How AI is Shaping the Future of Finance with Runik Mehrotra

Summary

In this episode, Evan J. Cholfin interviews Runik Mehrotra, co-founder of Vise, about how AI is transforming wealth management. They discuss hyper-personalization, technical challenges, growth strategies, and future trends in AI-driven finance.

Takeaways

Hyper-Personalization in Wealth Management
AI Infrastructure Evolution
Moore's Law of AI Capabilities

Soundbites

"Each client is unique in nature."
"Models are getting so much better, so fast."
"Enabling personalization at scale is our big goal."

Chapters

00:00
Introduction to AI Speed and Vize
03:14
The Vision Behind Vize and Hyper-Personalization
06:08
Challenges in Building AI Infrastructure
08:46
Defining Success and Future Aspirations
16:43
AI Speed Podcast Raw Outro 4K.mp4

Video

https://youtu.be/awNet7_DVaw

Transcript

SPEAKER_00 0:20

Welcome to AI Speed, the show where AI-powered companies talk about what actually works in the market right now. Business doesn't move at internet speed anymore, it moves at AI speed, and the people who figure it out how to turn models into money will own the next decade. I'm Evan J. Chalvin, founder of Luxamer and growth partner to high performing brand. Today I'm thrilled to be joined by Runick Mahotra, co-founder of VISE. Runick is building one of the leading AI-driven platforms in wealth management, helping financial advisors deliver personalized portfolios at scale through automation, data, and intelligent systems. His work is helping redefine how investment decisions are made and delivered in evolving financial landscape. Runick, thank you for being here.

SPEAKER_01 1:15

Of course. Thanks for having me.

SPEAKER_00 1:17

So you started building companies very early and eventually focused on financial services. What was that pivotal insight that led you to VIE?

SPEAKER_01 1:27

Yeah, it's a great question. I think there's very few industries that are kind of like absolutely massive in scale and size, like where there's like, you know, multiple hundred billion dollar companies that exist that like are going to be transformed. They'll look incredibly different 10 years from now than they do today. And I think that the, you know, with AI, like, you know, I think more industries fit the bill. But when we started the business 10 years ago, uh financial services and namely wealth management was an example of one where there was, it was, there was not a lot of amazing technology. Um, it was mostly people kind of delivering investment advice somewhat manually. And so we thought, okay, like how do we like modernize this? How do we how do we create a way better experience for the advisor and for their end client? You know, and and I think core to that thesis was, and I think this is AI is an interesting accelerant to this, but core to that thesis was that each person deserves a hyper-personalized portfolio tailored to their needs. So today the average person gets kind of a model portfolio. And so this is a portfolio of a bunch of mutual funds and ETFs that a advisor or investment professional has done some research on, and they create generally like five models. They'll create a moderate portfolio, an aggressive portfolio, a conservative portfolio, and then kind of two in between. And so those portfolios are fairly generic in nature, and then they'll say, Oh, you look like a moderate investor. You've got, you know, you're maybe retiring in 20 years, so you'll be a moderate investor. Oh, you're aggressive because you just started, you know, investing and and you have your whole life ahead of you from a from a um from a time horizon perspective. And everyone kind of gets put into those same model portfolios. Ultimately, we believe that like each client is unique in nature. They have outside investments, they have beliefs, taxes, potentially estate, other things that make them unique in who they are, investments and things that they like. They want and deserve portfolios that are tailored around their needs, right? The ultra-wealthy get kind of this concept of a family office where you hire a bunch of people and they deliver you a personalized kind of outcome based on all the things that are going on in your life. But the average kind of person gets these like very generic kind of portfolio templates, not necessarily something that's hyper-customized to them. So we sort of advise is kind of like how do we deliver that hyper-customization at scale? Uh, and our view is kind of that will be true across industries with AI. Like people are going to expect a hyper-personalized outcome tailored to them as the cost of intelligence drops down to zero over time.

SPEAKER_00 3:49

Yeah. Yeah, that's that's true, true, very true. Um so who do you serve best? What's the ideal advisor or firm profile that gets the most value from BIE?

SPEAKER_01 4:01

Yeah, it's a great question. I would say today we serve kind of large independent wealth management firms the best. So these are kind of what we call enterprise firms. They're the next generation of wealth management firms. Uh, so in the industry, you generally have kind of the institutional firms that have been around for a long time. Think of Merrill Lynch and Morgan Stanley and Wells Fargo. And uh, you know, again, I don't mean to knock those firms, but they've been around a long time. And so, in some sense, they're saddled by the legacy technology and capabilities that they kind of already have. Uh, you then have these next generation of firms. These are folks like wealth enhancement or creative planning or Corient. Um, if any of the listeners have heard of them, but uh they're they're the next generation of these kind of uh uh of these wealth management institutional firms. Uh we service those kind of next gen firms. There's about a trillion dollars of assets in next gen. We think that's going to kind of 10 trillion of assets across, you know, across all these different these different kinds of uh firm types. And uh these enterprise RAs are kind of our core segment. So we focus a lot on how do we build them, the operating system powered by AI, to you know, deliver their clients deeply customized investment advice at scale.

