Podcast: Play in new window | Download | Embed
Subscribe: RSS
Podcast Intro
Here at Ezra Group, we’re experts on everything wealthtech, including CRM, portfolio management, trading, rebalancing, performance reporting, just to name a few. When we start working with an RIA or broker dealer, the first thing we do is a comprehensive tech stack assessment. This provides a top to bottom view of all systems and processes, and it’s a critical part of the firm’s growth plan, since the tech stack is the foundation for building towards the future.
So if you’d like to see your tech stack converted from a liability into an asset you need to run not walk to our website, EzraGroup.com, and click on the golden Contact Us button at the top of the homepage, the experienced team at Ezra Group will conduct a detailed tech stack assessment for you, delivering targeted recommendations that will optimize your existing software platforms. Or we can run an RFP process and help you select and then implement a new solution to help take your firm to the next level. You can schedule a free consultation by going to EzraGroup.com.
A few quick housekeeping tasks before we continue.
- Please subscribe to the show wherever you listen to podcasts so you don’t miss an episode.
- Check out our sponsors, the Invest in Others Charitable Foundation. You can find them at InvestinOthers.org.
- If you are a wealthtech vendor, please register for the Ezra Group WealthTech Integration Score Portal where you can update your integration data in real-time and improve your score.
Now let’s kick this thing off!
Boost advisor confidence.
Episode Transcript
Craig: All right. I’m excited to introduce my next guest. This is Jeremi Karnell, Head of Envestnet Data Solutions. Jeremi, brother, good to have you on the program.
Jeremi: Thank you for inviting me on. I know it’s the end of the week during the holiday season, and it’s after a holiday party we had last night. Forgive me if I struggle.
Craig: I know. You’re paying the price for all the good fun you had.
Jeremi: A 1,000% percent.
Craig: I went to bed early last night.
Jeremi: Oh man. It was for the podcast. I’m sure you did. I’m shot out of a cannon, and I’m going to drag you along.
Craig: You’re going to need to.
Jeremi: You’re going to need to. That’s all good.
Craig: Where do we find you today, Jeremi?
Jeremi: Chicago.
Craig: I recognize those buildings out there, those car parks.
Jeremi: Those buildings—first time I ever visited Chicago, I think when I was 6 or 7, late 70s to date myself, was visiting one of those buildings.
Craig: Family had a friend in there. For people who can’t see, it’s those cylindrical buildings in downtown Chicago with the car parks in the first 20 stories—cars in a circle. I always take a picture of it when I’m there.
Jeremi: It’s iconic. Very cool.
Craig: All right, let’s jump in. Give a 30-second elevator pitch for Envestnet Data Solutions and how that differs from the rest of the business.
The Scale and Utility of Digital Exhaust
Jeremi: We process the digital exhaust that comes off the $7.4 trillion of assets under management.
Craig: You said it. You said “digital exhaust.”
Jeremi: Digital exhaust. I did, so take a drink. We said this earlier this week—it’s going into a drinking game. That’s a shot.
Craig: Absolutely.
Jeremi: I work with some of the finest data analysts and data scientists in the country. We help Envestnet aggregate the institutional wealth data that fuels our platform. Mutual fund and ETF data, brokerage data, managed account data—we stitch it together, reconcile it, pull it into our business and decision intelligence models, and feed it back to our three core segments: asset managers, wealth enterprises and their advisors, and RIAs. Also, this happened this past week: the Envestnet data integrations team got folded into Data Solutions. They take care of Open ENV, the developer portal, the APIs that go back out to customers, as well as third parties.
Craig: Does the data integrations team report to you?
Jeremi: Mm-hmm. 100%.
Craig: So I can yell at you when the integration team—
Jeremi: You can. You’d be within your rights. The Open ENV thing is working better.
Craig: I know.
Jeremi: I wanted to thank you and the Envestnet team for inviting me and my team to the consultants forum earlier in the week.
Craig: There were a couple interesting things that came out of it. We’ve known a lot about Data Intelligence, Data Solutions. Before you got there, we’d been working with the previous team on what it should be. I’m excited to see it come to fruition.
Jeremi:
Craig: Talk about the decision intelligence orchestration layer and how that’s going to benefit Envestnet clients.
