Gauthier Van Malderen - Founder and CEO of Perlego

0:00But it's great to have you. Thank you so much for joining us. Like, why don't you just kick off, give us a quick introduction to who you are and what you're building. Sure. So my name is Gauthier. I'm from Belgium, Ridgely. I've been living in the UK now for about seven years and I'm building a company called Pelego. Basically, it came out of a personal pain point of mine, which was really expensive price of textbooks. So when I was going to university, I'd spend like, you know, two, three hundred dollars on a book. I'd read two chapters of the book and never use it again. and I noticed that all my friends they would download content illegally like piracy or buy secondhand books so yeah whilst I was at uni I had a small business which I sold and I've been doing Pelago ever since kind of seven years on we've raised just over 90 million dollars we're about 130 in the team we're active in 191 countries and we've kind of been branded by the press as the Netflix or Spotify for books so essentially you pay 12 pounds a month here in the UK and you get access to millions of books and it's it's amazing like I was having a look at the website and I said this is this is an amazing thing right there like and the pain points that you were just talking about which is the pain point that every student at university has experienced and then you know there's almost like the secondhand marketplace where you'd buy textbooks from people in the year above you and then you'd sell them on the next year and they were ludicrously expensive um so one of the questions I wanted to ask is like how did you how did you manage to get I guess like the critical mass required to make this work where you have enough books or publications on the site that then people are willing to say okay this is my like primary source of the information that I need yeah no it's a great question I think look the the biggest sort of difficulty of building a company like Pelago it's very much of a chicken and the egg right you need to lock down the supply to get on the consumers and in the early days I'll be candid I knew nothing about publishing nothing about ed tech and all that kind of stuff and a lot of the publishers said it was ridiculous however to my benefit this industry is very similar to the music industry like in the early kind of 2008-2009 if you think about it publishers lose about 35 % due to the second hand book market so students selling second hand books and then a further 28 % due to piracy so to publishers the pitch was look we'll create a recurring revenue line where we pay for all the consumption we mitigate piracy from happening and we're creating a more digital sustainable solution and also for their gross profits it's much nicer right if you think about it in your print model there's a lot of wastage in the supply chain print distribution wholesale returns so what's interesting on a hundred pound book the publishers might receive 23 to 25 pounds and the author might only get four pounds so what our big pitch the publishers was is cut out all the middlemen go direct go you know pure digital um and it's a much more scalable model well i said i didn't realize that the margin is so low for them was that did that make it an easy sell or was there still a lot of like coaching hand-holding to try and get them to see the vision uh i think it took a bit of time so like now you know fast forward we've got about 9 600 publishers over one and a half million titles we've got about four million on our database so we've done a really good job on that on that level um kind of how i thought about it is what we've done is we took all the reading lists in the UK, Europe and the US and we worked on content coverage.

3:14So if I needed to do an intro to venture capital course these are the four books I would require and we negotiate with those publishers and then we started to sign global full collection deals. But in the beginning it was working from a small cohort and then building up from there. Interesting, did you pick specific universities and specific courses and just go after that group with those texts? Yeah, so the way we thought about it was, if we think about, we might have the chance to talk a bit about our international business and how we think about it, but the main focus was on the language level.

3:49So English as a language, which gave us the UK and the US. And in those markets in the UK, we have about 2.3 million students in the US, we have about 19.8 million students. And then as you know, like in Europe, you know, a lot of students, I studied my undergrad in Italy. All of my courses were in English. So English was the first cohort. And then what we did is we both based it on what are the biggest kind of undergrad subjects. So business economics and management is much bigger than veterinary science, as an example. So we'd go after those kind of tranches and then work our way from there.

4:23Got it. Very cool. Very cool.

