Disrupting Legacy CLMs | Alex Zilberman, Chamelio

In this episode of the Legal Intelligence Platform (LIP) series, Anna sits down with the founder of Chamelio, a LIP built for in-house legal teams.

The conversation goes deep into what it takes to build a modern contract review and negotiation platform, why legacy vendors struggle to adapt to AI, and what all of this means for legal teams that are trying to buy AI tools without getting misled by demos alone.

This video was made in partnership with the featured vendor. The vendor provided product access and reviewed the video for factual accuracy. Editorial opinions and conclusions remain those of Legal Benchmarks.

00:00Chameleia was quite bold in terms of taking on all of the existing incumbent CLM from the get- go. What would you say is that unique edge that you have when compared against the CLM giants like ironclad?
Speaker change: They got started on working on the AI the same time that the rest of the world did. Having thousands of customers for some actually stopped them from building a better solution. A legacy incumbent always moves like a big ship and we're a speedboat. So we can make a decision this morning and execute it in two three

00:30days and we do that and we never sleep. We've built a contract review negotiation platform that blows up out of the water anything they try to if I would to to say to my customers whatever they build now it's like buying a new product.
Speaker change: Hi welcome back to the legal intelligence platform series where I dive deep into what legal AI startups are up to today. Before I take you to my home studio where I'll be speaking with Alex Zilberman, the CEO of Chameleio, here's what you need to know about him. Alex is a seasoned operator

01:01who's led teams across product, finance, ops, and go to market. In our conversation today, I'll be speaking to Alex about why he is building a legal AI startup even though he's not a lawyer, where he thinks the future of legal work is going, and why he thinks the current CLM incumbents are lagging behind. Follow me and let's dive in.
Speaker change: Hi, Alex. So happy to have you with me today. I'd love to start with understanding where this name came from and what this platform is about.

01:31Speaker change: Yeah. So, the name Chameleio comes after I would say the second most popular way of naming companies and animal. And of course, it's a take of Chameleon. And very naturally, when we started building our legal intelligence platform, being able to mimic and learn the environment and the way that they operate is going to be key. And the name Million really kind of presented itself.
Speaker change: You are not a lawyer. Is my understanding correct?
Speaker change: That is correct. But we are a group of three co-founders and we have a lawyer amongst us. So one of my co-founder he's

02:02a recovering lawyer of 15 years. He did the three in-house roles when the last one he did he was the general counsel for my previous company. So we connected and we met also our third co-founder Gal who was back then at Sage the accounting giant. He got to stage by being bought up with his startup that did document intelligence. And so a an entrepreneur, a second time entrepreneur, a lawyer, and an applied AI researcher met in order to build Chameleio. And I think that each of us brings something completely different into the game.

02:32Speaker change: When you guys first came together and said, "Wow, Chad GBT is coming out and it's going to change so many professions." I mean, you have background in a wide range of professional services. Why legal in particular? Is it because you were very, very frustrated with the lawyers that you were dealing with? It was very clear to us that we want to touch something that is close to home. I had a very extensive ground in operations already back then due to my previous role. Um and when we started ideating that was one of the three ideas that we had in mind. So there were more than one idea that's not it wasn't a calling at that

03:03point but what we've done which I would recommend any to do as a beginning we did proper validation. So we went ahead and we spoke to more than 100 different professionals in each and every segment that we we were investigating and what was very obvious from the get-go that you could get general counsel chief legal officers to reply on a cold reach out on LinkedIn. So in three months time we're able to speak to 100 of these people of these professionals and after speaking to them we we really understood that they are very frustrated what they currently have they're really anxious

03:33scared worried excited about what the future can bring and these signals were very clear indication that this is a place that we can pursue business opportunity and we started you know pulling that thread and very quickly we decided to go all in
Speaker change: and you just closed a $10 million amp so congratulations on that. What were the other two ideas that you had? One was helping US-based companies manage the process of contracting with the US government. That's tens of billions of dollars opportunity and the other one was account receivables. But they were

04:03left in the editing room uh during the validation phase and was very obvious that this was the route to go. If you think about it, there is a very clear like connecting tissue between these ideas that goes to our founder fit and our clear technological advant advantages that we could bring to the table when building this solution. uh you decided to go all in on legal and I would assume you decided to go all in on in-house solutions.
Speaker change: What problem did you solve as the initial wedge into this market?

