The Finance Startup Bringing Agentic AI to Wall Street

0:00In the last seven days, we've signed the same number of contracts as we signed in the whole Q4. There is clear, tangible value being driven by these products, and it's only going to get better and quickly. Ultimately, you've got to be very passionate about what you're building. You've got to have that perseverance. And if that sounds good to you, then build a startup. If something logically makes sense, you should probably continue doing that thing, right? And not let anything stop you. And I think like the consistency that we've noticed of founders that we've invested in or work with is like the ones that kind of do that and really persevere tend to win.

0:39Today we're here with Arne and Chas Englander. They are the founders of Model ML from Winter24. Prior to Model ML, they started two other YC companies that both were successful and sold, Fancy and Fat Llama. And this is probably the first time I've worked with a company where both of the founders had had a previous successful YC company before. So I'm super excited to welcome Chas and Arne here to YC. Welcome back. Thanks for having us. Thanks for having us. Tell us what you guys are building. So Model ML is an AI workspace for financial services. So that's our one-liner. What that actually means in practice is we've built a workspace that's akin to kind of the office suite.

1:17So our own version of Word, PowerPoint, and Excel, with the major difference that it's built on top of an agentic system that kind of mirrors what a human has access to at the firms we work with. So quite specifically, if you're a human at Firm X, right, you will have access to your files and folder systems, your emails, your CRM, any data vendors that you might use and pay for, real-time publicly available information, public filings, your internal custom data sets, etc. So then we kind of build this, we call it a cognitive architecture. It's a fancy word of saying, kind of like a brain that mimics what you have access to digitally.

1:57And we overlay that with our user interface. The general idea being, well, if you had an Excel spreadsheet that was already connected into those data sources, you'd probably spend less time going and gathering information and analyzing it. I can tell that you guys are excited about how things are going right now. Would you put some words on how things are going? Vertical. look i mean in the last seven days we've signed the same number of contracts as we signed in the whole of q4 wow congratulations thanks very much and i think it's it's really just the turning point i think in the sector whereas as we keep saying it's like there is clear um tangible value being driven by these products and it's only going to get better and quickly what were people they're using model ml using before what were their tools that they were using in the daily work?

2:44So they would have their data sets and then they would spend a lot of time in the office suite or in Outlook, right? And what that meant was a lot of that process is super manual and super repetitive. I think the key here is we're definitely not saying that humans should never do these tasks, but if you're organizing logos in a PowerPoint presentation and you've done that hundreds of times before, and you're a very well-educated analyst or associate, it's probably not a great use of your time. Right. And I remember from, I think maybe from our interview or sometime in the first office hour, you had a unique story on why you wanted to solve this problem.

3:23Remember what that was? Yeah. So we sold our first two companies. And after we sold our second company, we did a bunch of kind of investing ourselves, which I think on the whole, we were pretty bad at. But we became very interested in automating as much of those processes as you possibly could right and so it started whereby we literally were when we would receive an opportunity via email and having sold a company you kind of tend to receive a bunch of opportunities every day it would then enter into this like a gensic system that to be honest we were just building for fun and kind of produced like a one pager for us but the interesting thing about the one pager is whilst you know we receive some information via email, what was in the one page, probably 90 % of that information was unrelated to what we would have received via email.

4:12In other words, let's say we got an opportunity to invest in a startup, right? This thing would go off and look at their LinkedIn and their background and then look at like comparable companies on Crunchbase or S &P and so on and kind of produce this one pager that you would probably go and do as a human as your first port of call. You know, little things like if it was a consumer company, it would go and look at review websites, just as an example. And yeah, we thought it was pretty cool. And then people became interested in it. And then we kind of agreed that we were quite bad investors and okay at building stuff.

4:44And that's how it started. And that then turned into you selling this product or a new version of that product to the top financial firms in the world, basically. Yeah. So, you know, we can say now that we're about 10 % of the largest private equity firms and investment banks in the world use our product. We have other customers, asset managers, sovereign wealth funds, a bunch of venture firms. But yeah, it's been pretty exciting. And how would they do their work before you start using Model ML? So as crazy as it sounds now, right? So let me use an example whereby, let's say you're tracking a public company, right?

