined use cases, verify outputs, and iterate before scaling. Simultaneously, upskill finance teams on AI fundamentals and embed governance checkpoints to avoid regulatory penalties. By treating people as the primary super‑power and aligning tools with existing financial controls, organizations can achieve sustainable ROI, maintain regulatory compliance, and future‑proof their data workflows against rapid AI advancements.