1. Podcasts
  2. WealthTech Today
  3. The Human in the Loop: AI Agents for Estate and Tax Planning with Danny Lohrfink of Wealth.com
WealthTech Today

7 Nov 2026 · 16 chapters

Episode artwork

The Human in the Loop: AI Agents for Estate and Tax Planning with Danny Lohrfink of Wealth.com

In short

Wealth.com’s integration into Anthropic’s “Claude for Financial Advisors” ecosystem, the shift from read-only AI to agent workflows that can perform CRUD actions, and governance requirements for a “human in the loop” before changing client data.

Guest backgrounds

Danny Lortfink, co-founder and chief product officer of Wealth.com; Wealth.com was founded by ML builders who previously built/sold a machine learning platform to LexisNexis.

Key claims

Anthropic invited Wealth.com after interviewing wire houses, broker-dealers, and large RIAs about estate/tax planning needs (not pay-to-play). Hallucination is reduced because Claude triggers Wealth.com tools/skills that run deterministic estate distribution calculations and return auditable numbers plus workflow backup. Advisors already use frontier AI tools (Wealth.com survey: ~60% use Claude/ChatGPT; ~95% use some frontier model tool), so firms should enable safer enterprise controls rather than banning. Agents should not autonomously execute create/update/delete without showing downstream impacts first.

Notable examples

Estate distribution analysis for beneficiaries using deterministic calculations; advisor workflow for a client moving from New York to Massachusetts requiring reruns of prior tax/estate reports.

Written by AI. May contain mistakes.

Chapters

2:33

Wealth.com Overview

Danny explains Wealth.com's focus on advanced planning software.

“It is none other than Danny Lortfink, co-founder and chief product officer of Wealth.com.”
3:58

AI in Financial Services

Discussion on the integration of AI in financial advisory services.

“Wealth.com is the leading advanced planning software for wealth management institutions.”
6:12

Integration with Anthropic

How Wealth.com became part of Anthropic's ecosystem.

“And the world's greatest CMO, Dan Bolton, who's on this call, but you can't hear him or see him, has also been traveling a lot.”
8:21

Data Control and Workflow

Explaining how Wealth.com manages data and workflow with AI tools.

“You're on the screen with Anthropic and all these other vendors.”
10:44

The Role of AI and Human Interaction

The importance of human input in AI-based financial tools.

“Then once it gets back into Claude, here are all the numbers.”
14:00

Understanding User Behavior in WealthTech

Learn why understanding user behavior is crucial for developing effective AI solutions.

“You're kind of stopping the innovation, right?”
14:31

The Evolution of AI in Advisor Workflows

Discover how AI can streamline advisor workflows through self-learning systems.

“And when you actually look at the human's real behavior, that's when you get the real unlock because you can see what their day to day actually looks like.”
15:38

Leveraging Wealth.com's AI Agent for Document Analysis

Explore how Wealth.com's AI agent enhances trust document analysis and client interaction.

“So I would argue that it's actually a good thing if you're looking at the behaviors of how advisors are operating.”
18:34

Integrating with Existing Systems

Learn how Wealth.com integrates its tools within existing financial advisor platforms.

“And then any information that you've loaded in there.”
19:16

Navigating the Confusion of AI Tools

Understand the differences and challenges financial advisors face with various AI tools.

“Some of our large broker-dealer clients, there's usually a big disconnect between the IT team, which loves Microsoft.”
Show all 16 chapters
21:37

The Control Over Client Data in AI Usage

Examine how financial firms can manage client data security while using AI tools.

“Now, you think about that like from a firm perspective.”
23:07

The Role of Human Oversight in AI Tasks

Discuss the importance of human oversight in managing sensitive operations performed by AI.

“But you were talking about a couple of things I wanted to bring up and kind of break apart some of what you just said.”
28:00

Understanding Agent Autonomy

Explore the complexities of AI agent autonomy in financial planning.

“But the question is, do you want the agent to run autonomously?”
29:41

Future of Wealth.com and Industry Trends

Discuss the future developments and industry trends related to Wealth.com.