5:24and every quarter, every time there's a release, maybe as part of your job as an analyst or associate, you'll go to the release, the filings, and you'll put together what's called an earnings summary. So think about it like a beautifully present, like a slide, effectively, a single slide with a bunch of information. But all that information is coming from the same piece of information. So for example, you might always get certain numbers from the filing itself. You might get other numbers like consensus or similar from a fact set and so on and they're put into these slides. and these things take ages we're talking probably like days to make the company just to pull together these slides right because everything has to be linked back to source and checked and so on and what you can do in model m hours so think of model m hours you know being used as the building blocks for this so you have an excel spreadsheet that's already connected into these data sources and you might want to export that into designs right and so rather than having to go and gather that information once that release happens this one pager effectively was actually three pages, a cover page, your main page, and a legal page, just appears in your SharePoint or Google Drive, right?

6:29And it's kind of 90, 95 % of the way there. And we think in some cases more accurate because actually trawling through these filings as a human, because you can also pull from multiple data sets for the same figure, yeah, it can be more accurate. I'm assuming that you built something in the beginning and there was a model that could do some amount of this work. because you did YC how long ago? A year ago. A year ago. So the models have probably progressed a lot in that year. What were things the models could do a year ago that were impressive and what can they do today and what are they going to be able to do in the future?

7:01The whole industry or the industry we're in right now, really from January is going straight line up. And I think the main difference that we've seen this year versus last year is last year was a year of testing. So everyone wanted to make sure they had some sort of exposure, as in they were trialing something, or looking into something, but it was still testing, right? And us having ran consumer companies, we don't really care about revenue figures. We've always been obsessed with like usage and retention. Right, right. That's good. That's good. That's the most valuable thing you just want to bring into B2B.

7:35Like B2B people don't always know the most important stuff. Which we think is madness. But this year, everyone kind of, the whole world went from testing to using. So there was, and there was a fundamental shift. I think these agentic systems, There were small things like the improvement and things like function calling with some of the newer models, but everything has become considerably better by itself. And that's the interesting thing, right? It's like, I think even if we do nothing, theoretically, our product will improve, which is a pretty interesting world to be in. Right. And so when you're constantly working on something and you're pushing the boundaries, you know, month to month, what's possible is like just wasn't possible the previous month.

8:12I also think the vision models. Yeah. I mean, when they first started coming out and improving, we were sitting next to each other. We were going crazy, weren't we? It was like the stuff that we could do with the vision models, if you think analyzing files, as an example, you know, OCR, okay, was amazing. You combine that with vision and its ability to read information from tables and charts, it really just changed the game. Yeah. And also, I think it's a misconception, right? So, like, I think the industry still thinks that AI or a large proportion of it is kind of where it was even six months or so ago, right?

8:47You know, what we're seeing today is, you know, if you look at some of these data providers, they've got humans reading information from, say, public filings and put it in that structured format. You know, the bulk of the work that we're doing, what we're seeing is models are already more accurate than humans in those sorts of tasks. And so I think that will take a little bit of time in terms of confidence, But I think some of the more lower level, more kind of data gathering and presenting types of tasks are fully already being automated at the top firms. This is interesting. I'll tell you, at YC, the evolution went through at the same time.

9:22So two years ago, we would say, oh, you can't really sell software to investment banks or private equity funds. They don't really buy software. Or if they do, they buy one piece of software every 10 years. And the same was true for lawyers. But something happened last year. They all got very AI curious. They were curious about AI. They were like, let's do pilots to play with the software and try it out. And you were saying that this is the year when that turns into actual contracts. Exactly. And you can no longer say these companies don't buy software. That's wrong. They're all buying software now.