4:25Gauthier Van Malderen:What is, sorry, I have to unmute myself in two places. And what is sort of your second act here? So do you just go deeper into other languages? Is there, you know, I mean, what you're basically working on in a way is also digitizing education to some degree, taking a lot of these physical books out of that system. How do you think about sort of the next move where you want to go with the company? yeah so if we think about our initial vision was aggregating your textbook material or book material in one beautiful space the second layer of the vision was how could you build an engagement or learning tools around the content layer so with AI and we hopefully will get to talk about the implications of AI in Pelego today we've built things like workspace so you can aggregate your slides your notes your essays all in one beautiful space the third layer for us is to have collaboration where you could work with your peers and your classmates on that content layer so really moving away from the Spotify for textbooks analogy and to a marketplace where you can find all your core learning material people can upload their own content and then you have a collaborative and and like an education layer around it it's also important to know in terms of our go-to-market we have a really big d2c business but we also have quite a big b2b business so pal where are you from if I may ask

5:54Germany I can't oh nice okay like in Germany we have quite a few nice so we have quite a few like business schools that buy for all their students etc so in your first day of class you get full access to Pelago and they basically incorporate that as part of the tuition fees and then what's really cool is now we started to bring a lot of like you know best-selling trade titles so you might get Pelago your first three years at university and then after you'll get access you know you might buy it as a consumer after yeah I would also be like super interested in

6:25Gauthier Van Malderen:because because I feel like a lot of self-education that happens um on the internet is sort of you know you don't really know where that data is coming from like is this actually state of the art for this field etc so I'd like I mean the the inventory that you're building up and sort of the potential for summary and like learning on top of it that sounds so interesting to me in college but also beyond college right when you're not stuck to these physical books but you can kind of sit on top is that something you you think about as well yeah so in october last year we launched one of our latest products which we call research assistant essentially this uses ai in two layers we've built a semantic search with rag on top so you can ask any question in palego so let's say the implications of interest rate in the VC market and it will generate a beautiful 200 to 300 word answer but unlike other providers you have all the sources and the material in there and then you can go straight to the book and now a product that's coming out in the next few weeks is talk to books the ability to ask questions and tutor and work with your book which is really cool to help for your learning so yeah now that you have the content layer you can build all these tool set layers on top and that's really where we're seeing a lot of engagement and you know a lot of growth in the business essentially i mean i think the natural next question is how how does that work together with the llms in the long run obviously you're kind of you or your publishers are sitting on a lot of training data as well um do you know how the publishers think about that or how do you think about it yeah um so we need to be really careful from from that perspective And you might have seen there some recent lawsuits from Meta, Anthropic, who essentially scraped pirated books.

8:16There are about 460 ,000 books that were scraped to train the LLM models. Now, Pelago can do any form of trading in a closed environment, which stays on our ecosystem. But we are not allowed to train or sell that data to external providers. Some publishers have done direct deals. But what I noticed is the majority of the publishers are reticent to do so. and right now the first phase is more lawsuits claiming back you know the ip that was infringed but i'm not allowed to to sell or do deals you know with other other providers but i've got all this great stuff on my side that i can leverage and use and essentially because one thing to know is we pay back royalties to our publishers we pay a rev share back to them that's a critical component is we pay for all consumption both when you're training in our models but also when you're consuming the content as a general reader okay so there's also on on your cost side it's really like based on what people are using and consuming yeah so the way we pay back our royalty model is based on consumption and the value of the content essentially you know an introduction to german history book might be worth less than an introduction to belgian history book so my book would be a bit more premium as an example and i'd get a bit more of the payments back you know thanks for that yeah i'm joking i just you know i know seba i couldn't pick on seb so yeah yeah no that's super interesting so yeah you you've kind of got access to build the models on top of the data sets that you've got but do it in a compliant and a legal way um which actually it feels like it could give you a huge advantage over those sort of like generalist llms um do you see that being the direction that i guess like education goes in i think you know the first one of the first ai things i ever built was an ask aristotle ai bot and it was a bot just solely trained on aristotle's works and that was just because you could find generally kind of find them without any sort of you know they were out of the world went out in the world and it was just for fun and it was just like oh could you create a version of aristotle just trained on his text and what would he say if i'm in a moral conundrum um like do you see that that kind of being the future of the business and of education in general where you can have these almost like bots or llms to answer really specific questions about academic texts and and areas um yes and no so will palego you know try and compete against the open ais of the world i don't think so but where there's a huge opportunity is to do it on a course or subject level so we could take all the philosophy books and then you can train and learn off all the you know hundreds of thousand philosophy books we have that is something that's really exciting that we're looking to um i'm not sure of the applications yet so the way i think about it is you have general search you'll then have search or training on a subject level and then you'll have in a book level and because you know you read these three books i I can recommend those two books, et cetera.