“So where do you see wealth.com, you know, if we had this same podcast next year, where do you think we're going to be?”
31:46

Build vs. Buy: Strategic Decisions in Tech

Evaluate the strategic considerations of building tech versus buying from vendors.

“And to have everything working in in orchestration.”
34:15

Deepening Capabilities in Tax and Estate Planning

Learn about Wealth.com's approach to enhancing tax and estate planning capabilities.

“Now, what are we actually going to build?”
A man walking through an airport, asking his phone a question

Ask about this episode

Get the gist, explore an idea or ask about something said.

Claude

On WealthTech Today, what are the key points of this episode?

Searching your listening

Transcript

Automatic transcript. May contain errors.

0:00Welcome to the Wealth Tech Today podcast. I'm your host, Craig Iskowitz, founder of Ezra Group Consulting. This podcast features interviews, news, and analysis on the trends and best practices all around wealth management technology. My guest for this episode was none other than Danny Lorfink, co-founder of Wealth.com. And we spoke about a couple of interesting topics. Number one, how Wealth.com got into the Cloud for Financial Advisors ecosystem, which was announced at FutureBruf in September and is continuing to be expanded. We're talking about all-in-one platforms versus best-of-breed tools and how advisory firms are using those.

0:59And finally, we're talking about why AI agents need some sort of human checking before they are allowed to change client data or make important decisions. Speaking of AI agents, every vendor in your tech stack is delivering some kind of AI-powered functionality. Your CRM, your planning software, your portfolio management system all have some kind of AI built in now and it's not stopping. You're receiving these features without even knowing which ones really work, how they connect to the rest of your tech stack, or whether your data is ready for them. Ezra Group's AI assessment for RIAs answers those questions.

1:32We'll review every system in your tech stack and rate the AI features you already own. We'll also test whether your data and integrations can support these new AI agents. You can check out our AI agents directory, which is tracking more than 250 agents and MCP servers so we know what's real and what's just a demo. You'll walk away with a roadmap that shows which features to turn on, which gaps to fill and which vendors to drop. So if you want to free up your firm for exponential growth, you need to run, not walk, to our website, azuregroup.com and click the big purple contact us button on the homepage to schedule a free consultation.

2:07All right, standard housekeeping chores, yada, yada, yada, you know what that is. Please make sure to check out our sponsor, of the Invest in Others Charitable Foundation. I won't yada, yada, yada, the Invest in Others Charitable Foundation at investinothers.org. And if you're a software vendor in the wealth management space, please make sure to register your product integration score vendor portal at ezragroup.com forward slash register so your product's score will always be up to date. And now let's kick this thing off.

2:46All right, all right, all right. I'm happy to introduce our next guest. It is none other than Danny Lortfink, co-founder and chief product officer of Wealth.com. Danny, it was great to see you again. Amazing to see you too, Craig. Amazing. Amazing. It's amazing to see you. You know, it is Friday. I told you earlier before we started the podcast that I have acquired my pumpkin spice latte, my first one of the season. So amazing feels it's pretty like that. That's the right word. Yeah. Some people are really into pumpkins. I'm not, I'm a little, I'm a little pumpkin ish, but I'm not like totally over the top of pumpkins.

3:26You know, it has its short window. I don't like that. Starbucks started advertising for it on September 1st. I think that's a little bit overkill, but for two or three weeks, it has its purpose. I like pumpkin pie. That's about it. That's the limit of my pumpkin-infused things. Oh, man. We can talk about pumpkins all day long, but this is not a pumpkin podcast. This is a tech podcast. So, Danny, can you please kick us off with a 30-second elevator pitch for Wealth.com? Sure. Wealth.com is the leading advanced planning software for wealth management institutions. So when you think about advanced planning, these are pieces of the puzzle that go into a broader financial advisory relationship, particularly tax planning and estate planning.

