Aparti is building an AI-powered operating system for family law that automates divorce processes. The founders include Anna Naidis (CPO, 17+ years in tech/product) and Igor Sheremet (CEO, 17+ years in entrepreneurship).

The Problem:
Nearly 1 million US couples divorce annually, spending $15 billion collectively. The process is financially devastating and operationally outdated, with lawyers drowning in non-billable administrative work.

The Solution:
Aparti’s platform has three main components:

1. Conversational AI for client intake and data collection
2. Asset division engine for predicting divorce outcomes
   Automated document generation with e-filing capability

Business Model:

• B2B focused, serving law firms (not direct-to-consumer)
• $500 per lawyer seat (covers 10 cases/month)
• $100 per additional case
• Target market: ~200,000 family lawyers in the US

Current Status:

• Working with 10 law firms, primarily in California, one in Oklahoma
• Training models on 140,000+ open California divorce cases plus proprietary firm data
• Integrating with case management systems like Clio and MyCase
• Backed by Antler and angel investors

Key Value Proposition:
Law firms can unlock $140K+ in new annual revenue by bringing financial analysis work in-house (previously outsourced to accountants) and reducing non-billable client communication time.

Challenges Discussed:

• Unauthorized practice of law concerns
• Liability and AI hallucination risks
• Scaling across different states with varying laws
• Determining whether paralegals or lawyers will be primary users

The founders are seeking connections with open-minded family lawyers interested in prediction models and AI in legal practice.

Watch CodeX group meeting

Transcript

Roland Vogl – Executive Director CodeX

It’s 1:30 pm January 8th, 2026. This is our first CodeX group meeting. Welcome to everyone. We’re excited to get our group meetings going again. This is always such a fun group where we bring in early-stage legal tech entrepreneurs or researchers to share what they’re working on and have a little bit of discussion around that.

Today we have our first presenters of this year. We have Anna Naidis and Igor Sheremet, co-founders of Aparti, which is a new AI-based technology for family law. We’re excited to learn more about it from you.

With that welcome Anna and Igor. I’ll turn it over to you. 

Anna Naidis

Thanks for letting us join in your first session of the year. It’s an honor for us. Let me talk a little bit about myself and about our co-founders. Igor will share more about our company. My name is Anna, and I have over 17 years of experience in technology. My main specialization is product management, strategy, and operations. I used to work for Fortune 500 companies like DHLThomson Reuters and then top YC startups in Silicon Valley, and recently in one of the AI startups. I am chief chief product officer, actually. 

We have another co-founder, Ingrid. She’s our chief technology officer, who also has over 12 years of experience in technology. She was one of the first Uber engineers and she was one of the Netflix engineers and Chime. We used to work with her together in a YC startup. She was my engineer. I was her product manager. That was the time when we became also friends and I learned that Ingrid started going through a divorce. That was around five years ago. Around three years ago, I met our third co-founder, Igor, our CEO.

Igor Sheremet

That’s me. I have more than 17 years of entrepreneurship. And then, first of all, thank you so much for having us. We’re really excited. I have more than 17 years of entrepreneurship and international business development experience. I built two companies before, sold one of the companies. Closed companies that, unfortunately, didn’t work out. I’m also divorced. And I’m CEO of Aparti, where we believe divorce is not the end. We believe it’s a fresh start. 

Let me tell you, divorce sucks. It’s heartbreaking, financially devastating, operationally extremely arcane. Nearly 1 million couples in the United States are going through this process every single year. They’re spending $15 billion and divorces cost the U.S. economy over $300 billion in lost productivity. I’s also very painful for lawyers. They’re drowning in the admin work they cannot bill for. We got together and decided that we’re going to change it. But we’re not trying to fix what’s not working in this process. We’re actually aiming to rebuild it from the ground up.

Aparti is building the operating system for family law. We’re building an AI platform that automates the most painful parts of the process. Here’s how it works. First, intake. Our conversational AI guides clients of law firms through the intake process, collecting all financial and personal data. Second is asset division. Our asset division engine that we’re building helps to provide predictability on their potential outcomes of the divorce. Law firms have the ability to predict what the revenue is going to be. Families are able to predict what the potential fair split of assets will be. The third part is we generate ready documents with direct e-filing.