SPEAKER_00 5:12

Yeah. And what would you say has been your biggest challenge growing wise this past year?

SPEAKER_01 5:18

I mean, I think the biggest challenge is how quickly things are changing, right? I think that naturally you build and assume things are static, whether it's in engineering or in your business functions. And so you kind of have to build systems and infrastructure. And the process of building that system and infrastructure oftentimes takes weeks, if not months. So on the technical side, you'd say, oh, these agents are really interesting, but you know, the these kind of coding agents are really interesting, but maybe not necessarily experienced enough to be able to ship code themselves. And so you build a lot of infrastructure around how you prompt them to be able to ship code manual human review with the processes. Three months later, actually the harness has evolved and these agents are actually quite good and they can ship code themselves. Okay, now you have to rebuild all that infrastructure. So you tear all the existing stuff down and you kind of have to rebuild it from there. And then you find out three months from now, oh wow, like the the models actually can test themselves and improve themselves. So, you know, all this invested in you know, investment that I've done on this like kind of benchmarking and testing framework is no longer necessary. But then there are other problems that form. Okay, like now these models need environments that they need to be able to operate inside of. How do you build them their own environments? How do you kind of separate the code that AI writes versus the code that people write in your deployment pipelines? Um, and so from a technical perspective, like everything's constantly changing. And so all the infrastructure you build, all the things that you're doing, you're kind of you have to constantly think about what that will look like. And you have to make bets on where you believe the models and capabilities will go, because the amount of time it takes to build the scaffolding and infrastructure around the models is roughly the same amount of time it takes for the models to evolve and kind of stepwise change. You know, and so I think like you know, whether it's like Claude Mythos coming out or any of the next gen, you know, uh Meta launched uh a new capability of models today. And so, you know, whether it's you know, Meta's new models or Cloud's new model, you know, Enthropics new models or OpenAI's new models, those additional models kind of enable new capabilities that you know could either subsume your infrastructure or require brand new infrastructure you haven't built. On the non-technical side, it's mostly education. Like the models are just getting so much better, so much faster that I think we're finding that like each month we have to re-educate everybody in our organization how to best use them, how to use the new generation of harness, like Claude Cowork, where you're interacting with files and generating content and having kind of this mini, somewhat neutered coding agent that generates that information, how you interact with Vise's kind of databases and you know to get data yourself, how you interact with existing agents, building agents to kind of operationalize current workflows. All of those things are like changing every three months. And so, and as the as the models change on the technical side, they enable a bunch of new non-technical things also. And so we've spent a lot of time kind of building a training infrastructure. We call it Vise Accelerate. Uh, we uh you know we have someone his name is David, he runs Vise Accelerate, and the job of Vise Accelerate is really just to internally transform the organization. You know, and on the technical side, it's how do we think about the infrastructure that supports that? And on the non-technical side, it's literally training the people every day to get better at using AI.

SPEAKER_00 8:22

Yeah, yeah. It's amazing just the turnover, how fast things are moving with AI and just keeping up with it uh you know, every three months or so you're seeing just these rapid changes and you know, keeping up with that is certainly a challenge. I'm curious what challenges you face raising awareness about your company and its products.

SPEAKER_01 8:46

Yeah, I mean, I think that uh um I think that awareness is an important component. I think one of the things that, you know, we we were just talking about this the other day. Um I think there's two kind of components that I think are really interesting. The first component, or you know, what one side of the coin is how do you use AI to get awareness, right? How does marketing transform with AI? And then two, um as the pace of shipping things improve, you know, today, like the pace of how quickly you ship is the limiting regent to get your customers to understand the work that you're doing, not how quickly you can actually talk to them about that work. Um, and so I'll give you an example. Like if you're the average technology company, you ship maybe three or four really big things that year. So your marketing team has enough time to basically educate your customer base on the three or four things that are coming. They have three months for each thing. But if that accelerates from three or four big things that year to like eight or nine big things that year, now you don't really have all of the time to explain to your team, here are all the things that are going on and your customers. And so, this process of how do you give customers awareness of all the things that you're shipping? That's like the limiting region for us. Like that the blocker is like, how do we do that well? Um, and I think today we've thought a lot about what does marketing look like? How do you build automated pipelines that educate customers on you know the capabilities that we have available? And some of the answer is like maybe you give up. Maybe you like trust that the agent for that customer is gonna have the most relevant information. So we work with a handful of firms that have deployed Enthropic and they have Claude kind of on their desktops. Um, and instead of saying, hey, I want my, you know, each of the people using our product to know exactly what features are up to date, we've thought about, well, okay, can we ship updates to our model context protocol or can we ship updates, you know, to this site where if someone reads the site, you know, an agent reads the site, they get the context so that the agent knows what new capabilities are available and can educate, you know, the the advisor. And then the other one is like good old-fashioned email. Like we've, you know, we've automated emailing each advisor personally explaining the new feature we've launched and why it's relevant. Um, and you can automate that whole pipeline. So instead of people having to do that, that process is automated, it feels customized. It's tailored to each person, tailored to the feature they're working on, and tailored to what we know about that customer and why they would use that feature in a relevant way. But you have to kind of rebuild all of the pipelines and awareness processes to uh to kind of match the time horizon of improvements on the business side.