04:33Speaker change: The first place that we got started which I'm happy that we did was really contract review and why I'm happy about it because it made us understand that in order to do it better and different and more efficient, we need to understand the contract. I mean there are very clear rules in a in a high percentage of the contracts. There is a predictable structure to them and the appearance or lack of certain clauses indicate certain things. So and that was the first start. We then reuse that engine to do many other things. I mean when we uh and we

05:04started building our smart repository again same engine come to place when you actually want to analyze store and track that data. And we're now moving on and building outside of the contract space by using same principles but applied differently because while contracts are structured in our view other types of legal documents become slightly harder to dissect and analyze the same way but there we are using and relying he more heavily on L versus a more engineered approach like we do with contracts which

05:35is different than what you would get today in the market. When you review a contract, you as a human being, you read all the clauses and in your mind, it's like you're checking off each clause, saying, "Okay, I read that, that's fine. I read that, that's fine. Okay, here's an issue." That's how usually most people review contracts. LLMs, when you take a document and you just put it into an LLM and you prompt it with the questions or a playbook, which is a list of questions that in most systems they use, it doesn't do that. It it doesn't do the same type of process. it will translate the document into its

06:05embeddings into its like mathematical representation whatever and it will an input and will get an output spitting out the result to that prompt I mean LLM are kind of magic no doubt but still it's a very clear process what we do in order to mimic that operations of a human much more and to achieve better results is we first break down the document so the contract isn't ingested as a whole to another model and then outputs an answer that we spit out nicely on the screen we first make sure that we account for each and every part

06:35of that contract. So what we would do that you won't get anywhere else for example as a matter of a method is if clause 3.2 isn't covered by your playbook. Okay, it just doesn't say anything, right? It's silent about it. We won't ignore it. We will still we will be able to know it wasn't accounted for in your playbook and we can do some extra checks on it. And what we will do is we'll run it against our own proprietary algorithm using LLM of course and we'll spit out an answer. Breaking down into the steps requires you understanding what what each clause might be by classifying comparing each

07:06clause to understand is it part of my playbook or not before you even do that. And then when you actually draft the language there's another process that tries to understand do I need to take into account there are two different items in my playbook that might give similar or slightly contradicting recommendations to draft a language to answer these requirements. And all of these and I gave just four. All of these processes are done in that pipeline that then generates these results that people seems to like. So I hope it uncovers a little bit from the that black box we've

07:37built. I definitely can see how much happens under the hood and behind the scenes of not just building a product because everyone can build let's say a feature like contract review but the real difference and the differentiator is in how intuitive it is like you said how proactive it is in identifying what the users want without the user having to explicitly state that and then of course the quality of the outputs as well. So it's very important I guess for legal teams to be aware of that and also to test it out for themselves when they're comparing the different platforms. One thing I wanted to share

08:07which is my observation and it could be completely wrong but I felt like Chamilleia was quite bold in terms of tackling CLM or taking on all of the existing incumbent CLM from the get- go. So tell me a little bit more about that.
Speaker change: Just a year ago there weren't as many customers using our solution and we were just getting started. Again it's always nice to think about from our perspective to see how much has been successfully completed in the last year and it's it's really exciting. So I think that what we understood early on is just building a point solution isn't going to cut it.

