At the 2026 SIFMA Operations Conference in Marco Island, Florida, financial services leaders and compliance professionals gathered to examine one of the most significant questions facing the industry: how do you modernize financial services infrastructure at scale, without compromising the resilience that clients and regulators depend on?
The answer that emerged was not a technology roadmap. The tools are available, the urgency is real, and the talent exists. What the industry is working through is how to sequence the investments, build the right foundations, and ensure the people behind it are set up to lead rather than keep pace. The themes that surfaced align closely with what North Highland's financial services team is seeing with clients nationwide.
Panelists from various Financial Service Institutions each made the same point in different terms: AI is not limited by model sophistication. It is limited by the quality, accessibility, and trustworthiness of the data it runs on. The firms that are seeing measurable impact from AI deployments are not the ones with the most advanced tools. They are the ones that spent the necessary time on data architecture before they built anything on top of it.
For operations leaders, this is a reframing worth taking seriously. Data governance is not a box to check off before AI investment; it is the investment. As one session made clear, when something goes wrong in an AI-enabled process, the first questions from regulators and auditors will not be about the model. They will be about where the data came from, how it was transformed, whether it was contaminated, and who approved it. Firms that can answer those questions confidently are the ones positioned to move quickly, while firms that cannot will find themselves slowing down just as competitors accelerate.
The compliance and operations session and the 2036 panel converged on a warning that resonates with how North Highland approaches transformation: layering AI onto an unchanged operating model does not produce transformation. It produces a more expensive version of the same outcome.
Real-time compliance requires real-time controls. A batch processing control framework applied to a real-time data environment is not a cautious approach; it’s a gap. The same logic applies to client expectations, where the standard set by consumer digital platforms now extends to financial services. Clients expect clear, consistent, explainable outcomes. Legacy infrastructures built on manual handoffs and fragmented data systems cannot deliver them, regardless of what AI is layered on top.
For operational leaders, the most valuable investment right now is in the operating model itself, including digital workflows, modern data architecture, and the human oversight structures that sit above the automation layer. Technology follows from that foundation. Without it, technology accelerates the wrong things.
The AI governance session was the most direct of the conference. The core argument was simple and consequential: governance cannot be owned by the technology team because technology is not accountable for business outcomes. The business units deploying AI on their processes own those processes, including the risk of what happens when the AI gets something wrong. Assigning governance to the technology function creates a structural gap between the people who set the rules and the people who carry the consequences.
Stifel's Ron Kruszewski pressed further. He raised two questions he said no one in the room, or within the regulatory authorities, has answered yet:
Those questions aren’t hypothetical. They’re the next set of challenges this industry will face, and the people with the experience to work through them are the same ones who delivered T+1 (Trade settlement in one business day) without breaking the market.
The 2036 panel was explicit: operational roles are evolving, not disappearing. The work is shifting from transactional processing toward advisory and oversight functions that require judgment, curiosity, and the ability to ask the right questions of systems generating real-time insights at scale.
The governance session added an important caution. Firms that block AI access do not prevent their people from using it. They push that use into the shadow, outside any oversight or learning loop. The more productive approach is to invest in internal education, reward the intellectual curiosity that drives good AI use, and build the governance structures that make experimentation safe rather than prohibited.
What this requires of leaders is a different way of thinking about their teams. The people who will thrive aren’t necessarily the deepest experts in a single vertical. They are the ones who think across the end-to-end flow, understand how data moves from the front office to settlement and back, and can exercise judgment when AI systems surface something unexpected.
The SIFMA Operations Conference 2026 didn’t produce a consensus technology roadmap. What it produced was something more durable: a shared understanding of where the work is, what is blocking it, and what it will take to move forward. The firms building now are investing in data architecture, operating model redesign, governance structures with business ownership, and the people programs that make all of it stick. These are not new themes, but the urgency is new. The firms that are Built for AI will be the ones setting the standard in 2036.
At North Highland, we work alongside financial services organizations on exactly this kind of transformation, from operating model strategy and change adoption to AI implementation, data governance, and workforce optimization. If any of these conversations resonate with the challenges your team is navigating, we welcome the chance to continue the discussion.