Decision Intelligence and the Evolution of Advisor Judgment
Jeremi: Our viewpoint on AI and machine learning adoption in wealth management is that the industry is focused on episodic tools—AI note-taking, etc. We have a broader strategic view. Clients don’t pay advisors for speed. They pay them for judgment. They expect advisors to help them make some of the most important decisions of their life. Envestnet sits in a unique spot because of the data we aggregate, stitch together, and provide back as next best actions, so advisors can use limited time and serve the broadest set of households with the best thing they should be doing—portfolio, financial planning, etc. We help them make grounded decisions: data-driven, leveraging client and advisor knowledge graphs, propensity models, and predictive analytics. We’ve created a scaled platform that generates 25 million insights a day for advisors. We continue to evolve it—building machine-learning-based predictive analytics—offering the ability to engage with that data through agentic frameworks, dynamically generate content, and generate views that may not exist within our API libraries. We’ll continue building on that with AI explainability next year and democratizing access to that data in native AI enterprise platforms so they don’t always have to log in to UMP, Tamarac, MoneyGuide, or the Wealth Data Platform. They can get it natively within Copilot, Gemini, or Claude. That’s where we’re headed.
Craig: When I rub my chin this way, I need you to take a break so I can get a question in, Jeremi. I’ve got to fit a couple more.
Jeremi: Sorry about that. I’m joking.
Craig: It’s all about you.
Jeremi: You said a couple great things I was going to talk about anyway.
Craig: Talking about decision intelligence as a category: do you see competitors doing this, and will it become table stakes in WealthTech platforms?
Jeremi: I see some advanced competitors headed down this path. We weren’t first to market. I think that was Morgan Stanley.
Craig: I—
Jeremi: It makes sense. I think big fintech competitors are heading there. I think wealth enterprises and RIAs that are building solutions—our hope is they continue to depend on us for this. I don’t see signs they’re heading down this path. They’re still lagging in many ways as far as innovation in this regard. That’s fine with us because that’s where we’re investing our time and energy, and we hope they take us up on it.
Craig: You mentioned knowledge graph and propensity models. Explain what those are.
Knowledge Graphs and the Precision of Predictive Modeling
Jeremi: A knowledge graph is a big map. We’ve got different entities mapped—financial products and services—and we map their relationships: investor to investment, assets to portfolios, etc. It’s made up of over 100,000 advisors—about one-third of the industry—and over 12 million investor accounts. Envestnet sits on about 24 to 25 million investor accounts. This represents about 50% of that, if not greater. That’s the cornerstone and engine of our propensity models and predictive analytics. Earlier this year, we made predictions based on that knowledge graph. The top 5,000 predictions were 90% precise in predicting a buy transaction, and 98% were net new. Our FSP and SMA predictions—15,000 of them—were 98% and 99% accurate and represented almost 70% of net inflows for Envestnet in the first 6 months.
Craig: You said 15,000 insights and 98% accurate. Give me an example of an SMA insight that was accurate.
Jeremi: They were predictions of which advisor within a specific enterprise or RIA was going to make a purchase of an SMA or an FSP project.
Craig: Predicted certain advisors would.
Jeremi: We knew exactly who would do it. With 98% to 99% precision in the first 6 months, we knew exactly who would do it.
Craig: Precision meaning buying an SMA, or the amount they buy?
Jeremi: Both, but it was going to buy.
Craig: A dollar or a billion dollars is a big difference.
Jeremi: I get that. We knew who was going to do it, and it represented 70% of the inflows into Envestnet. Whether we predicted the exact amount—I have to go back and look—but we knew exactly who would do it.
Craig: Asset managers must be excited by that.
Jeremi: That’s our advisor engagement data product that we work with asset managers on, and also home offices for wealth enterprises that want to work more closely with advisors around these topics.
Craig: If asset managers can know with some certainty that certain inflows are coming in, that’s beneficial to their business plan.
Jeremi: When we piloted this with BlackRock about 2 years ago, we worked with a mid-sized broker-dealer and their advisors. We looked at inflows in the previous 12 months for FSP and UMA and compared it to the 3 months after we launched. It was programmatic: BlackRock wholesalers in the field engaging directly with advisors, with home office support. It was a 35% and 38% increase in net inflows in the first 3 months of that program versus what they were doing over 12 months.