08:37And the way out of a point solution of a transactional only result is by building a stateful solution. Stateful solution being the system that has your data that can then be used in a ongoing manner throughout the life cycle of whatever you want to solve. Yes, it's bold. I mean early on we were like he's going to replace their CLM. It's too sticky. It's like they love it. Well, you know, they actually don't. They don't because it's it's a software that was built uh with outdated technology with a lot of

09:08constraints that have now been alleviated in the last couple of years and because they already have a decent amount of customers. It's very hard to change that. You really need to rebuild the whole thing if you want to do it the way that modern software is now being built. So we understood that if we want to build something important and that's going to be really useful and going to be useful for the long term we got to go all in and we got to go for the system of records and that's what we've done and so far it's proven itself as a as as the smart direction as a good idea. I won't be surprised if you'll see as you mentioned more companies now walk

09:39towards that way. I think we have somewhat of a head start and I also think that we've made some design choices on how to tackle what what CLM used to do in a very rigid way in a more original way that can tap into that amazing flexibility that AI can bring today that just wasn't available or or relevant just two three years ago. So I can't blame them for building it the way they did. I would probably do something similar then. And so then what would you say is that unique edge that you have when compared against the CLM giants

10:10like Iron Clyde for example? Why are they no competition for Chameleio today?
Speaker change: Well, first of all, they are. So let's not get ahead of ourselves. They are they're a great company. They they have a a strong solution. I think that it's a it's a classical situation. A legacy incumbent always moves like a big ship and we're a speedboat. So we can make a decision morning and execute it in two three days and we do that and we never sleep literally. I mean we are all in and that's like I think the fourth time I'm saying it because that's really how

10:40we feel. Now what does that mean? It means very directly we've built a contract review negotiation platform that blows up out of the water anything they try to do. They've been trying and not and I'm not referring to Iron Cut specifically. I'm referring to the CLMs out there as a whole. have been trying to add AI for the last 3 years as well and only now you're starting to see the first like things of something that might mean anything and and I understand that because there is a huge change that you need to make in order for that to become really effective and there's a lot of impact on how the software behaves in order to really do something

11:11that's meaningful in the I would say architecture of the software. So the existing incumbents in the CLM space, they have already been the system of record for a lot of these enterprise companies for many many years. So the question then becomes when AI comes along, why can't they just slap on, you know, AI functionality to tap into all of that knowledge base and be like, you know, way better and more accurate and have higher quality results than someone like you that's building your solution that doesn't have as much data points.
Speaker change: CLMs, they have no access to the data,

11:42so let's not kid ourselves. It's not the fact that they were there years ago. It's not like they could access I hope you know but they should not have access to their customers contracts or anything like that. So they got started on working on the AI thing probably the same time that the rest of the world did. So their advantage is their existing customers already customers. These are their advantage. It is an advantage by the way but they did not capitalize on it properly. Not yet because slower company, not not the right talent, not the right architecture. At some point, and I think

12:12that we're at that point now, having these hundreds of of of customers, thousands of customers for some actually stop them from building a better solution because you have backlog and commitments to customers and implementations to do and and a list long of things that you have to do before you have that ability to even add AI properly into your system, into your product. And I mean, they'll get to it at some point. But if I would to to say to my customers like I do and to the rest of the viewers, I mean, you need to

12:42future proof yourself. You need to buy a solution that's been built with the latest and greatest technology and architecture and AI in mind and not something that has been added afterwards. Because to be honest, whatever they build now is built now. It's like buying a completely new product. And then if you're already doing that, you should go to Anna and you should use her benchmarking guidelines and check what's out there. What's the best solution for your needs? Really, I mean, just because you already bought one of the competitor CLMs,

13:13should you stay with it?
Speaker change: Speaking of going shopping right now, how do legal teams find out about Chameleio? Because you were bootstrapped before. You obviously didn't have that many marketing dollars to put on Google ads or LinkedIn ads. And so, how did customers find you? And what were the problems that they initially come to you with?
Speaker change: Yeah, so the number one source for customers for us currently is other customers referrals. That's the number one source and we are lucky enough to have amazing customers that are willing to to recommend to their friends and colleagues in other companies. So that's