11:15For now, we focused on search, study tools, and then we're also going to do what we call talk to books. And that's the third layer. The subject layer is a bit more complex. And the reason for that is, you know, it's a silly thing, but like some people's perception of this philosophy book might be against another one. So how do you balance and how do you think about the hallucinations from that perspective, which adds a bit more complexity and, you know, something we're testing and playing around with. Yeah, super interesting. That was kind of why I wanted to build this agent because I was like, I just I don't want hallucinations.

11:49I don't want other people's opinions. But maybe I want to find a good quote about what Aristotle thinks the good means or whatever. And that you will be able to do in Pelago in the next few weeks. Yeah, that's really cool. I can, you know, as a student, you can really see the use cases if you're looking for specific quotes to back up your argument and you haven't done the full reading recommended by your tutor. What's your take on like AI in general impacting and changing education? Do you think it's going to be something that fundamentally disrupts the way we educate our children? You see it as like a tool.

12:18What's your kind of overall take? yeah um so what is the question do i think ai is going to completely disrupt like um sorry could you could you repeat that in a bit more detail sorry yeah i guess we're sort of in this transition phase where students are now using ai across the board but it feels like the curriculum or the way that we test students hasn't changed despite them having access to these new tools and those new models do you do you think the way that we are educating and assessing children should change um the short of it is yes if i think about my own business which is more higher education and professional one thing that's really interesting is i'm not sure if you've got i noticed you guys started up and i'm sure you're getting applications to join uh and you probably get all these applications and they're all ai written and it's just like oh wow There's no originality to anything anymore.

13:12What I do think is AI could probably help in three main ways. One, personalization. The way you, Seb, learn is very different to the way I learn. And I think what there's a huge opportunities to create more gamification, more adaptive learning that you could learn faster and me less fast as an example. So number one, personalization will be really good. Number two, when I look at the education space as a whole, one of the biggest bottlenecks or fragments is actually you as a professor, as a student, you're very time poor. Right. And so can AI help you drive efficiencies from that perspective with automated grading or maybe like student progress plans, etc.

13:48And then I think the third one for me, like it's a it's a it's a big thing. But like. On Pelego today, there's a huge push towards accessibility. So we use premium AI voices where you can listen to all the books on the go. And if we think about the audio book market as a whole, only four percent of books are made available digitally. but because we have all these books in the background you can now listen to all those books so again i just think you know there's so many cool things you can do about make it more accessible with text-to-speech or you know real-time translations those kind of things so those are my three things that i'm excited about however i do want to reiterate like you know i'm seeing people are learning much less than just using chat gbt for everything and that just as a macro trend i'm a bit more worried about um yeah yeah there's definitely a growing trend uh like an overall pushback against higher education it feels like there's a lot of people are now advocating to skip university and i think there's a part a part of that is uh how much easier it's becoming with ai tools and how the assessment criteria hasn't changed so did you go to university if i may ask yeah i was in durham nice um nice i've had that's a very nice university it's lovely it's like a very quaint town very small um it's like the type of university you'd like your child to go to because it's like known for its being safe and boring um what about that's that's um so i studied in buccone and in cambridge but basically what i'm trying to say is education at the end of the day it's about an experience right do i think i got the value of my money from going to buccone and spending x amount no but it's going to come down to two things in the future one the quality of the classes and the experience you have as a student and you know i really think it's been very formative to create great friendships and things like that and that is something i would put place a lot of money on do i use my degree in my day-to-day business not much but i think the perception of education probably will change as an experience more than you know because you can learn everything online and candy everything i've built at palego be it raising money etc you pretty much learn it on the go right so i think yeah we're going to see a huge shift around higher education in the next 10 years and it's going to be much more like what is my value for money yeah absolutely super interesting

16:06Gauthier Van Malderen:and that way it's great because um you can bleed into the you know further adult life because it's not about just using the textbooks while using them and selling them to the next person but actually using them to get good answers and self-educate maybe also beyond those formative years yeah lifelong learning is the biggest trend in education and we see this with our own users you finish your university degree and then you might want to read the cfa or the gmat book or a book on hr policies etc and so what's interesting is a lot of our b2b clients like are buying for their employees and stuff so that is another trend is i think you won't just study once anymore you'll re-skill and upskill during your career like maybe three four times um so that's another thing i think will change amazing well i think we're out of time great um but thank you so much really great to meet you my pleasure um super exciting what you're building uh and yeah we'll be following along thanks guys have a good one ciao bye