4:14We specialize in those two domains. And so we deliver specialized technology that helps advisors to automate the process of both helping clients understand their current circumstances and then optimize on a forward going basis. Of course, we were founded right at the start of the AI boom. My co-founders, they built and sold a machine learning platform to LexisNexis. So that was kind of like in our ethos. So, of course, the entire process of analyzing and then optimizing has AI infused throughout to make things seamless for advisors. Awesome. It is truly a new and exciting age that we live in. Things are happening so fast.

5:03My favorite quote is, the rate of change of technology has never been as fast as it is today. Yeah. But it also will never be as slow as it is today. Oh, totally. I mean, you just see that hockey stick curve, and people always like to throw this out there, but the fastest mode of transportation, If you're looking at the Roman era versus you're looking at the 1400s going into even the 1700s when we had our friend George Washington just right down the street from me crossing the Delaware River, fastest mode of transportation. What is it? It is the horse, right? And then all of a sudden things start to take off.

5:47And we're in this phase right now where it feels like things are moving so slow, but also so fast at the same time. It's almost like that movie Interstellar where it's like you're on a different planet and you're living your life as if things are just moving at the normal rate of change. But then on Earth back home, 32 years have passed. That's kind of the way that it feels right now operating in this industry. That's a great one. I never thought. I never heard that analogy before. That's perfect. Yeah, that's really good. And I've been traveling a lot. And the world's greatest CMO, Dan Bolton, who's on this call, but you can't hear him or see him, has also been traveling a lot.

6:32And I wanted to talk about Future Proof, which was two weeks ago already, as time is flying. And there was a big announcement, I think, at the conference with Anthropic. And you guys were part of that, where you're one of the vendors that are now integrated into the desktop at Anthropic. Can you talk about how that came about? What did they promise you to get in there? How did you manage to get, you know, to become one of the vendors? Is it a pay to play? How does the whole thing work? And what's the benefits for advisors? Yeah, so they came to us. It's definitely not a pay to play. They came to us after they went around and for the better part of a year, were talking to leaders of wealth management institutions, big and small.

7:20So they went to all of the various wire houses, talked to the presidents and CEOs of the private wealth businesses at those wire houses. They went to all of the broker dealers, talked to the senior level executives at those broker dealers and the largest RIAs in the country. And they simply were asking, like, what technology would you want to be part of this Claude for Financial Advisors initiative? And, of course, estate and tax planning comes up as a necessary piece of the puzzle for servicing clients' needs. And so then they ask the natural follow-up question of, all right, great, like, what vendor are you using and which vendor are you considering using?

8:03And that's ultimately why they came to us. It's as simple as that. They just looked at distribution and demand and who is kind of like innovating the fastest, at least from what they told us. And we were very grateful to have that invitation. Excellent. So being part of that limited set seems pretty prestigious. You're on the screen with Anthropic and all these other vendors. So now you're sort of in this whole ecosystem where you're not directly interacting with the client. So the advisor is now in Claude doing his work. And he might say, oh, now ask wealth.com for this information. So you're sort of in between.

8:44So how do you know the data you're providing or the answers you're providing are being delivered right to the advisor and Claude is hallucinating somehow? Yeah, well, because it's our tools. And so the way that it works is you have tools that are then stitched together to create a skill. Right. And so a tool call could be like, all right, first, we need to get a list of all of our clients. Then we need to select a client. Then we need to look at that client's balance sheet. Then we need to look at that client's estate planning documents. Then we need to run an analysis that analyzes how that balance sheet, which is a snapshot in time, will then distribute to the next of kin, right, as one person dies or another person dies based off of the terms of those documents.

9:28And that's not being done in the head of the LLM. Claude's not doing that. That's wealth.com, right? And so the advisor might say, hey, run a estate analysis for my client. It then triggers the skill to run. It hits all those tools. So list client, identify this specific client, look at the balance sheet, look at the documents, and then you feed that into our calculation engines. And so when you think about all the different components that go into a state distribution, you're talking about direct beneficiary listings. You're talking about distributions at certain trigger events to the beneficiaries.

