How UBS Built An AI Insight Flywheel

Many firms build an AI strategy. The smart ones build a business strategy and use AI to power it. By examining the practices of its most effective advisors and reconstructing those workflows with intelligence embedded throughout, UBS created a flywheel that strengthens with each use. The result is an insights engine with an 80% advisor adoption rate that now delivers 17 million insights annually.

The architect behind the flywheel is Joe Cordeira, Chief Data and Analytics Officer for UBS Wealth Management Americas. His perspective is shaped by two decades in financial services, including senior roles at Morgan Stanley and Merrill Lynch, where he led large-scale data, analytics, and CRM programs that transformed how advisors used information to serve clients. That foundation set the stage for the advisor-driven model he championed inside UBS.

"The platform is built by advisors, for advisors. We worked backward from how our best advisors serve their best clients—what they look for, what they wish they had—and built an insights engine around that," said Cordeira. UBS aimed to address a familiar issue in wealth management. Advisors were juggling diverse client needs and nonstop market information, while also seeking to link the firm's broad platform to each client's situation.

• Complexity crunch: "Our advisors have many clients with intricate financial lives and a vast global platform. The difficulty is connecting the right solutions from that platform to the right clients across their entire book of business." The challenge, Cordeira said, was finding a way to turn raw information into something advisors could act on without slowing them down.

To address this, UBS developed a "client experience engine" that scours internal and external data. The platform uses a knowledge graph to create a 360-degree view of every client, powering over 400 different types of proactive insights. These can range from acting as a compliance backstop for business-owner clients unaware of state-level retirement plan regulations, to spotting opportunities for hedging a concentrated stock position, to surfacing personal milestones like a client's child starting college—a detail that matters to the human relationship.

• Flywheel kicks in: That analytics platform created a flywheel effect. As the tool delivered value, advisors contributed more ideas, which were then scaled into new insights for the entire firm, driving further adoption. "This flywheel kicks in once the platform becomes so valuable that advisors are not only using it, but are actively contributing more and more ideas to grow the engine," Cordeira said. "The best ideas from our best advisors can then be scaled and leveraged across the whole business."
• A $20M insight: One of the clearest examples comes from a recurring pattern the engine surfaces. Advisors often discover client cash sitting in low-yield external accounts, which creates an immediate opening to offer a better solution. As Cordeira explained, "Sometimes it's idle cash in an external account that is barely earning anything. When the advisor sees it, they can step in with a better solution. That exact mechanism surfaced a twenty million dollar opportunity not long ago."

But none of this works without trust. The platform's effectiveness lies in a hybrid model that keeps the advisor, not the algorithm, at the center of the client relationship. The advisor is positioned as the ultimate decision-maker, backed by a rigorous governance process.

• Governance in action: "We have very rigorous both data governance and AI governance policies with concrete safeguards to ensure we are protecting our clients," Cordeira said. "It's a multistage process that requires approvals from many different parts of UBS: legal, data privacy, compliance, ethics, etc. Everything is reviewed holistically."

• The longest pole: "Every one of those 400 insights goes through that same governance process. That process is the longest pole in our development cycle, and it exists to make sure what we're doing is safe," he continued. The process is designed to ensure that only well-tested and fully vetted insights make it into the platform.

• Human in the loop: "The 'human in the loop' is critical and will continue to be," Cordeira said. "The trust between the advisor and client is no different than it was without this tool; we're simply giving them better information to better serve their clients. The advisor is the one who uses their judgment to determine how to use that information."

What’s next is an evolution from insights to action. The firm's credibility on this front stems from a core insights engine built on traditional machine learning, a long-term strategy that predates the recent Generative AI boom. That foundation now allows the firm to empower advisors with digital tools and "leapfrog" simple chatbot interfaces by expanding the current "Advisor Insights" engine into "Advisor Assist."

The future suite of agentic AI tools is designed to automate day-to-day front-office tasks, with plans for a "meeting scheduling agent" to handle logistics or an agent that can pull reports from the chief investment office. It’s a vision that connects advancing technology back to the firm’s timeless business mission: "We're trying to save advisors time from administrative tasks so they can spend more time building relationships, better serving clients, and prospecting to drive growth," Cordeira concludes.