Financial services firms are turning to generative artificial intelligence to provide advisors with a crucial advantage: streamlined access to institutional knowledge. Last week, Raymond James announced the launch of its own gen AI-powered proprietary system to access information stored across its internal databases.
Stuart M. Feld, the firm’s first-ever chief artificial intelligence officer, recently explained to Financial Planning that the firm’s new intranet search tech allows employees to ask natural language questions.
“Like all firms, there’s a lot of data there, and it was a keyword search,” he said. “Now you can ask questions and get natural language responses. Before you had to pick up the phone and call a help desk or ask somebody else.”
READ MORE: Raymond James’ new AI chief opens up about tech at the firm
And while this has the potential to provide advisors with additional time to spend with clients, experts say the results of gen-AI queries should be reviewed with a skeptical eye.
In-house AI tools gain traction
The development of in-house AI tools to enhance knowledge management and service delivery is a trend likely to grow within and outside of financial services firms, said William Trout, director of securities and investments at technology data firm Datos Insights.
“This allows employees to quickly access institutional knowledge without needing to navigate complex information repositories or wait for support from colleagues,” he said.
READ MORE: Raymond James’ wealth unit again posts record profits, assets
This type of application represents one of the primary ways companies are investing in AI, said Trout. Internal knowledge search is “practical” and “high-ROI,” allowing a firm to capitalize on existing company knowledge bases to deliver immediate productivity improvements. It has a clear use case with measurable benefits and can be implemented without massive changes to existing workflows, according to Trout.
Raymond James’ annual investment of $975 million in technology, of which AI search is one component, demonstrates how seriously the firm takes technological innovation, he said.
“Similar applications are likely to become standard across industries that manage large amounts of specialized information,” he said.
In fact, internal search platforms may be the tip of the iceberg when it comes to firms using generative AI to enhance search and service capabilities.
“For larger enterprises like Raymond James, the real promise is cutting down the time it takes to find relevant client information across massive internal systems, which can directly improve both advisor efficiency and client experience,” said Jordan Gilberti, founder of Sage Wealth Group in New York City.
Freeing up more time for client interactions
AI, like any other technological advance, has the potential to enable advisors to spend more time with clients in meaningful interactions while spending less time on other tasks such as fact gathering or analysis, said Meredith H. Schneider, founder of Schneider Wealth Management in Palo Alto, California.
More firms will follow Raymond James’s lead by rolling out time-saving generative AI tools, said Joshua Mangoubi, founder and wealth manager at Considerate Capital in Chicago. His firm has been working on developing and internally testing its own AI models.
“Our AI models help us organize data and analyze complex reports,” he said. AI demonstrates proficiency in simple tasks yet encounters difficulties when interpreting nuance and delivering professional judgment. AI-generated hallucinations create challenges because they produce incorrect data with high confidence.”
AI results should not be blindly trusted
Though results from gen AI search tools can deepen advisors’ understanding and improve efficiency, the potential for errors abounds. Trout said users should view generative AI search results with “informed caution.”
The key is not just speed but also accuracy and trust, as these systems still need strong human oversight, said Gilberti.
“If the AI returns a fast answer but the advisor doesn’t have the training or confidence to verify and contextualize that output, it can backfire quickly,” he said. “The firms that will win in this space aren’t just the ones who invest in AI, but those who build the right checks, balances and training alongside it.”
While these systems can be highly effective, they face challenges, including the potential for hallucinations or incorrect information synthesis, difficulty with nuanced interpretation of complex financial regulations and possibly outdated information if knowledge bases aren’t regularly updated.
“I once used AI to evaluate an investment opportunity, and the results were patently wrong,” said Schneider. “They may have been accurate a few years ago, but they were not accurate today. I asked for a reference to see if I could figure out why the results were wrong, and the references simply did not support the output.”
Potential pitfalls around hallucinations and security issues can make AI systems tough investments to manage, especially at scale and with a longer-term perspective in mind. As a result, Eva Nahari, chief product officer at AI agent and assistant platform Vectara, said she sees enterprises keeping many of these services internal-only.
“Their AI applications are designed to serve a human agent whose task gets assisted and accelerated, but in the end, the human is the control checkpoint for what information gets externally exposed,” she said. “As long as hallucinations are involved, we don’t see a path to fully roll out agentic or other AI services without human intervention.”
In the case of Raymond James, Trout said humans are still in the loop by using the AI as a tool that advisors control, not an autonomous decision-maker; by implementing explicit human checkpoints in the process; by creating real-time feedback mechanisms where users can flag incorrect or problematic responses; and by developing the system in-house to maintain transparency and control.
“Advisors should approach these tools critically and seek to use AI-generated information as a starting point for analysis rather than accepting outputs as definitive answers,” said Trout. “This maintains the human judgment and expertise that clients ultimately value while at the same time, leveraging AI to enhance efficiency and information access.”
When used with care, AI can serve as a valuable addition to advisory work, said Mangoubi. He compared AI to collision avoidance systems in modern cars in that AI doesn’t drive decisions but can indicate potential issues that require attention.
“The AI support functions as an additional verification step, enabling advisors to conduct more detailed analyses while keeping human decision-making at the forefront,” he said.
