Why should financial advisors use AI? To hear AI providers tell it, it’s because the technology will save time on rote tasks that would be better spent with clients.
But for those who have taken these claims at face value, many find that common tasks assigned to AI would be completed faster and more accurately if they had simply done it themselves in the first place.
AI can be efficient, or it can be “an absolute time sink,” said Eric Croak, president of Croak Capital in Toledo, Ohio, who has been taking time to study both “the promise and the problems” of AI in the industry.
“It depends on how you use it, and honestly, what your expectations are,” he said. “When AI tools are good, they are fast. But when they are bad, it feels like dragging a broken sled up a hill.”
As a law firm partner who regularly manages complex financial matters for both the firm and clients, William “Bill” London, partner at Kimura London & White in Irvine, California, said he has experimented extensively with AI tools to streamline research, draft documents and even do some financial analysis.
READ MORE: AI-powered firm’s tech could expand planning market exponentially
“While AI has clear strengths, it doesn’t always deliver efficiency — and in certain cases, it actually slows things down,” he said.
Client communications require care
What should take 10 minutes can often turn into 45, especially when an advisor is trying to generate client-facing content that does not sound robotic or generic, said Croak.
“You end up rewriting half the draft and tweaking every other sentence,” he said. “That is not efficient.”
When advisors try to use tools like ChatGPT to save time on client-facing content such as portfolio summaries and planning emails, it can take longer to “fix” the responses than to just spend the time to write it all themselves, said Wyatt Mayham, CEO and co-founder of Northwest AI Consulting. He works with RIAs and financial platforms looking to integrate generative AI into their marketing and client communication workflows. Mayham said the root issue is that most of these tools aren’t trained on the advisor’s real client data or voice.
“Without that context on the relationship, financial advisors end up re-prompting multiple times and still getting outputs that feel generic or off-brand,” he said. “And with relationships as personal as the one between a financial advisor and their client, you can’t have that.”
READ MORE: AI-powered search is re-writing SEO rules, and advisors must adjust
Scrubbing for compliance eats up time
Compliance-heavy content can eat up the most time, said Croak. He said he has seen AI generate a seemingly clean tax summary in two minutes, but then had to spend 30 minutes scrubbing it for outdated thresholds, vague language and “stuff you cannot legally say.”
“You have to re-prompt over and over,” he said. “Change tone, tweak disclaimers, remove ‘advice’ phrasing. Eventually, it’s faster to write it from scratch.”
Drafting an initial financial risk memo related to litigation strategy has proven particularly problematic, said London.
“What should’ve taken 30 minutes to write manually ended up taking closer to two hours,” he said. “I had to re-prompt the AI repeatedly to clarify vague language, correct factual inaccuracies and restructure the argument in a way that complied with legal standards. By the time it was usable, I had essentially rewritten it myself.”
The biggest time-wasting issues tend to come up in areas that require nuance and compliance — like financial modeling, legal analysis or anything involving jurisdiction-specific rules, said London.
“These tools don’t yet understand the regulatory context or professional tone required, and often hallucinate or generalize,” he said. “That creates risk and rework.”
Generalities are still efficient
Where Mayham said he typically sees AI being most efficient for financial advisors is on the back-office side. Tasks like summarizing meeting notes and drafting disclosures have tighter boundaries and don’t require as much nuance.
If all an advisor is looking for is a generic template, AI can still deliver, said Croak.
“Need a draft outline for a 529 comparison? Great. Want to analyze a Roth conversion strategy? Meh,” he said. “It can spit out numbers, but the nuance just is not there. So, when you are dealing with tax strategies or layered estate issues, with real variables, you are still better off doing the thinking manually.”
Where AI excels is in ideation, formatting and repetitive tasks like summarizing large amounts of information or generating basic content outlines, said London.
“In those cases, it can significantly speed up workflows and free up cognitive bandwidth for higher-value work,” he said.
This “speed trap” happens because AI does not think like a planner, said Croak.
“It can regurgitate averages, but it has zero sense of what’s actually practical,” he said. “So instead of solving problems, it creates new ones.”
How AI providers could improve their tech
As for improvement, London said AI providers could vastly enhance the experience by training tools to better understand context-specific standards and by giving users more control over tone, formatting and fact prioritization.
“Until then, professionals will continue to find that AI is sometimes helpful — but not always efficient,” he said.
To improve, vendors need to prioritize vertical fine-tuning and allow firms to train on internal content, said Mayham. If that doesn’t happen, “AI will keep being a bit of a mixed bag for advisors, depending on where it’s used,” he said.
If providers want to make AI less painful, Croak said, they should stop trying to make it clever and instead make it consistent.
“Give users a no-jargon switch,” he said. “Let us lock tone, style and compliance settings so we do not have to babysit every output. Because right now, it’s like using an intern who talks too much and does not listen.”