9:49And that's because this thing happened in the last 12 months. Exactly. And last year was the year of proof concepts. And now our average contracts are in the years, first of all. And I just think in general, people are seeing a lot more shorter term or instant value. You know, I also think just, you know, on the note of these firms historically, inclusive of law firms and others not buying software, that is, I think, true. I think the big difference here is this is like number one thing from the very top. This is like the number one thing on everyone's agenda. So we are not selling like the next CRM or like the next like data vendor tool or anything like that.

10:28We are selling what we're describing as the most advanced sort of AI solution for financial services in the world, right? And I think if you are a CEO or an exec in general, you've kind of got to take that call. You know what I mean? And I think that's really played to the advantage for all startups, for sure. You describe some of these sales meetings, like who's the decider? Like who decides to buy this offer and what are they like? CEO level or in general, just the most senior people at the firm, regardless of if you're a top five or top 10 investment bank, priority firm, as I said, or sovereign wealth and whatever it might be um which i think at the start we found like really strange because we we sort of thought that this would be looked at at a team or or group level it's really not i think it's so important um firm wide that everyone has to be involved from from the very top and actually what we found is you've really got to get that buy-in um you know from the right person at the top and then you've also got to get buy-in from the people that are ultimately going to be implementing the tool.

11:30So you guys are flying to meet all these firms where they are? Wherever they are. So we have a bunch of people working out of Hong Kong and Singapore now. We've just opened a small office in India. About half of our overall team are based in London and we have an office in New York. Again, I think it's clear the product is impactful, right? Because when we're demoing, we're laptop out, we're showing real use cases with real data, high use case frequency. So then if you think about why they wouldn't sign with you, a big part of that I think is trust and building that relationship. Because it's also a different dynamic.

12:10It's like a lot of times you're speaking to folks that if they make a wrong call here, they could get fired. And so we spend a lot of time building that trust. I think a big part of that stress is FaceTime and getting in front of people, obviously, but then spending the time on the demo and how that's justified and really investing in things that are specific to the customer. I want to go back a little bit to the two previous companies that you guys ran, Pat Llama and Fancy. Before we get into the details of what they were doing, what are some of the learnings you took from those companies into Model ML?

12:44I would say you've definitely got to enjoy it. You've got to enjoy building companies. is my overarching thought. What about you? I think you've just got to be prepared for the worst. You've got to be prepared for the worst and the most ridiculous rollercoaster experience. I think all startup founders will say the same. I think it's going to be a lot of ups and downs, a lot of the time downs, and you've just got to be prepared for that and know it's coming. And as Chas says, just enjoy it because it's fun. And is there a sense that when you start a third company that you've gone through that before.

13:18So you've gotten used to the ups and downs and you can just kind of like be calmer about it. You can definitely be calmer. Definitely. Definitely calmer. But there is certain things that you just cannot prepare yourself for. You know, I think like, you know, just in general, no matter how much you've kind of said to yourself, this is normal, you're going up and down, that'll be the odd thing. It will be employee related or product related, whatever, that will still surprise you. The fundamental thing that I think we realize that we've at least brought into this company is this concept of perseverance.

13:50Not blind perseverance, but perseverance. And I think there's a clear difference where if something logically makes sense, right? Like if you just, if in an unemotional way you're approaching a problem and it logically make sense right then you should probably continue doing that thing right and um not let anything stop you and i think like the consistency that we've noticed that of of founders that we've some of that we've invested in or work with is like the ones that kind of do that and really persevere tend to win so i think that's probably the biggest learning for us just persevere at all costs yeah for sure i mean one of the big ones for me in particular is definitely hiring i think when you know I was CEO of Fancy I was 22 23 so hiring was new to me and I think to be honest I didn't really have a clue what I was doing and we still don't yeah hiring is always tricky I think one of the biggest things right now and we say a lot like probably the biggest thing that comes up in an interview is is how much do you think you'll enjoy working with that person I think you know Fancy, again, being inexperienced and maybe naive, it was all about where they worked before, you know, what's on their CV, and you kind of ignored the little things that maybe you shouldn't.