10:07But then there are proportions that go into one bucket versus another bucket. One could be filled with the available federal exemption. The other is with the remainder or something like that. And all that needs to follow this like deterministic calculation. Right. And so that's what we have. We have a deterministic calculation engine where the data from the balance sheet and the documents get shaped into that calculation engine. We run the numbers. We then pass that back to Claude. Claude surfaces the number. Right. And so there's no hallucination in that process because we have full control over that entire skill and that entire workflow pipeline.

10:44Then once it gets back into Claude, here are all the numbers. Here's the entire workflow process that you can then audit because that is part of the skill. We give you not just the final output, like Johnny gets$10 and Susie gets$20. We give you all the backup data of how we got there. And then Claude has that at its disposal to be able to then go on with the advisor's questioning. So being part of that ecosystem, does the RIA, first to get this advanced technology on Anthropic, they have to be a Claude for enterprise client. So they've got to pay for that. then are there any special tiers of the Wealth.com subscription that they need?

11:34Yeah. So good question. In fact, they don't necessarily need to be a Claude Enterprise. They could be on a Claude Teams account. They could be on a Claude Enterprise account. In fact, we actually do know that there are some people that are just on Claude individual accounts that are able to go in to Claude for Financial Advisors and turn on the plugin. Here's the big difference, though. When it comes to enterprise level security, the Teams account and the enterprise accounts, those are the ones where by default, you have data sharing and model training and all that stuff and data storage turned off.

12:11Right. And so that's why you see almost every institution saying, hey, look, we're supportive of you leveraging AI. But what we need to make sure of is that your client's circumstances are not being used to train the model. And so that's also our recommendation, obviously, within wealth.com, like no one's no no client data is being used to train any models because we have the reps and warranties from whether it's Gemini or or or Claude or OpenAI that there's no training agreements in place and there's no data storage agreements in place as well. So that pushes me to ask another question, which I think is sort of, there's a fine line here between what is happening with these AI tools.

13:03Yes, they're saying, we're not training on your client data. Aren't they really training on everything else? Every question the advisor asks, every report they will look for, every way they want to look at the data, everything they're entering, all the inputs, all the prompts they are training on. I think if you look at query response pairs, you are getting a better sense for how a financial advisor is trying to accomplish their work. Right. And so if you are rolling out Claude for financial advisors, you want it to be as effective as possible for that financial advisor in the work that they're conducting for their clients.

13:51And so if you're not looking at the query response pair and figuring out, all right, how do we actually then surface the next question preemptively to the advisor, knowing that this is their likely next step? You're kind of stopping the innovation, right? And so I think it's actually a really important thing that you are continuing to learn about the actual behaviors of the users, not training on any client data, but understanding how advisors really like to operate. Because one thing that we talked about before this podcast started, Craig, is humans are really bad predictors of their own behavior.

14:37And when you actually look at the human's real behavior, that's when you get the real unlock because you can see what their day to day actually looks like. Right. And then you can develop solutions to match that in the age of AI. that can be done via a self-learning loop, as opposed to it taking a software engineer months at a time to sit with an advisor, like a PM, like a product manager, to sit with an advisor, understand their processes, like download their conceptual workflows, translate that over, even though that could be biased, translate that over into a written PRD or product requirements document, work with a designer to build out this end-to-end workflow of how the client advisor wants to work, then pass it over to an engineer to code and build.

15:29Instead, it's literally just a self-learning loop where now the model can get smarter in anticipating the next step that the advisor is going to want to take. So I would argue that it's actually a good thing if you're looking at the behaviors of how advisors are operating. The clod for enterprise connectivity is through wealth.com's MCP server? Would that be a correct statement? Yeah. So is that MCP server available for wealth manager firms that don't use Claude? Yeah. So we are turning it on for folks that are leveraging

16:14ChatGPT, so OpenAI, Gemini. What's interesting is Copilot. People might be curious about Copilot. They have basically not the level of sophistication today as the other frontier model providers. And so you have to jump through a million different loops to basically package your plugin as an application that can then get deployed. It's not slick. We're just waiting until they kind of catch up to everyone else before we make it available through Copilot. co-pilot. With that being said, Craig, we actually also have something else, which is our SDK. SDK, for people that don't know, it's software development kit.