Craig: Because they were using the program. It increased 35% to 38% versus what they thought they were going to get.
Jeremi: Exactly. Using our insights.
Craig: That’s right.
Jeremi: With great field support by the asset manager. In this case, BlackRock.
Craig: Great field support. This assumes certain market conditions. If the market crashed tomorrow, your predictions would go out the window.
Jeremi: Probably. Black swan events would impact that.
Craig: Tough question. Envestnet has the innovator’s dilemma. You’ve been successful over the past 25 years, so you have legacy infrastructure and you’re struggling to replace it, like all legacy vendors. If you take the AI layer away, are there other structural advantages Envestnet has that a well-funded startup can’t replicate?
Jeremi: The data. No one has that size and scale, especially startups. The digital exhaust on $7.4 trillion of assets under management—that’s what no one else has.
Craig: Great. That’s the data.
Jeremi: Craig, it’s good data. People run into the stat that 90% of AI POCs are failing. There are a lot of factors, but the biggest is data. People don’t understand AI needs good data to be effective. The data we aggregate goes through strong reconciliation at scale. We feed AI models, machine learning models, deep neural networks, and the knowledge graph on trade-ready data. That’s a structural advantage.
Craig: You mentioned you generate 25 million insights per day. You’ve seen predictions come true. What about insights advisors don’t act on? What does that tell you about trust versus workflow friction?
Jeremi: It’s a good question. You’re too strong for me in the morning.
Craig: It is.
Jeremi: We know the insights they’re acting on—portfolio optimization insights, tech strategy insights. The ones they don’t get to, we don’t know if they’re ignoring them.
Craig: It’s a Venn diagram. You’ve got insights they act on and insights they don’t act on. You should know which they are.
Jeremi: I know the ones they act on. You just got me the questions and I didn’t have time to prepare.
Craig: If they didn’t act on it, does that mean it’s the wisdom of crowds, or something they’re missing?
Jeremi: It’s time. Advisors are struggling with time to do the work needed to service their book. It’s going to get more acute because the advisor population is shrinking. The industry is going through a big transition. We’re not going to fulfill the same service levels over the next 10 years to meet current demand. There will be fewer advisors servicing bigger books. They’ll have to leverage tools to do it efficiently and effectively. If they don’t get to specific insights, it’s time. We’ll have to help solve for that.
Craig: Are you seeing areas where AI is breaking down in real usage—where the demo looked great, but when you start to advise it didn’t work?
Jeremi: On our side or in general?
Craig: I prefer Envestnet, but it could be anywhere.
Jeremi: I don’t want to throw anyone under the bus. I don’t think we’ve seen that because we’re in early adoption. We don’t have a longitudinal view into something that hit the market long enough to say it didn’t work. Everyone is testing and going through compliance review. We’re early in adoption.
Craig: I don’t think we have enough data to answer that. How about prompting? A lot of tools say you can prompt and ask any question. Advisors aren’t great prompt engineers.
Jeremi: They’re not. That’s why you lean into AI. The chat experiences we’ve built come with 3 to 6 prompts out of the box that are contextually relevant to that advisor’s book of business based on our analysis. Those prompts change as they select a prompt or give their own prompt. We released version 4.0 of the Wealth Data Platform last week. That introduced a robust agentic framework. It’s 6 dedicated agents with specific roles: receiving a prompt, researching the data, understanding the prompt context, generating a response, and another agent reviewing the response in detail. Does it make sense? Does it fit compliance frameworks? Is it a good response before presenting it back? It’s a system that will continue to support them.
Craig: Thank you for bringing that up. What is the Wealth Data Platform? Why is it important for Envestnet clients? What’s so great about the 4.0 infrastructure?
Empowering RIAs Through the Wealth Data Platform 4.0
Jeremi: We started building the Wealth Data Platform about 3 years ago to move our business intelligence experiences into a modern era. It’s a dedicated BI portal within the Envestnet WealthTech stack and it services all three segments: asset managers, wealth enterprises and their advisors, and RIAs. Version 4.0 addressed RIA Analytics. We moved Tamarac customers into this new infrastructure. Every segment is in the Wealth Data Platform. More importantly, we invested time this year decomposing the Wealth Data Platform and integrating it into UMP, Tamarac, MoneyGuide, and the proposal engine to reduce swivel-chair for advisors. Home offices are heavy users of the Wealth Data Platform, so it makes sense for them to go into that dedicated experience. Advisors have limited time. We want insights embedded into daily workflows with minimal effort to see them and act on them.