15:09And now it's kind of the main thing that we look for, you know, we're going to be spending a lot of time with these people, you know, we work a lot and everyone on the team does, and you've got a lot of work with this person. So we often meet in person multiple times and make sure they're the right cultural fit and it really makes a difference I think right now you know we hire very slowly um and we try and make sure we hire the right people um and it feels like we we've got better at that as time's gone on and humans are just so much more impactful now as well so I think it's it's just even more important than it has been before um and yeah the other aspect dare I say is work ethic.

15:48I think we, you know, I think we're known for it. I mean, right now we're still working seven days a week, as we said, and we have done for about 18 months. And I think certainly in that initial period that you've got to do that. And I think, you know, our team works six days a week at the moment. Look, that may change, I'm sure, as time goes on. But I think part of that point about enjoying working with the person is they've also got to really enjoy what they do, right? So we've started to ask more questions around, Look, all the other sort of standard stuff that has to be done, I think it's important to have the kind of standard interview type structure.

16:22But are they going to enjoy their day to day? You know, the question that we ask people is like, you know, what have they been building? You know, regardless of whether they're in engineering, right? As soon as they start to mention things like, oh, they've been doing a bit of vibe coding or this, that and the other. We're like, OK, they're going to enjoy what they're doing. Right. And I think that's really important. It feels like there's a moment in a company's lifetime where it's suddenly it's working, but you haven't won the market yet. It sounds like at that time it really matters to work hard because you're not the only one trying to go after this market.

16:52Yeah, yeah. We just don't like losing. Maybe go back to Winter 17, Fat Llama. Tell us about Fat Llama. So Fat Llama was a marketplace that allowed people to rent items from people in their buy. But the main difference is you were insured. So if you're my neighbor and I lent you a camera, a$10 ,000 camera, I'm insured. If I lent you a drill, the same thing. That was what we did differently with the model. That model had been tried a lot of times before. But it still took us three years to find product market fit. I sort of defined product market fit loosely as, you know, the unit economics add up.

17:27People want, you know, what you have in terms of product in a sort of relatively large addressable market, right? But it took us three years to get there. And I think that was, you know, we learned so much during that period. And so, yeah, that was Fat Llama. And I can talk more about that in a second. but what about Fancy? Yeah so Fancy again consumer business so we were a last mile grocery delivery business which now everyone's probably bored of speaking about them but back then so this was end of 2019 start of 2020 so our model was a little bit different to a DoorDash or an Instacart so we were what's called a vertically integrated model which means we had our own warehouses we held the stock ourselves and this was relatively new in Europe I think there was one other player doing it out of Turkey, but in the UK in particular, we were the first ones.

18:19And really it started it as a delivery app for students. I was a student at the time and it really came about the need. I was really sitting there one day thinking, God, I could do some Pringles or some beer, to be honest. And I didn't want to walk to the corner shop that was like five minutes away. And this was during COVID? So this was just before COVID. This was like three months before COVID. We kind of built the app MVP in sort of four to six weeks. I studied computer science at uni and so it was just before COVID and really almost instantly as opposed to Fat Lama we kind of found product market fit almost overnight which is like sounds obvious right it's like you're delivering sort of beer ice cream to students in at the same cost as they would get it from the corner shop.

19:01So we took off and that was obviously really exciting and then we got on to YC and then COVID happened and COVID you know for our business was really that shot of adrenaline. I mean, COVID happened, everyone was staying at home. You know, the business really then started to take off. It was an amazing business. It's an extremely difficult business. I'm sure we'll come on to it in a bit, but we got hit with, we always like to say a lot of cricket bats to the face, you know, a lot of stuff can go wrong. But it was awesome. So we continued to grow in the UK. We raised money out of YC. And then we acquisition offer from go puff i think it was about 18 months after we started um and go puff at the time a market leader um they were very very big in the us and they wanted a um you know opportunity to come into europe and we were best placed for that um so it really was a an incredible ride yeah yeah it was like the at least for that amount of time felt like the perfect kind of startup ride i'd say uh but again i think fat llama was the polar opposite the story i always to talk about with Fat Llama in terms of perseverance was, you know, I was, you know, maybe very early 20s, obviously.