16:59Basically, we give you like old school, Danny. Danny, you guys are going old school. Going old school. So think of it this way. We have our AI agent in wealth.com. You upload a trust document and you say, summarize this. We'll give you the full summary, not just going to the LLM and doing what a normal LLM will do. But like we have like infrastructure and guardrails around it to say, look for this stuff, look for that stuff. If it's a grant, we're going to look for annuity terms. If it's two slats, we're going to look to see if they violate the reciprocal trust doctrine, right? Then the output is also structured the same strict way every single time.

17:34So it's reliable and dependable. But then we produce flow charts, right? And so then we'll go and produce, if you, again, just upload a single document and say summarizes, we'll produce a flow chart that shows you exactly how everything works, that is ready to go and print to go show to your client. And that right there within our agent is actually embeddable within your website. So if you are Premier Wealth Partners and you have your website that you log into and now this is your desktop upon which you operate, rather than you having to necessarily log into wealth.com to upload a trust document to then get the summary and get the flowchart and whatnot, You just have this embedded little button that says like estate and tax analysis.

18:21You click it and it opens up our wealth.com's agent, essentially. And you're able to do all that work right there without ever leaving your Premier Wealth Partners desktop. Right. And then any information that you've loaded in there. So if it was like the client's trust document or if it's a client's tax package that still gets stored into the wealth.com vault. So if you want to do more robust analysis and work when you log into wealth.com, so now you're ready to swivel, right? Like before you're like, I'm just doing a quick thing, but now you're ready to swivel. That work that you did before is not lost.

18:56Like that doesn't just like go lost in the ether. It actually gets transferred into the system. So I give that response, Craig, because even for people that might be using CoWork, we also have this other embeddable option so that we're able to reduce that swivel of having to log into 10 systems at a time. Funny you mentioned how difficult Microsoft is to work with. We found similar issues. Some of our large broker-dealer clients, there's usually a big disconnect between the IT team, which loves Microsoft. and they're all Microsoft shop. They want to push Copilot everywhere. And the advisors usually want Claude or OpenAI directly access.

19:39And they're always fighting. But what people don't realize is that Copilot is just a harness on top of Claude. For the most part. So you're really getting the same thing. If you want Claude and you've got Copilot, well, it's kind of the same thing. They just have implemented it very differently and use a lot of different terminology. So it makes it very confusing. and pricing is different and delivery is different and all different programs are different. I actually wrote a blog post about it on a WealthTech Today blog that compares Claude versus Copilot and how they talk differently. So yeah, this can be very confusing for clients to try to wade through these different tools, even if they're using the same underlying technology, trying to make it work, figure out what's going to cost more and how you can deliver it and what they can do with it is confusing.

20:28Definitely. Right. And one thing that I'd say, because I mean, you kind of brought it up, advisors, they are using Claude and OpenAI today. Like, make no mistake about it. Like, they can have all the technology in their tech stack. They can have, like, you know, all the cool AI tools out there. They still are opening up ChatGPT and Claude. How do we know? Well, Wealth.com actually ran a in-app survey of advisors. And we partner with tens of thousands of advisors around the country, right? So we got a lot of responses. And we saw Claude and ChatGPT commanding roughly 60-something percent of advisors that were using one of those two tools.

21:14And then you have Copilot and some others that round it out so that 95%, 95 % of financial advisors, at least that are Wealth.com partners, are using one of these tools, frontier model tools. OK. And so like going directly to Claude or to ChatGPT, 95%. Now, you think about that like from a firm perspective. Like, let's say you're major RIA, you have 500 advisors under your umbrella. Do you want to be the firm that just kind of like turns your head and says, no, no, like we don't want advisors using Claude at all. And so we're just not going to approve Claude for financial advisors. and we're not going to approve the enterprise licensing because we don't want them dumping client information into one of these models.