13:44the number one source currently. And you are right Anna, we have not spent a lot of dollars on marketing so far. we are now much more out there going to events and sponsoring things and you know trying to gain better brand recognition so more people are aware that we exist. So that's that. But another thing that we've done early on in the few dollars that we decided to invest in marketing is education. So sponsor educational events, webinars, anything in that nature. We wanted to be where curious lawyers are and you know hungry for

14:14innovation, hungry for education, legal teams persist because they are more clearly interested in learning, buying, adopting and moving forward. So that worked also quite well for us and we still continue to host these webinars. In February, we'll have a few in March and in every month we'll have a few webinars about not only about AI but also about like education and how legal benefit of course from AI but also just general education of different legal

14:45sorts because we we think that interacting with the community and the the education part of it is a very strong way to be better recognized. So for the pilots that you do get into with your customers, do they tend to usually be in the category of like CLMs or is it usually in the category of like legal intelligence, contract intelligence or do you see a mix of both?
Speaker change: It's a mix of both. So really it is a mix of both to be honest. It started with the contract intelligence sides of things but over time as we expanded our capabilities and we now compete directly

15:15on the same market also with the CLMs. I would say that 50 to 60% of our pilots today are in that
Speaker change: out of the pilots that you have lost. What were the reasons you found out for losing those pilots?
Speaker change: Yeah. So, first of all, I would uh I can say that we win on average 85% of our pilots. So, I encourage anyone just to try it out. Why not? I mean, it's three pilots. Sign up, see what happens. Let's see. We we did lose a few pilots and we are always looking into that and that's something that we really care about

15:45because I think you have to learn from your customers. That's the whole idea of building a solution. The main I would say the main reason if we lose a pilot would be that a person that tried to onboard a tool didn't necessarily had enough I would say internal buyin to to get that done and then we couldn't get enough people to try using the solution. So it's a very clear I can look at at at at a customer's graph after a week and know if they are con they can if they're going to convert or not just because if

16:15if they log in they literally if they try they're going to buy but some don't want to try. So when you have a team of like 20 people and you have a legal ops person that's like really excited about the solution and he wants to solve this and that but his team maybe even his GC doesn't doesn't necessarily back that effort then it's likely to be a losing situation. Fortunately enough, we didn't come to this situation many times and I can say that even if we identify that that might be the case, we will try to provide with like extra resources and

16:46ideas and try to engage with the people because you know it's it's just about getting started. It's just about trying and and and you know logging in but it it's not doesn't always work.
Speaker change: I think the challenge is that for many of these contract intelligence platforms they could look quite overwhelming for the users when they log in. They don't know what's possible and a lot of education is needed. So what are you doing on that front to help these users get a quick win so they stick around.
Speaker change: Yeah. So that's a great question and I'll say I'll answer it in two in with two points. One, we we don't we won't

17:18work with you unless you are unless you understand what the pilot looks like. So unless you understand that there is a session of training which is mandatory. If you want a user that's going to log into the pilot, it's mandatory. You're not on the session, you don't get a user. Simple as that. It's it's already proven in terms of if you don't attend the session, first of all, the likelihood that you're actually interested in trying is low. So, we want to force just a tiny bit the situation. The second thing that we do, we insist on having a Slack or Teams channel with 100% of our customers and it starts in

17:48the pilot period and then transitions automatically to our the way that we support. We also insist on that. Sometimes they they go like, well, it's going to take me time for it to approve the channel and stuff like that. I said, we'll wait. That's fine. because we know that being there for our customers is a very big and important part of success for for any solution but especially with something new. So we insist on having a clear and direct communication channel with the actual end users. They don't need to open a ticket. They just put something on Slack. They'll get an

18:18answer in no time. And once others see what was the question, what was the answer, how responsive we are, it creates a great of interaction.
Speaker change: So part of what we're doing here at Legal Benchmarks is to try and make this procurement process easier for both the vendors and the legal teams who are looking to select the best tool that fits their needs. I wonder what do you think are some of the misconceptions or the factors that legal teams are not taking into consideration enough when they're looking at the different options that would make their life a whole lot easier if they knew in advance and also maybe make your life easier as well. I