The value prop for family law firms is pretty powerful. We calculated with our design partners that with our solution, they unlock over $140,000 in new annual revenue by automating work that they can bill for. We also bring a lot of financial analysis work that’s usually been outsourced to third parties back in-house to the law firm. For example, our co-founder Ingrid’s divorce, she already spent over $500,000 on her divorce. Around $80,000 of that money went to third-party accountants. This is what we’re building will allow law firms to bring this back. So Ingrid could have spent, let’s say, $20,000 instead of $80,000.

We’re right now working with open California data,over 140,000 divorce cases. Plus, we use the information from data from our design partner firms. The way we’re building our solution is that every case that’s going through our system, every case that a law firm uses Aparti for makes our model better and improves it.

Pretty straightforward business model. We’re charging $500 per seat for a lawyer, basically allowing them to manage ten cases through the system. Every extra case is an additional $100 for us. There are approximately 200,000 lawyers in the United States. So it’s a pretty big market for us.

In terms of competitive landscape, there are a lot of solutions that are focusing on the user retail side, like Hello Divorce, Divorce.com. There are solutions that are working with consumers. We’re strictly focusing on B2B. On the B2B side, there are other solutions like case management systems—like Clio, MyCase, FileVine. We look at them as partners. We’re actually in conversation with Clio right now to build a partnership.

We’re working with ten law firms right now, in conversations with the largest firms in California and Oklahoma to utilize our solution. We have pretty good backing by Antler SVC, like fantastic angels. Yeah. And that’s where we are right now.

Roland Vogl

Great. Do you think you can show us a little bit of how the system works and do a little bit of a demo?

Anna Naidis

I will need a couple of seconds to be honest, to set up. I was not prepared for the demo. And just because it’s also will be translated on YouTube, I will kind of be careful about this, but we can set up a demo for everybody who wants it. Just because as a startup, you need to be careful of what’s going on and which technology you are exposing.

Roland Vogl

Let’s open it up to see if folks have any questions. I’m curious as to what is the general workflow here? You’re not bringing in customers to the firm, right? The firm has this sort of as a back-office technology, in a sense. To help with what exactly? I know that’s like some separation of assets and those kinds of calculations, but is it helping with negotiations? Is it helping with—what exactly does the system do?

Anna Naidis

Let’s just split the question into two parts. Yes, we’re working with B2B, and we’re helping law firms to streamline the workflows. But we also understand that there are a lot of people who are going through divorce, and we’re helping people who come to us to also connect them with the trusted law firms we are working with. It’s like our great ability to connect the dots.

Igor Sheremet

Yeah. To be fairly honest, we started also as a DIY solution for consumers. We went crazy viral on Twitter. We had like 400,000 views in 48 hours as a divorce-as-a-service solution. We have some thoughts on the way how we can channel clients to our clients, right, to refer clients to the law firms, because we already have a pretty good inbound of regular people coming in.

There’s a lot of back-office. So right now, like the biggest firms, what they’re facing is they’re not able to properly process the inbound that they have of clients. They don’t know how to identify and qualify those leads. That’s part of the prediction model that we’re building. Based on the intake information, we have the ability to provide them with this information. I hope nobody’s familiar with this process. But if you’re familiar with this process, there’s a lot of anxiety and going back and forth with the clients. So we’re building—we’re the first solution that actually has a portal for the client to engage in conversational AI to not take time from lawyers, because what we learned from them is that they want to focus on lawyering rather than just endless.

Roland Vogl

Bea is asking whether you had—have you faced any unauthorized practice of law issues thus far. Could law firms kind of replicate what you do using LLMs and what really is your kind of secret sauce that makes this an investable startup? Is it a divorce law firm, operating system only or what is it? David also would like to know more about whether you had any thoughts on unauthorized practice of law issues and what’s the greatest liability risk that you see?

Anna Naidis

Yeah. Let’s start from, like, reverse, about the liability. We see that—and this is a great question. Thank you so much. Great questions. First of all, there’s of course the usage of LLMs. Always for the law, it’s the question of hallucination and accuracy of the suggestions. We are working very carefully with the data. Family law firms already have digitized the cases they have to understand the historical outcomes of the cases. Our predictions, our recommendations—like how to process the case, what could be a possible outcome, what could be a possible length. And for the law firm, what could be the revenue model.