22:14Well, what do you think they're doing? If 95 % of them are saying that they're already using these tools, what do you think they're doing? They have their individual accounts and they're dumping the client information in without necessarily opting out of the training and the data sharing and whatnot. And so if you are a RIA, hey, just know that like you can, this is totally in your control to say, all right, no, we know that you're already using your individual accounts. Let's just actually enable a enterprise level access where we then have control over the security measures, right? Because that way is the more dangerous way, right?

22:57This way is the safer way. We know that advisors are doing it. We know that their clients are doing it. Their clients are putting their data directly into one of these models. So it's sort of the Wild West. But you were talking about a couple of things I wanted to bring up and kind of break apart some of what you just said. So we've got the ability to get to wealth.com through Claude for Enterprise, that harness. You've got the SDK where firms can build capabilities directly connected through if they've got their own tools and technologies or dashboards and things and plug it in. and you mentioned agents.

23:36Now, your CEO, Rafael, was on the podcast last year, and we talked about a future where advisors let agents handle specific tasks. And the future is now, right? So we're already there. So there's another way that we can do this work. So do you have a capability with your agents that advisors can create multi-level tasks for use cases for some of the things you mentioned around estate planning and tax management and have the agents do it by interacting with like Esther or other parts of wealth.com? Yeah. And so you can construct like a multi-step workflow. We like to ensure that there's always a human in the loop check before any like CRUD action is performed.

24:19And so when we say CRUD action, a CRUD action is create. It is... CRUD. What's that? You're getting really techie here. CRUD? Yeah. Read, update, or destroy. It is create, read, update, delete, right? And so these are CRUD actions. And what capabilities do we give to the agent? I think in the early days, the read capability was the thing, right? You can read the data that is in the system, and you can now serve as that orchestrator on top to take the data that lives in this backend repository and the backend repository over here and now view them in one single lens, right? That was like the read action.

25:04And that's, I think, how a lot of WealthTechs started was building basically the chatbots, right? The chatbots that gave you the ability to read data that already existed in the databases. But where we're moving towards and what's actually already here is the other actions. So create, update, and delete. This becomes much more what I would consider to be dangerous if you are trusting a new flash-in-the-pan wealth tech that does not have that institutional-level security protocols or engineering team in place because they're going to be bypassing a lot of those checks and balances where if you are creating a destructive action, aka deleting something, you need to make sure that there's a human in the loop there where it's like, hey, before I delete this thing, like, please confirm this deletion will not only delete this item, it has all these downstream repercussions, right?

26:04And then listing out all those downstream repercussions, like you need that. It's not just saying, hey, are you sure you want to delete this item? It's like, tell me what all the downstream impacts would be of deleting this item, right? Similarly, if you're going to have a agent create a new record. So let's say I chat into the agent and I say, the client moved from Massachusetts to New York, please update in the profile. Okay. Like it can clearly update a field, right? Because basically what's happening is when I say that they moved to Massachusetts, it then takes either MA or Massachusetts. It knows that that, and I can spell it incorrectly, it's fine.

26:45It knows that what I'm trying to do is Massachusetts. It then goes and looks at the database and says, all right, what field do I now need to update? And then it says, all right, I'm going to update this thing, right? So now I'm going to update this to Massachusetts. But again, when I'm making that update, that's going to have serious repercussions downstream because if I'm going from New York to Massachusetts and my analyses from a income tax standpoint, from an estate tax standpoint, we're all based off of New York tax law, right? Now we need to clarify with the advisor, all right, these things are all going to update.

27:18Do we want to update any previously run analyses? And so if I created a report for a client and I have that sitting there ready to deliver, right? And let's say it was a combined estate and tax report that looks at their income tax circumstances, their estate tax implications of living in a state with the New York cliff, right? Do I want to re-trigger that report to now have this new fact, this new input, right, that overrode the last one? And do I then want to recalculate and re-trigger all the downstream operations? That is where we're at right now, Craig, is like this moment where agents can do these things.