18:49would one thing that comes to mind that is I would say particularly repetitive is implementation complexity timeline and the process of implementing the solution. I think this is either completely overlooked in terms of what's going to happen after we say this tool. It's being treated as something that was probably in the past like a super traumatic event in a way that they're like no no no no it's going to be that's going to be 2 years again we can't have that we'll stick to what we have. I think by the way we heard that so much

19:20during the validation and it still goes on just made us even more excited and that we even build part of our solution in a smart way to avoid a lot of the the pain point in implementation and and our onboarding is like I would say probably the shortest in the market period full stop. I mean it's one of the values of getting Camille because we have a lot of AI and automation goes into the onboarding and it's definitely not being uh I would say um paraded around because it seems like okay whatever it's like when we decide to buy it but I have to I have to say I think you should probably

19:51ask it in the middle you know it's like when you shortlist your solutions and we are commonly being shortlisted with at least one or maybe sometimes two other solutions before we go into bigger projects I then say explain that like proactively if I'm not asked because I think that understanding what implementation looks like and how painful it could be or not.
Speaker change: So about this implementation timeline and setting expectations for it. Most vendors on their website all advertise to say implementation is easy. It will just like one click or like we'll handle

20:22everything for you. So what would you say is a realistic expectation let's say for a legal team that already has existing legacy CLA provider for them to migrate to a new um LIIP's CI CLA evolution what are the realistic expectations they should have or the timeline or you know the factors they need to consider so they can be smarter about this
Speaker change: so that's a great question I think that if that's I'll start from the end of your question I think that's like a going to be most helpful for them so question that they should ask actively is are you going to dduplicate automatically my documents ments. Are

20:52you going to automatically find all the drafts and all the documents that aren't really contracts and the stuff that are lying in there either in my current CLM setup or in the drive or or shareepoint folders? Are you going to make sure that it's going to analyze it according to the relevant document type that I have? All of these things in many many setups are being done or not done but manually nonetheless. So still and then it drives multiple issues. One, time of course. Two, professional services costs,

21:22unpredictable costs in terms of what would it actually means and also it requires a lot of effort from the side which is hard to predict if you'll have the time or not because you will be implementing the solution during not your summer break but during your regular working hours and days and there will be the end of the month and the end of the quarter days and you won't be able to progress it. You'll need to you need to make sure what the system you are enlisting to what it provides from a technological perspective of the bat. So you don't sign up to something that just

21:52means relying on human resources to be able to complete that because this is always a question mark and of course more cost associated that's definitely one thing that I would I would d into deeper and understand better. Now in terms of like expectations if you have already a current CLM setup and it it actually alleviates one of the biggest time constraints that we see in projects which is legal teams and legal ops deciding what they want to do and how they want to correspond and interact

22:22with other departments in the organization. You'd be surprised to know that when presented with unlimited capabilities theoretically they then start think okay maybe we should do this and then they change their mind and and so on and so forth. I think that being able to uh start from an an existing system realistically speaking I think you should expect a couple of weeks four weeks time implement a CLM CLM like migration that has like I don't know like 30 40,000 documents with like 15 20 workflows it should take about a

22:52week four weeks or so I need to get that time that cost is also one of the factors of this whole process that maybe legal teams don't fully understand upfront can you share a little bit about whether for a lot of these migrations they are a separate cost a separate bucket of cost that legal teams should account for or is that something that some of these legal intelligence platforms I mean I'm wondering even for Chamilleio will cover the cost of that migration how should they account for the the work and the cost associated with it
Speaker change: yeah so the it varies between companies

