RIAs need to be using AI tools 'yesterday,' tech leaders say

Advisors should start practicing with AI tools starting yesterday, according to Sindhu Joseph, founder and CEO at wealthtech firm CogniCor Technologies. 

Joseph was responding partly in jest to an audience question during a webcast, hosted by Ezra Group CEO Craig Iskowitz, about whether all RIAs will eventually be using AI. A survey by Nvidia earlier this year found more than 90% of financial services companies said they were either assessing AI or already using it in production.

"The answer is: Yesterday — all RIA firms will be using AI yesterday," Joseph said. The July 23 webcast featured industry leaders discussing how AI tools can generate returns for advisors. "There are many advisors who are using it," Joseph said, and they are "getting far ahead both in terms of efficiency gains as well as servicing the clients better, providing better products for the clients."

Joseph was speaking alongside some of the top tech leaders — and AI developers — in the advisor space, including Lee Davidson, chief analytics officer at Morningstar, and Henry Zelikovsky, CEO of financial software engineering firm Softlab360. 

While AI is increasingly being put into practice, it also poses challenges in terms of training the learning technology to be able to give correct responses and outputs specific to the wealth management industry. This might include how an AI tool reads and interprets a complex tax or estate planning document, or how it summarizes a client meeting.

READ MORE: From writing assistance to presentation builders: Top AI tools picked by wealth leaders

Morningstar, which launched its OpenAI-powered chatbot assistant Mo last year, went through the process of training its AI on more than 130 million PDF documents and in 30 different languages, Davidson said. 

"One of the big pain points for us in terms of managing the data was … ingesting large quantities of unstructured data and structuring it," he said. "There's going to be lots of other use cases similar to that across wealth management, where we're dealing with more and more unstructured data." 

READ MORE: In AI we trust? The peril and potential of embracing AI in wealth

And the costs to train these AI tools add up. The International Data Corporation projects spending on Generative AI globally will exceed $40 billion this year and surge to more than $150 billion by 2027. That's why it's critical any AI tool that an advisory firm practices can generate a return on the investment.

One way to do that, Joseph said, is to take a step back and assess how certain AI tools fit into the client experience from beginning to end. She gave the example of Morgan Stanley Wealth Management's new AI notetaker, AI @ Morgan Stanley Debrief, which generates notes on client meetings and surfaces action items for advisors. 

With Morgan Stanley's Debrief, "You are just taking the notes and putting that in the CRM, which is nice, but how about looking up in the CRM and seeing who to meet with, to begin with? And then preparing for the meeting, preparing the talking points, and then sitting in the meeting to capture the summary of that meeting, to creating tasks and connecting that to workflows," she said. "While there are many apps that are around, you should probably look at what is the end-to-end experience."

READ MORE: AI tools that find answers, suggest tax strategies and draft emails for you

Another critical component to applying AI tools for advisors is establishing guardrails within the technology, particularly if it's a large language model (LLM), to ensure the output does not inadvertently create regulatory risk or give unsound advice. That means teaching the LLM the types of questions it should not answer in addition to the questions it should, Davidson said. 

"There might be regulatory compliance reasons why you want to do that, or quality reasons," he said. "Avoiding questions is just as important as testing the accuracy of the answers that you're giving."