Craig: Can you explain what these agents are? Will the home office know they’re doing it? Do they have to interact with them or configure them?
Jeremi: They operate in the background. You were in my session at Elevate in Vegas earlier this year.
Craig: I heckled you a little bit.
Jeremi: You did. When we released Insights AI, it required a lot from home office users and advisors—prompting back and forth. We wanted to reduce that workload. Instead of users driving to the outcome through multiple prompts, we put in a new architecture with dedicated agents under the hood. No one needs to engage with them separately. We’ve got an agent that deals with a prompt, an agent that researches, an agent that generates content, and an agent that reviews the outcome before it’s shared.
Craig: How does that help clients?
Jeremi: Better insights faster for the advisor. Less workload going back and forth with AI. Best output in the shortest time.
Craig: They’re not going to see things running back up.
Jeremi: No. That’s how the sausage is made.
Craig: That’s exactly right.
Jeremi: Craig, I’ve got to do this for them.
Craig: I’m an agent. We’re running out of time.
Jeremi: I am. I’m so sad.
Craig: Biggest benefit from your presentation was the AI dashboard builder. You can ask the AI: I’m an operations person, I spend most of my time on onboarding, build me a dashboard with what I need—and it’ll do it.
Jeremi: That’s generative business intelligence. Before that, if an enterprise wanted to see data and it didn’t fit within our fit-to-purpose APIs, there was friction. It would have been time and money. It would have been professional services to create something bespoke. That’s tech debt for us. To shift focus toward decision intelligence, we leveraged AI to pick up that work. We created natural language query: ask for a graph, widget, dashboard. If we have the data, it will send that query. We’re leveraging OpenAI structured output to dynamically generate that API in real time and display the data how they want. No professional services, no fees, instantaneous.
Craig: It gets us out of manufacturing that stuff.
Craig: Last question. Looking 3 years out, where do you see AI adoption not meeting expectations, and where do you see it exceeding?
Jeremi: In 3 to 5 years, AI will shine in dealing with downward pressure on advisor time—fewer advisors, more clients. Real adoption will come when tools are native in environments advisors already use—Copilot, Outlook, PowerPoint, Excel—so they don’t have to go anywhere. The data is securely in that environment and they can leverage multiple models. That’s where it shines—when it’s baked in. Huge adoption. Where it falls short—I’d be hesitant to predict that.
Craig: It’s never going to replace advisors.
Jeremi: We’re in a liminal phase where the advisor is central, and AI is a co-pilot. Traditional platforms Envestnet, Orion, and Pershing are advisor-focused and driving AI adoption. Robinhood is tech- and AI-focused out of the gate and will bring advisor experiences as young investors accumulate wealth. The types of clients advisors service will be AI native.
Craig: They’re going to come to meetings with their own co-pilot.
Jeremi: If advisors aren’t leveraging the technology, they’ll see decompression and massive churn, because investors will run circles around them.
Craig: I agree. You’ve said it all, Jeremi. Tell everyone where they can find more information about Envestnet Data Solutions.
Jeremi: Go to Envestnet.com or search the Wealth Data Platform or Insights Engine. Look me up on LinkedIn. It’s Jeremi with an I.
Craig: Jeremi, you’ve been a great guest. Thanks so much.
Jeremi: Thank you. Appreciate it.
Summary: The Future of Decision Intelligence in Wealth Management
This episode highlights the critical role of data reconciliation and AI-driven insights in modernizing the wealth management industry. By leveraging the Wealth Data Platform 4.0 and advanced predictive modeling, firms can reduce “swivel-chair” friction for advisors and provide more personalized, data-backed financial planning. As the industry moves toward a “liminal phase” where AI serves as a co-pilot, the ability to integrate these tools natively into everyday environments like Excel and Outlook will define the next generation of successful RIAs.
Boost advisor confidence.