23:23most that have heard of that we've been against they charge separately for the implementation and on boarding and all of this dduplication process that I explained they charge completely separately Some they even send you to an implementation company to do the implementation for it for them for you. Not sure who for who but uh I think that this leaves a very interesting gap out there. I think that these consultancy agencies external are a very great way to help legal teams understand what they want to get done and how and think about it professionally. That's great and

23:53there's some great ones out there. But you need to have as much control as possible as a legal team to what's actually going to happen during this implementation process because you want to be able to maintenance and maintain it yourself later. So the legal ops person the system should be simple enough that after short training they could replicate or build out their workflows by themselves. We really encourage that. So we'll do it with them for example because we want them to learn and to get a handle of how to manage and use that platform. our

24:24platform. So that's that's that. Now yes, we do include onboarding and implementation as part of our cost of the platform. There are no hidden fees. You just you pay for the solution and you get it including implementation primarily because we've invested a lot of time in building our product in a way that this on boarding isn't a big deal for us. We understand what the thing that I mentioned about the duplications and the drafts. Everybody has the same problem really. Now once we saw that we just spawn off a small team that built an onboarding engine that takes care of these things. So we are less worried

24:55about god what's going to be the case with that customer or this customer. No, we understand that the vast majority of the challenges should be solved in a repetitable manner and the rest of the manual stuff we will assist the teams in solving and we also have a team of legal engineers that we bring on board and you get a dedicated named legal engineer to you know to work with you on the project and to care of the things that has to be taken care of many not accountants. This sounds like a very hands-on process cuz you're like you're really getting into the nitty-gritty of helping these

25:25companies not only with their contract solutions but also with their data management and data migration. And I wonder if that's something that Claude Co-work or like all of these general purpose chat bots or these Frontier model labs would be willing to do because right now there is a lot of skepticism towards legal AI companies after Claude all the other Frontier um Frontier model labs chatbots release legal related functions and one of the challenges is that legal AI companies are just a bunch of features built on

25:55top of these models. So I mean first of all people always like to romanticize how future how simple the future might look like. If we think about our conversation today the amount of time I've spoken about the pipeline and the engineering side of things and on boarding and understanding the real pain points. I think that simplifying legal department work by saying that okay we got contract review done and thinking that this is what they do period full stop or that it's done is just funny in

26:27a way I mean it's really looking from the outside in on a problem I think it's like any other person looking at at a company and saying what's the problem docu sign e signature solve problem I can build it in two days I can vibe code it code it overnight on and done. I think I saw like 50 posts like that. Well, you can. There are so many things that goes into building a company, porting, understanding their needs, making it enterprisegrade in a way that real users can use on a daily basis that

26:59that having one specific function done in this way or another, it's really not going to cut it. And it also requires so much attention to details that I don't think that general purpose AI or an LLM frontier model company can or should do because they are building the raw capability and around this raw capability you can build many solutions and these solutions are first and foremost are tools that going to be used by professionals that going to be

27:29integrated into their daily work into their daily life and that go and that's that's transcend and extends so much out of a specific task and it raises an awareness. But I don't really see lawyers building their own CLMs soon.
Speaker change: Would you recommend your kids to go to law school today?
Speaker change: Would I recommend my kids to go to law school? Yes, only if they are going also to take a science. if it's going to be pure law school uh not I think that the

28:00ability to think uh logically in and when I say logically I mean from an algorithm algorithmic standpoint becomes a a a stakes need and requirement from every worker it's just not enough to be your a subject matter expert in a specific topic that you are amazing at you also really have to understand better how computer systems operate and you need to do that because you will be operating a computer system that's going

28:30to be infused with your knowledge and experience and you will be judging its results and I think that as time goes by yeah people will probably move from their current roles to other roles and it's really hard to know the very far future looks like it's there's a big
Speaker change: you think it's possible that people will move from legal roles potentially not into other aspects of legal roles but into non-legal roles as these this technology is changing and replacing some of the work.
Speaker change: I think I think that the first um I would say move that I I would witness