Can they do this themselves? If they have a big technological team, probably they can start. But the beauty of being technologically advanced and having great experience in the past is that the law firm continues doing work with their clients. We’re helping streamline, as Igor said, we’re helping streamline the processes. We’re helping that they can take more cases on board because people are a bottleneck in any workflow for administrative work.

About unauthorized law practice—I think this is something we will learn down the way together with the law firms. There’s no, like, a strict answer. We’re not giving any legal advice or something like that. We’re literally building the system that allows a law firm to operate in a more—in a better way. Yeah. And we’re working with them to understand what law firms need. It’s not coming from the idea of let’s build this without talking with the customers. 

Roland Vogl

Legal tech companies frequently sort of draw the line between the business of law side of things and the sort of the practice of law side. Right. The practice of law is where you’re dispensing legal advice and doing legal reasoning, and the business of law is like all the sort of optimizing the operations and the workflows of a law firm business. And it feels a little bit like you’re a little bit doing both. I guess if you’re helping to learn from the past kind of cases from a law firm, then the idea would be that looking through these documents or whatever, divorce settlement documents and negotiation documents, then you’re trying to train your AI to sort of replicate a little bit what lawyers do. Right? 

You still have a human lawyer presumably providing oversight and debugging and doing quality assurance. Then so that the end product will will be reviewed by a human. And so maybe that’s kind of less of a risk of unauthorized practice of law. But there’s still the issue that if your document that you’re generating or something that is meant to reflect what a lawyer does and the lawyer just doesn’t do proper quality assurance, then it’s a liability for the law firm. You have to, like have some kind of kind of way to think about that, that risk. 

Igor Sheremet

Yeah, right now I think the way we’ll look at this, we are providing this information to the law firm and their dashboards. Right. What they’re using. It’s as you mentioned, it’s for them mostly. The documents that we’re generating, these are standard forms that are utilized, information that we collect from the intake. And it goes also always under review of the end user, who is the law firm.

That’s a great question. In terms of, like, best—of what makes a product investable, I think what makes a product investable is scalability of this. And as I mentioned before, there’s, like, there’s a huge market. We’re members of San Francisco Bar Association. We are actively participating in events and organization of the American Academy of Matrimonial Lawyers. So we’re actively working with them. We see how this specific product built for this pretty big niche is investable and scalable. 

Roland Vogl

You’re piloting it at the moment with some of these early adopter law firms. And what is it that that turns out to be easier than you anticipated, or what turns out to be harder than you anticipated in this process?

Anna Naidis

Yeah. So that’s also a good one. We are piloting right now with a few solo practitioners, and we’re collecting feedback on the way how they’re operating. I think what’s interesting is, as you mentioned, working with the more enterprise-size law firms because they have a different way of setting up their workflows. They have more people involved. They usually have the paralegals. They usually have lawyers who are doing the intake process, lawyers who are doing different setups. 

I think what also a few other colleagues mentioned—the way how the lawyer will be engaging with the software—is it a paralegal engaging with the software or is it a lawyer engaging with the software? I think there will be a lot of understanding of the features which are scalable because we want to build the product which will be used by multiple law firms. We’re starting from California, and we’re planning to be multistate compliant. I think that’s another interesting challenge—how to cover law in the different states and in different counties.

Roland Vogl

Got it. So B is also asking about how do you scale it. It’s more about replicable workflows. Or do you also predict the outcomes, like do you make recommendations around kind of asset divisions, and—yeah, and I also ask, why Oklahoma? 

Igor Sheremet

It naturally happened. We met at the Legal Tech Summit. Same way I met with the largest similar LLM firm in Oklahoma at the same event. That was a pure accident, but they’re very tech-savvy and advanced, and they’re looking into a different type, not built into like billable hours. They’re looking to provide this service on a flat-fee basis. They’re looking into a more efficient way to do that. They have their data set. So that’s why we have this one firm in Oklahoma. Everybody else is in California. That’s where our major focus is. It’s just an interesting use case. 

In terms of workflows, yes, there’s a part with workflows. If you dive deeper into family law, it’s crazy how outdated everything there is. Literally solutions that right now—some of the solutions on the market, they look like Windows 98 and they’re repeatable—as you mentioned, they’re repeatable across the different states. And that’s for prediction of the length of the case, there could be some tweaks based on the state and county. 

Answering the question about the state—we’re starting with California, but we already know how to do this in a multistate-compliant way. But step by step. First thing is California. 

There are also some technological back-end secrets,how to get, like, updated information about the different states.