20:08And, you know, at that time when you believe in something and you're raising a small angel around, so this is pre-YC, right, is, you know, we were raising, I think maybe$50 ,000, maybe$100 ,000, which, you know, nowadays is like, I suppose back then, at least then it meant everything to us. You know, we were scraping around, we were doing half an hour pitches with an angel that might put in two thousand dollars right you know we were just doing everything we could we fundamentally believed in this idea and um anyway so we built the mvp we launched and i started accounting so so i was big on like yeah other numbers gonna make sense right so i had my like forecasted average transaction value my retention everything else so we launched and pretty much straight away like within the first day we got a rental and the rental was for$600.

20:56So I'm like over the moon. I'm updating. One of the first things I did is I'm straight into the financial model and I'm updating the average order value. And this thing is looking crazy. I'm like, we're going to the moon. We're going to the moon. Right? And so that was on Friday. It was due to be returned on Sunday lunchtime. And remember the criticism I got, I probably did 100 pitches to raise money and maybe one out of 100. I must have done close to 1 ,000 in the end. Which is, but the other 99 was like, no one's going to lend out an item because everyone's going to steal them. And even if you have insurance, the insurance is going to work, right?

21:26Anyway, so we had, and that's all I'm thinking, you know, it's all we're thinking about at the time. So we had this first rental. Anyway, it got to Sunday and the lender of the item called us and said like, I can't get a hold of the borrower of this item. So we're like, okay, it's fine. And then we realized after a couple of hours, like he wasn't responding and we start to get a bit nervous. One of the things we did, it was, you know, it was a native app and it's when people search because it helps with what you share in the search results. If we save or we look at the latitude and longitude, an accurate geolocation of the individual so we had that so i was like yeah i'm i'm going straight out there we've got to go and find this item and this was it this was like an 1800 pound drone right so for us this was like everything right so we went up anyway so i arrived on this random street in north london and there was just a door open to this house i don't think we've ever told this story because it's just like it's that awful and and the door was there door was open i couldn't believe it so i opened the door there it was the drone was just there on the side so i picked up this drone i got back in the car and i just went straight down to the to the lender's house i arrived at the lender's house but you've got to picture this right is what's going through my head is everything that we've pitched every person i pitched family and friends we had spent months probably close to a year at this point convincing people that this would work and they believed us and they believed danz and i and they bought into this and so all i'm thinking is like i've lied to them you know this is like my whole world was falling apart at this point and so anyway i arrived at the borrower's house uh the lender's house rather opened the door and he was like hey you're not a lender and i was like oh i know i gave him the drone he was like and you're wearing a fat alarm t-shirt i was like yeah we we deliver the item back automatically he was like this is awesome man like close the door and so we it did add up you know in terms of the insurance and we focused on verification but it fundamentally came to that that perseverance piece is like we believe this is something that should exist and we believe from a tech from a tech perspective we could make it work and it ended up doing so.

23:19But yeah, it was tough. Both outcomes seems like really, really great though. Like it's like such a well for this company. Absolutely. I mean, Arns did what the first 1 ,200 deliveries I think. Yeah, I think with Fancier from the outside it looks like a very fairytale story. Everything like after the fact. Exactly, exactly. They're like, wow, it's like I should start a startup and I'm it's like you should but really it doesn't really it's not all what it seems you know with fancy we had so many disasters uh you know and so many times where we were like god is this really going to work um you know even when covid happened uh you know we had to stop delivering for for a couple of weeks you know there was a point i remember i actually remember this sort of maybe two three months into the company uh so we were working with stripe as our payment provider and one day they just shut us off because apparently there was a breach in their terms of service, which later was a misunderstanding.