29:01and I kind of see it starting now already is more people would move from being just lawyers to being legal. Um I I see it already. I mean more lawyers become legal ops versus that was not the case a couple of years ago. You would you won't have lawyers become legal ops. You see it now. I think by the way that this whole concept of like AI ops in in a department it's it's very interesting. Is it going to be like per department? It's going to be wide, companywide. I'm I'm not sure. I'm kind of more inclined to think it's going to be per department. So, there will be like an

29:32expert on your department's AI. They tend to think it's going to be the subject matter expert that were was able to adopt to this new reality and now he's the herd masters of the AI agents and he understands how to control them, manage them and and what they do and how they interact with other AI agents. But it might all go wrong and we go back to the stone ages. So
Speaker change: what's something on the road map for Chamilleio in 2026 that will get legal teams really really excited about trying it and make the other legal intelligence

30:04platforms really scared and take you take Chameleio even more seriously.
Speaker change: Okay. So I think that there is one thing that's coming very soon. So, and then it's a new legal AI memory layer that we've built that allows us to really capture and learn the essence of your legal preferences from every action that you do across the platform. So, when you going to accept or not a certain redline recommendation that we give you in Word or you're going to change it, we're going to learn from it. And when you're going to review a URL of a terms and

30:35conditions in our web interface and how you're going to interact with it, we're going to learn from it. when you're going to look for certain things throughout your contracts in the smart repository, you're going to learn from it. When you interact with our workflow and our slack bot agent, we're going to learn from from it. What we see in early bet testers is that this interactive learning creates an amazing experience and I'll give you one example citing without mentioning the customer name is that uh our customers she went on and written good morning to the agent just random good morning and the agent with this memory function enabled replied the

31:05following answer. They said, "Good morning person. How was the call with Amazon yesterday? Did you end up agreeing on that you were debating today? And would you like us to try and pick up the work that you have not completed on the contract you started reviewing yesterday at midnight?" Now, this is amazing stuff. So, fast forward a couple of months, maybe you'll be getting these good mornings proactively from saying, "Hey, Anna, I know you didn't finish that thing. Would you like us to continue together?" Think about it. And that's pretty cool. And that's actually already in the works.

31:35Speaker change: Thank you for giving me a sneak peek, Alex, and also the other lawyers out there of what is possible in the near future and also in the far away future. And I just want to wrap up with asking you this question, which I ask every single founder I speak to. What are you using behind the scenes to run and build this legal AI empire of yours?
Speaker change: I use Claude, I use Chad Gupt, and I use Gemini as well. every day. I kind of nail down each of their secret extra good capabilities and uh and and I

32:05really use them in a very intertwined way to to get everything that I need done from analysis of Excel files through drafting letters and proposals. So, I use it like every single day almost on every task that I do. It's it's a second nature. It allows me to do what I feel like at least three full-time roles instead of just the one.
Speaker change: Out of the three chat bots, which one is your favorite right now? I know this could shift like with any moment with any new model release, but what's your favorite one?
Speaker change: It's I don't know. I mean, they would be offended if I say this or another. I

32:36mean, the chat bots, not on the companies. So, um if you really uh my arm, I'd say I'd say that Chad GPT I really like and I learn a lot. Um really like because they add many um dual components to their outputs. So, I like it from a product perspective, you know. So necessarily the results are much better than the rest but I like how easy the interface is and I learn a lot from that and gives me a lot of inspires a lot of the things that we also add to our uh results and so that's definitely one of them but I I really use all of

33:06them like every single day. It's like your favorite child. You can't like they all have their quirks, but you can't live without any of them.
Speaker change: You know, and it's it's weird, but it's not not far from that because each really does something that for me that I can't get the same level with others. By the way, that's why we use all three providers in our pipeline. You can't really get an amazing result. Just use
Speaker change: Thank you so much, Alex, for having this chat, sharing the vision, sharing what you're building and the challenges that come with it.
Speaker change: Thank you for the time, Anna. Thank you

33:36for your question. It was really