Okay. Good question about the PAI,I will also cover this. Yes. We are working with two types of data, as you mentioned. One is available data. There are about 140,000 open cases, California cases, which come with the metadata which you can use and train based on the information there—like, what are the events, how long those events are taking. Then the second one is proprietary data from the law firms. That one, we are using only redacted information. We don’t need the name, gender, or this kind of information—anything like personal identification. The data do not need to build good outcomes. This is also one of our discussions with a law firm—that it should be redacted when they’re passing it to us.

Roland Vogl

Okay. So David made a quick comment. He said he recommended to look at Professor Anthony Niblett’s research on the difficulties of using AI for prediction. And some useful insights there.

I want to go back to something you said before, Igor, that using your system, law firms have discovered ways to monetize things that they normally cannot monetize. Can you just kind of elaborate on that a little bit?

Igor Sheremet

It’s a little—in a way. So the way it works, there are a few ways. Right. The way they can monetize is there’s a lot of financial analysis involved in, although, more complex cases with asset division. So a lot of this work is usually being outsourced. And with our solution, we allow this to be brought back in-house. So all the analysis of our financial—financial situation—okay, this is normally to handle something, but it’s usually handled by third-party accountants. Right now you can do this in-house and basically help families to save money and get extra income for them. But also, the work that they cannot bill for is, like, this ton of small interactions back and forth with clients, which we almost fully eliminate with our conversational part.

Roland Vogl

The next question is what integrations does the platform currently support? 

Igor Sheremet

We are working with a few companies right now to integrate. That will help us foster adoption with the clients, with the medium to enterprise-size law firms. It’s Clio and MyCase, which means we can offer it to the law firms who are already using those systems. The data stored in those systems could also be read by our system. They don’t need to integrate anything, transfer anything. It will be much easier. And then we can do some customization of the analytics and information inside those systems.

Roland Vogl

Do you expect them to just move everything into your system, or is it—are you tying into Clio and you’re providing your functionality on top of, like, their Clio infrastructure?

Igor Sheremet

We will work with them, with the companies right now, and with the law firm right now who is using those systems. Then we will see how, like, how the adoption is going and what they’re looking for, because I think some law firms are—they’re looking for automation. They’re looking more for the white-glove experience. It’s all down the road.

Roland Vogl

Okay. There are a few—there are questions, yeah, on motion drafting. If you support motion drafting, and it’s a—maybe because divorce could go to court or be handled outside of courts. Does that, like—are you agnostic to that, or are you, like, specializing in any aspect of that?

Igor Sheremet

So, about drafting our approach here—we’re not going to do the drafting. There are 10,000 startups right now using AI to do drafting. And it’s just not our business. Maybe we’ll use it in the future—one of the solutions that wins the market. But I think it’s not—no need to compete with something that ChatGPT already can do, basically. And a lot of lawyers use, like, not officially, but are using it and then just rereading and reviewing it.

Roland Vogl

Okay. And then maybe the last question about the business model what’s your pricing strategy? You talked a little bit about it before. Maybe you want to elaborate? You’re offering it to B2C and B2B, but it’s really only B2B at this point, right?

Igor Sheremet

Correct. We—while our initial idea was B2C, that’s how we went viral. But then we fully focused on B2B. And right now, our ICP for a while was small and medium practices, where our plan was to charge, like, $500 per ten cases that they’re managing every month through our system. And every extra case, an additional hundred dollars. And that’s the initial one. And now with enterprise clients, we’re in conversations with these large firms to define what’s going to be the best structure for them in terms of their processing thousands of cases. Right. And they’ve been processing—I mean, they have, like, 1,000 cases every month that they’re working on. And it’s a long process.

Roland Vogl

What’s the biggest need you have right now? And how could we potentially help you along the way as a community?

Igor Sheremet

Yeah, as we discussed, first of all, we’re super thankful for the questions. It’s a great opportunity. We’re super excited to dive deeper and see how we can work together, because our goals are going, like, beyond just simply doing some AI in family law. We think there’s an opportunity for us to work together. The main needs right now—we’re looking for great, open-minded family lawyers who are interested in this. And we’re happy to talk about building prediction models in the law.

Roland Vogl

Thank you so much for joining us and for sharing Aparti’s story. Please keep us posted. Good luck. Hopefully some of the folks in this group will reach out and help you along the way.