28:01But the question is, do you want the agent to run autonomously? And that is, I think, the moment and the question of the moment right now. Do we want to give the agent the ability to actually do the CUD actions autonomously? The read actions probably we're all good with, right? But the other ones, that's, I think, a little bit of a gray area at this point. Great answer. That was a very detailed response to that question. And we're almost out of time. I can only fit in maybe one more question. So you answered a couple of other questions in that as well. So they can do multi-step actions. They can launch agents to handle these tasks.

28:52They can give it different capabilities. Here at Ezra Group, we launched our AI agent directory to start tracking all the, we're over 100 agents already that we're tracking. Plus we have 75 MCP servers, including wealth.com's MCP server that's now online. And we're also coming out of the report next month, this month, on AI agents. And as you mentioned, CRUD, that the agents have to have those, some sort of way to control that governance. What can you do? What power do they have? What can they update or not update? And how much autonomy do they have? So understanding those is important. But I want to just kind of wrap up with where are we going from here?

29:41So where do you see wealth.com, you know, if we had this same podcast next year, where do you think we're going to be? What's going to be the biggest change or the biggest product release you've got scheduled? Yeah, I mean, so I also want to address and I promise I'm going to answer your question. I want to address this question of like, all right, what's better? Do we have one tool that kind of builds all the capabilities, financial planning, tax planning, estate planning, Everything is built in and meeting notes and blah, blah, blah. And everything is stitched in together and you could trigger account openings and wires and ACATs and all of it is all in like one tool.

30:18It's like, is that the best, right? Or is it better to have a orchestration layer and then a bunch of specialized applications underneath it? I think, honestly, we're going to have this conversation in a year's time. A lot of the loud voices in the room that are throwing out like opinions will now be able to root opinions in the broader population will be able to root those opinions in now gradable fact. Like where has the industry actually evolved to over the past 12 months? And my personal take here, Craig, is that capitalism involves specialization for a reason, right? Like there's a reason why communism is not the driving force.

31:06Yeah, all-in-one systems are communism. You don't want those. My point here is specialization is the backbone of American industry. And it's because you can't do everything to 100 % capacity all at the same time. It's just not possible. And so you have to pick and choose and determine what are you going to be the best at? Like you can't be truly the best at everything. And I think for financial advisory firms, like my personal take is you look around and you select the best and then create an orchestration layer on top of that from which you are getting the best to talk to one another. Right. And to have everything working in in orchestration.

31:59Of course, you're a little biased on this, right? I think I'm Orion or Investnet or they would disagree with you. I'm not biased because we could build anything. Truly. Like you think that you think that with all the capital that we've brought in from outside from our investors, we've raised over 100 million dollars. Like you think that we can't go and build everything under the sun? Of course we can. But the question is, how do we deploy that capital? Right. And I also look at like our our large enterprise partners. Right. We work with many of the top 10 private banks in the country, three of the top five private banks, wire houses, that is.

32:38If you were to throw out a bunch of names, we're probably working with them. And they have so much money. They have literally billions of dollars of tech budget, billions of dollars. So the question is, can they build everything? Could they build a wealth.com? Could they build, name your other technology? Yeah, of course they could if that was their core competency, right? If that's where they want to spend their time, of course they could do that. And my argument is, I mean, maybe it's contrary to a lot of other perspectives out there. My argument is just it's actually not worth said large institution's time to have to not only build the original thing, but then continue to maintain and update.

33:28I agree with that. Yeah, I've made that argument in front of the board of a number of large broker dealers. And they asked me what they should build versus buy. They asked me you should build almost nothing because you don't have access. Every one of your vendors has access to hundreds of other firms that they're bringing in best practices from all them, incorporating them into their software. You can build anything. You could always build anything, as you mentioned. It's even cheaper now. Yeah. It's free, but you're not going to have access to the same information and best practices from across the industry that a vendor would have.

