Editor’s note: This article is based on insights from a podcast series. The views expressed in the podcast reflect the speakers’ perspectives and do not necessarily represent those of this publication. Readers are encouraged to explore the full podcast for additional context.
Artificial intelligence has gone through many hype cycles but according to Stuart Feld, Senior Vice President at Raymond James, this one is fundamentally different. In a recent episode of the “CAIO Connect” podcast, hosted by Sanjay Puri, Feld shared a grounded, experience-driven perspective on what it really takes to deploy AI responsibly at scale, especially in a highly regulated industry like financial services.
What made this conversation stand out was not futuristic speculation, but practical insight from someone actively leading AI transformation inside a large enterprise.
Feld makes a compelling observation: AI is the first truly transformative technology that was adopted by the general public before the enterprise. Unlike cloud, mobile, or algorithmic trading, AI tools like ChatGPT became mainstream almost overnight forcing enterprises to react rather than lead. With billions of users engaging with AI daily, organizations can no longer afford to treat it as experimental.
This reality shaped Raymond James’ approach and ultimately led to the creation of a dedicated Chief AI Officer role in early 2025.
AI wasn’t new to Raymond James. Feld and his teams had been delivering machine learning solutions for years, particularly in back-office compliance and supervision areas where AI delivered measurable dividends. The difference was credibility: Feld wasn’t advocating AI in theory; he was solving real business problems with it.
READ: Inside New York’s AI Vision with Shreya Amin (
That track record made the CAIO role a natural evolution rather than a leap of faith. The mandate wasn’t hype, it was scale, governance, and accountability.
One of the most relatable moments in the conversation was Feld’s reframing of a common fear that whatever we build today will be obsolete in 12 months, “what we’ve done with machine learning that we started five years ago. We do it differently now. It’s just better… The tried and true machine learning, some of the things we’re doing with gen AI, that’s just going to get better. As 5.0 models now, a year from now, we might be at 8.0 models now.”
His response? Obsolete doesn’t mean useless.
AI systems, when designed well, get better over time. Core machine learning foundations and GenAI capabilities evolve rather than being thrown away. While experimental layers like agent-to-agent communication or model context protocols may change rapidly, the underlying architecture compounds in value.
Feld is clear that at Raymond James, “We are not looking to replace advice or judgment. We stick humans in the loop. What I like to say is, we’re not doing anything an advisor can’t do. We’re just doing it lightning fast.”
AI should do what humans already do but faster. Tasks that take an advisor an hour can be done in seconds, freeing time for higher-value work. AI surfaces information, highlights opportunities, and reduces low-joy administrative effort but the final decision always rests with a human.
This “human-in-the-loop” approach is non-negotiable, especially in financial services where explainability, traceability, and accountability matter as much as accuracy.
Feld outlines a simple but powerful trajectory for enterprise AI adoption:
- Surface general information
- Surface personalized insights
- Recommend actions
- Eventually execute actions with human approval
READ: Treating AI like ‘a junior employee’: Lessons from Marc Kermish (
Most organizations, including Raymond James, are firmly in the first two stages today. Action comes later, once trust, governance, and explainability are firmly in place.
Despite the buzz, Feld notes that very few organizations are running truly autonomous AI agents in production. Most are experimenting with testing security, scalability, and integration with existing APIs.
For mission-critical functions, Feld doesn’t yet see autonomous agents operating without human oversight. And that caution isn’t just regulatory, it’s philosophical.
AI remains deeply underrated. There is enormous low-hanging fruit that can deliver immediate value without moonshots. Feld’s advice is simple and powerful: focus less on proofs of concept and more on real implementation.
Boards don’t want to hear what AI might do someday. They want to know what it’s doing now.
And as this “CAIO Connect” podcast episode makes clear, responsible, human-centered AI isn’t slower, it’s the fastest path to sustainable scale.