Altruist CEO Jason Wenk Talks AI After ‘Unusual’ Wall Street Selloff

Wealthtech firms don’t usually move the markets, but don’t tell that to Jason Wenk. 

The CEO of the digital custodian Altruist was enjoying a long weekend in Mexico earlier this month when news of the launch of his company’s new AI tax agent shaved millions of dollars from competitors’ market caps all before the chips and guac got served. Charles Schwab’s stock sank almost 11% in the days following the announcement, while the prices of brokerages like LPL Financial and Ameriprise dropped 13%. “We had one really nice and relaxing day, and then the next morning, I woke up to complete chaos,” Wenk said.

What had Wall Street on edge? A new artificial intelligence-powered tax tool that’s designed to create strategies for clients based entirely on their documents, and actually writes unique code for each individual client. It analyzes 1040s, account statements, pay stubs and other forms, and could become the “canary in the coal mine,” Wenk said. While some believe the selloff was a market overreaction, the tools are certainly beginning to worry investors. “It’s the sort of thing that people have feared the most about AI,” he told Advisor Upside. 

The tax agent may be just the beginning for the Culver City, California-based company that landed $152 million in its latest funding round in April at a $1.9 billion valuation. Altruist is already working on a financial planning agent slated for release this year. Advisor Upside sat down with Wenk to talk about AI and how the tech is primed to reshape wealth management.

AU: You were traveling when the Hazel announcement hit. How did that play out?

JW: I was supposed to be taking a three-day-long weekend. I had gotten an invite, ironically, from an advisor who’s become a friend, one of the early Altruist users. My wife and I went down with our son, and we got one really nice and relaxing day. But, we didn’t really get much of a vacation. I ended up spending most of that second day on the phone, and email, and on Slack. It was a cool experience I’ll never forget. If I had known it would be quite as big as it was, I may not have booked a trip. 

The irony is that most people recognize the place, Zihuatanejo, only from Shawshank Redemption. It’s still quite underdeveloped. I didn’t even have enough Wi-Fi to do a video meeting. If I wanted to do streaming video, it wasn’t even possible. It was an unusual series of events.

Why did this AI announcement land differently than other tech announcements?

There were two elements. Different people are promoting different AI products in wealth management all the time. The interesting part is that most of them aren’t that good, which is just the reality. It’s usually a couple-person startup with minimal resources putting a very simple front end on an LLM. Of course, every entrepreneur thinks their thing is going to change the world, but most of those tools are not very sophisticated. The other thing is that they’re not the highest impact. Most AI in wealth management today is note-takers. It records a meeting, creates a transcript, syncs notes to your CRM. That’s useful, but what’s really the GDP [impact] of democratizing access to good notes? Pretty trivial. That’s not making Morgan Stanley quiver. 

Hazel’s tax agent is the first meaningful agent built for wealth management in that it does actual human work. It does something that would normally take a long time and be quite expensive and makes it available to anybody for an extremely disruptive price. So, again, what’s the delta on note-taking? But if you’re a high-net-worth team, advanced tax planning might involve two to four people spending dozens of hours and charging clients tens of thousands of dollars. Hazel can do that work in two to three minutes. So, it’s actually disruptive to the advantage that a large high-net-worth firm can have. 

The second piece, which I’m just speculating about after talking to analysts, is that this came from Altruist. We’re not a small AI startup with a seed round. We’ve raised $650 million. We’re critical infrastructure for wealth management. Our core business competes directly with Schwab, LPL, Raymond James — some of the names that got hit the hardest. Names that have trillions of dollars in assets, that are intermediated assets. All of the sudden this powerful agent is released by a wealth platform business, people start thinking about asset flows. If you have the right wealth platform and a powerful suite of AI tools, why would advisors stay where they are if they can reduce costs by 80% or 90% and produce higher quality output? 

If that’s the case, how about walking us through exactly what the tool actually does?

Hazel connects to an advisor’s existing data: CRM, email, notes. With the tax agent, you drop in a tax return or any tax-related document. Hazel analyzes the documents along with connected data and creates an elegant tax summary written so a consumer can understand it. Then it creates recommendations to lower the tax bill. It works for any level of complexity. It can handle a simple W-2 filer. It can also handle a 500-page tax return with multiple Schedule Cs, K-1s, private business interests with complex depreciation schedules. It reads everything, extracts the data, and puts it into a cleaner database. Then it sources all relevant external information, federal, state, local tax code, specific to that client. If they live in Chicago, it looks at Cook County, the state of Illinois, the IRS. If they own property in Hawaii, it looks there, too. It doesn’t use the LLM to calculate the output. It uses the LLM to read and synthesize the data, then writes custom software code to generate the plan. That’s why it can replicate human-level work. Off-the-shelf software gives the same output for every client. Hazel writes software specific to each client.

Zooming out, what does AI mean for financial advisors over the next five?

Five years ago, most people wouldn’t have imagined what we can do today. AI wasn’t even in the conversation five years ago. In two and a half years, the progress has been extraordinary. The capital investment in AI is massive. Multiple vectors are compounding at the same time — you have the power supply, the compute power, and then the actual training models themselves. These models will be multiple orders of magnitude more powerful and dramatically cheaper within five years.

Basically, two things will happen. First, a large number of consumers will go directly to AI for wealth management and bifurcate humans altogether. That may skew younger, but it will happen. It’s not good enough today in terms of autopiloting everything. But, I would guess a 20-year-old today may never have a human financial advisor ever in their life. Second, there will still be a large segment, especially older clients in their 50s and 60s, who want a human relationship. Probably two-thirds of the private wealth in America is held by these people and they’re probably more reluctant to go full AI. There’s still going to be a big need for human wealth management, but the advisor’s job will change.

Administrative work will be gone within 12 to 18 months. Advanced planning, like objectively watching over every detail of your clients’ financial life, optimizing every little thing so no clients pays a dollar more in tax than they should, optimizing subscription usages, virtually anything that makes your financial life better, these agents will do. Advisors who don’t adopt AI will be more expensive and lower quality by comparison. What’s interesting is that AI today is already smarter than the collection of every CFP who has ever passed the exam. What we need are people who have these incredible soft skills that AI can’t do. The harder problem we’re going to face is figuring out how to solve that.