34:04Yeah, that's exactly right. That's exactly right. So you asked the question. My answer was a little bit in a different – I took it in a different direction talking about like – I noticed you took it in a different direction. Now, what are we actually going to build? I mentioned before, we are going to continue to lean into our core competency around advanced planning. Now, notice we didn't start out with tax planning. We moved into that space. Right. So we were in estate planning and we do three things in estate planning. We create the actual estate planning documents. We analyze estate planning documents and we help you to optimize a client's estate plan further.

34:42Right. On tax planning, we analyze and we allow you to optimize. Right. So how do we think about continuing to deepen our capabilities in both of those realms, but also considering it as a unified problem? Right. Income tax planning, estate tax planning. These are combined disciplines. Right. Where you're managing against either the step up in basis or you're managing against the gift and estate tax. Right. You have to think about those in tandem. And when you think about it from a infrastructural standpoint behind the scenes, it's much easier to build two point solutions, build tax planning, build estate planning.

35:25But to get these things to fully integrate and talk together is a completely different consideration. And then what are the other tangential things that go into that, right? You think about equity compensation, you think of and how that impacts income tax planning, right? You think about a lot of other stuff, cash flow, cash flow management, for how that relates to a Roth conversion and the lifetime benefit of a Roth conversion, right? There's all these things around that. So you can bet that we're going to continue to invest in the core of what we specialize in today, which is tax planning and estate planning.

36:03But we're just going to continue to make it more robust. And by doing so, we naturally will start to spill into other areas. Danny, you've said it all. And I can go on for another hour or two, I think. I've got another 10 questions I didn't even get to ask, but we are out of time. And I don't even need to tell people where they can find more information about wealth.com. Go to wealth.com. That's right. You can see the greatest CMO of all time, Dan Bolton, at some conference somewhere, I'm sure. And I'm sure you can run to Danny at some conference somewhere. And Danny, thanks so much for being here.

36:40Great to see you. Thanks, Greg.

36:45It's Greg again. Here are my top three takeaways from this episode. Number one, Wealth.com's engine does the math and Cloud reports the results. So of course, Anthropik came to Wealth.com after leaders at wire houses, broker dealers, and large RIAs named it as their estate and tax vendor they would really want to see in the Cloud ecosystem. So Danny said it wasn't pay to play. And Wealth.com's own calculation engine works out every estate distribution. So advisors get repeatable numbers with the full calculation behind them. Number two, advisors are already using AI, so your firm should really give them a safe way to do it.

37:21A Wealth.com in-app survey found that almost or maybe about 95 % of advisors across their platform are already using some sort of AI tool. Now, a firm that refuses enterprise licenses doesn't stop that use. It just pushes client data into personal accounts and may not have the model trading turned off. Number three, agents should show the impact before they change anything. Agents can now create, update, and delete records as well as read them. Denny wants each of those changes to show the advisor the downstream impacts first. His example was a client who moves from New York to Massachusetts, which would mean rerunning every tax and estate report that their advisor had done for them.

38:01All right, that's it. Congratulations. You've reached the end of another episode of the Wealth Tech Today podcast. But before you go, please head to our website, azuregroup.com, and scroll to the bottom of the homepage and sign up for our newsletter. Once a month, you receive an email chock full of both management goodness, news, information, and updates. You won't be disappointed. Thanks again for listening, and talk to you all again next time.

From the publisher

The post The Human in the Loop: AI Agents for Estate and Tax Planning with Danny Lohrfink of Wealth.com appeared first on and was written by Elana Iskowitz.

Meet Our Guest Danny Lohrfink is co-founder and Chief Product Officer of Wealth.com, an advanced planning platform for wealth management institutions that focuses on estate planning and tax planning. His co-founders previously built and sold a machine learning platform to LexisNexis, and AI has been part of how Wealth.com analyzes and optimizes client plans since […]

The post The Human in the Loop: AI Agents for Estate and Tax Planning with Danny Lohrfink of Wealth.com appeared first on and was written by Elana Iskowitz.

More from WealthTech Today

All episodes