Keebeck Wealth Management, a Chicago-based multi-family office managing almost $2 billion for 40 families, has joined the ranks of independent firms betting on the latest advancements in technology to improve client communications, streamline research and portfolio construction, and achieve scale without the historically inexorable overhead.
“During COVID, everyone was calling all at once—and all the time,” said Keebeck founder and CEO Bruce K. Lee. “That just isn’t sustainable in the long haul. To continue to grow, we needed this kind of solution. The ChatGPTs of the world came out in the nick of time.”
Lee, who leads a team of six, said he spoke with around a dozen potential technology providers to find a tool that could ease those bottlenecks without leaving anyone’s questions unanswered.
He selected Qdeck, a new Seattle-based financial technology firm offering an investment and client relationship management platform supported by financial research curated from vetted internet sources that can be accessed in real-time by advisors and clients alike. The technology uses machine learning and proprietary natural language processing models and can be accessed on the web or embedded into existing technology for a white-labeled experience.
A key aspect of the platform is its use of “auditable primary sources,” according to CEO Jagdeesh Prakasam. Qdeck has filed patents on some of its natural language techniques for looking at real-time financial news and available internal data to help advisors make decisions and keep clients informed. It doesn’t matter what large language model is used, Prakasam said, the heavy lifting to ensure results aren’t hallucinated is done in-house.
“We have trained our own machine learning models that are doing all the processing on our side,” explained Qdeck’s Chief Product Officer Timothy Ireland. “It’s easy to swap out the layer that is doing the language interface, so clients can use whatever they prefer. It means we can stay on the bleeding edge of whichever is the most advanced model, and even which model is the best for certain capabilities.”
All current users are interacting via Chat GPT interfaces, but Ireland said startup Anthropic’s Claude 3 family of language models works better in some use cases and that Google’s Gemini and Microsoft’s Copilot are also areas of focus for integration.
Built by a team of AI researchers at a quantitative fund manager, Qdeck’s underlying institutional investment platform has been under development for almost two decades. It includes simulation, reporting, account management tools and an expanding library of model portfolio overlays. A language processing piece that will translate to portfolio construction is currently under development.
The team pivoted a few years ago after realizing they were creating something that would be valued by advisors. They sought feedback from investment officers at several large firms and developed the customizable client relationship software before officially launching last year.
Lee, who manages finances for more than 100 individuals and several organizations, said the technology is well-suited to his needs. The research function is being integrated with existing client communication software provided by Moxo and a white-labeled app the Keebeck team calls Ask Keebs (also powered by Moxo technology and a play on Lee's college nickname and the old Ask Jeeves internet search engine) to push relevant information out to clients without the need for immediate advisor involvement.
“We have adopted communication through technology as our digital army,” said Lee. “The AI is not meant to replace the advisor, but to help complement the advisor by providing some of the data their clients are looking for.”
Answers generated by the robo-service are shown in a different color font, so clients know what they’re looking at, and the Keebeck team receives notifications when follow-ups are warranted.
A former wirehouse rock star, Lee did stints with UBS, Lehman Brothers and Bear Stearns before the turn of the century. He then spent four years with Morgan Stanley and seven at Credit Suisse before joining Merrill Lynch in 2011. After being terminated from Merrill amid allegations he had failed to personally complete compliance training (he was sanctioned by FINRA in 2019), and a subsequent health scare, Lee decided to drop his brokerage license altogether and leave the wirehouse world behind.
In a nod to his family’s Korean surname, meaning “to ascend,” Keebeck was launched in 2018 on the Dynasty Financial Partners platform with $525 million in managed assets. The firm conducts all equities trading through its custodians, Charles Schwab and BNY | Mellon Pershing.
“I went through some institutional stress, for lack of better words, in getting to the next level and decided to start my own multifamily office with a vision of how I wanted to see clients get served,” Lee told WealthManagement.com. “We wanted to be very bespoke and offer investments, opportunities and access that I just didn’t think Wall Street really had.”
“The type of transactions we’re getting access to are something I know for a fact would never be marketed on Wall Street,” he added. “And I think that’s been validating.”
Lee is one of the growing number of advisors looking to technology to create a hybrid model that uses automation and AI, in tandem with personal client relationships, to fill service gaps and generally improve client outcomes. Last year, Carson Group declared it was on a mission to create “cyborg advisors,” and newer firms such as Savvy and Compound Planning are intentionally creating technology stacks pre-loaded with AI capabilities, API connectivity and a host of other tools meant to streamline advisors’ most time-consuming tasks.
At the same time, third-party providers such as Alphathena, Catchlight, Portrait Analytics, SIFA and others are offering up a proliferating array of AI-enabled tools that can help with everything from lead generation and personalized outreach to portfolio construction.
Others, such as Toggle AI, also offer research functions but Ireland said the only tool that has come close to rivaling Qdeck’s capabilities is Perplexity AI. Qdeck goes a step further than Perplexity by licensing the data from the sources it cites, he said.
“The difference is that’s essentially scraping the web for top-level, surface facts about your question,” he explained. “In our case, we’re paying for every single primary source document we have. We’re taking them and fully processing them ahead of time and so, even if the answer is 10 pages into a 10,000-word report from the Wall Street Journal, we’re going to give you that answer and point to that exact sentence and link to the original article.”
Lee said he’s “not just putting buckets out there hoping to collect the rain.”
“I personally believe the fundamental business of going about hiring advisors is antiquated,” he said. “I believe that the new generation of investors will be highly technologically advanced and more dependent on technology. So, the way we’re attracting clients is very simple. If we have better technology, quicker answers and better ‘first rail’ investments, word of mouth is going to grow fairly quickly.
“We’ve had good performance and healthy growth that’s been exclusively through referrals,” Lee said. “Based on the projects we’re working on and interest that has already been expressed, we can see being at $3.5 billion by the end of 2025. If we’re lucky, maybe $4 billion.”