SPEAKER_00 11:14

Yeah, definitely. And so what are your main growth targets for this year for Vise, both in terms of adoption and platform expansion?

SPEAKER_01 11:24

Yeah, I mean, so we have 80 billion of assets that run through our platform today. The goal is to grow that to 200 billion this year. You know, and then we also have AI-powered asset management strategies that we run. We have just under 8 billion in those strategies. We want to grow that to 20 billion this year. So those are kind of the numbers. You know, but I think the numbers are only as useful as the story, right? And the goal of the story is you know, how do you enable more customization for clients across all different types of assets? Uh, and so today, the average client kind of only has mutual funds and ETFs in their portfolio. They don't really get to buy individual names or investments because the advisor doesn't have time to buy those things themselves. I can't do that alongside mutual funds and ETFs, you know, whether I want to run direct indexing or an individual kind of uh actively managed SMA strategy, separately managed account strategy. Um, if I want to invest in private markets, if I want to invest in KKR, Apollo, Blackstone, that's pretty difficult to do all in one account in a way that scales in a mass affluent manner. And so I think the way we get to those numbers is by enabling kind of what we call the whole portfolio solution, one singular platform that lets you kind of construct portfolios of all different types of assets, whether that's you know public market assets, private assets, ETF, mutual funds, individual securities, all in one singular account.

SPEAKER_00 12:44

Yeah, definitely. Well, along those lines, what trends are you seeing in AI and wealth management that excite you the most?

SPEAKER_01 12:50

Yeah, great question. I would say that the biggest trends that we're seeing in AI is like, I mean, AI is a broad term, right? So the question is like what parts, but I think the biggest trend that we see today is uh is is what we call it kind of like, you know, i I would say it's the new generation or new version of Moore's law, which is if you plot the the amount of time it takes to do a task, um, you know, a critical task, and the AI's ability to take on that task. So think about a simple task, maybe it takes a minute. You know, a complex task maybe takes three minutes, a really hard task maybe takes a couple of hours. It's interesting because Moore's Law is this concept of every year, roughly, the amount of transistors on a chip doubles. And so the amount of chip capability kind of goes up and up and up and up in a meaningful way. And so if I'm a kind of like a, you know, someone thinking about computing, you know, every two years, you know, I have a much more powerful chip. Sorry, I think it's every two years, maybe not every year. You know, I think what's happening is the quality of tasks that you can do is changing every three months. We actually think the size of tasks doubles every three months. And so, you know, the AI model can can can fundamentally kind of take on a much more sophisticated task now than it could three months ago. And then, you know, in three months, it will take on even more sophisticated tasks and an even more sophisticated task. Um, and with the advent of these kind of long-running agents, which are agents that work on the same problem for a really long time, our view is that it's likely that an agent can fully solve a task that would take a human one month end-to-end to do, that they can't do. We think that's gonna happen in the next maybe three to six months. And then it might be that an agent can do something that would take a person one year to do, that they they can kind of do end-to-end without any sort of kind of interference. And so I think what's just gonna happen is the complexity of work that AI can take on will go up and up and up and up in a really big way. Um, and I think that that will mean that you know, people to be successful as the world grows is like will need to uplevel their capabilities to be able to do more and more sophisticated things. Because if agents are doing more sophisticated things, then in some sense, people have to be one step ahead. They have to do the like next higher level thing, um, which means that like people just have to grow really fast. If these agents are growing really fast, can people learn as quickly as the agents can? And I think that's probably my biggest fear. I'm not sure. Like, I think that it will be really hard. The pace of evolution of the learning curve that is going to be expected by our team and by people is going to be really, really steep.

SPEAKER_00 15:11

Yeah, yeah, that's fascinating. Um, so if we were to have this conversation again in 12 months, what would you need to happen for you to feel like it's a big win?

SPEAKER_01 15:22

Yeah, I think for our company, I think what win looks like is you know, it is really how do you enable personalization at scale? Like we want to be able to enable a hyper personalized experience for each of our and clients. Um, you know, we have 30,000 different clients today. You know, we want to grow that to 100,000 or 200,000 clients. And so I think that there's two dimensions of that. How many people can we deliver personalization for? And two, what's the depth of the personalization we can deliver? Like, how can you kind of democratize what would have been available to ultra high net worth or really sophisticated investors down to kind of like the average person?

SPEAKER_00 15:59

Yeah, absolutely. Well, very cool. Um, well, that's it for today's episode of AI Speed. A huge thank you to Runick for sharing his invaluable insights into how Vise is helping financial advisors deliver personalized portfolios at scale and for navigating the future of AI-powered investing. If you're building or leading an AI native company or service business that uses AI under the hood and you care about revenue adoption and market share, make sure to subscribe to AI Speed. Learn how the best AI operators ship faster, sell smarter, and stay ahead. Thanks for listening. Until next time, keep building, keep selling, and keep moving at AI speed.