24:15But anyway, they shut us off for probably three or four days. Bear in mind, we couldn't take any payments, right? And at this point, I can't remember how many orders we were doing. Maybe not massive, but maybe 500 orders a week. Inventory. So there wasn't a problem with that, at least. Yeah, exactly. So at this point, what do we do? We've got orders coming in. We can't take payments. So we ended up just taking payments over the phone, taking payments with cash, taking payments from PayPal and there were so many of these stories and if you think of our business you know the amount of issues we had with delivery drivers with warehouses it really was an intense situation but that's really what made it fun right I think you know building a startup you just become so thick-skinned and you know problems come about you know on a daily hourly basis and you know kind of your mood is like this all the time but you know it's epic right?

25:04On the payments one we emailed Patrick directly and he responded like within an hour he sorted us out yeah it was awesome so if Patrick listens to us thank you Patrick he's sorted us out um but I think on both of those the the general story is and on you know the the team delivered the first 1500 orders right and so uh you know ourselves so it was very much a case of you know you're here but we rather than coding at night we're coding during the day because really our most of our orders were kind of in the evening, right? But what that meant was we were speaking to a customer for every single order, right?

25:36And so the customer feedback wasn't just like real time that you hear about. It was every order. We knew what was working, what wasn't working, right? Which meant that, you know, when we started to implement drivers, you know, we always say that you just, you knew exactly how long it was going to take from the warehouse to that house because you'd done the route about 50 times before. And I think across all three businesses, if you sort of set up and maintain the foundations of being incredibly customer centric, things tend to go okay. Yeah. There is a pattern though with startups, like as they grow, they move the builders further and further away from customers.

26:12And there's a bunch of other roles in between. And eventually you kind of have to interpret the information you'd learn from customers. How are you going to avoid that? Well, it's, I mean, you speak to everyone you can talk about on the product side. I think on the sales side particularly, because also we have a proof of concept phase still, right? So we still have a trial phase, which is this nice blend between kind of a sales stage, but also a pure customer feedback stage. Right, right. So naturally, and I'm 100 % involved with that. And it's also what I enjoy as well, right? you know spending time you know I don't like demoing on a screen I like sitting with a user with a laptop how they're gonna work and working on the product together and that also is the best way we found to sell but you also doing customer feedback calls constantly right yeah I mean you know it's always a YC mantra I think you know you gotta speak and listen to customers and that's really what we try and do you know on a daily basis you know really figure out what are their pain points with using model ML what does their day look like and then think internally, you know, how can we then productize that and get that into their hands for them to try?

27:20And it's that constant iteration, right? We always say, you know, the quicker we can ship things, the quicker we can learn. And really, we want to try and stay as lean as possible. You know, we always have this theory. We want to be the, maybe not the first, but, you know, one of the first, you know, 10-person, billion-dollar company. And it's just about staying so close to the customer and in the details. YC's next batch is now taking applications. Got a startup in you? apply at ycombinator.com slash apply. It's never too early and filling out the app will level up your idea. Okay, back to the video.

27:52So during YC, we have this group officer topic where we talk about what motivates you to build a company. And it's mostly to surface sort of the motivations for founders and their co-founders. So when things are really tough, you know why you're there. Do you guys remember what was your initial motivation when you started the first two companies? and then maybe what it is today? You know, when we have our one-to-ones, we're not brainstorming how to raise money. We're brainstorming actually how we can do this for the rest of our lives because it is so rewarding. You know, building something for individuals, making money, I think, you know, it just gets boring so quickly, right?

28:27You know, it's not really that motivating. You know, whereas building something that makes people smile, that's very impactful and a lot of people are using it, that's just so motivating to us. And I think, like, that really has been the story across all the companies you know I think what's been nice about this and surprising about this is we were we questioned ourselves going from a two consumer brand heavy type businesses into a b2b but actually particularly in the world of ai the most important element is like the b2c element and and that is to us the most rewarding element like when you do a demo to someone or they click run and they do something and it is just there's no money that can buy those moments.