Transcript: Maayan Yavne | MasterSchool | How to Build an EdTech Product

Eyal David: [00:00] Hi, this is Eyal, welcome to Product Builder, the home of the most interesting product conversations. I'm the owner of UserFlowzz and, among other things, the host of this podcast, and my goal here is for us to improve our product skills together. If this is your first time here, I hope you enjoy it, and if you've already listened and gotten value, I'd really appreciate it if you shared the podcast with a friend. So let's get started. Hi Maayan, how's it going? Hello. Maayan, who parked in the blue-and-white zone. Totally. I'm actually really the first. So with us is Maayan, who is VP of Career Training at Masterschool, did I say it right? Exactly.

Who is Maayan: from iOS development through Stanford to EdTech

Maayan Yavne: [00:43] Yes. So tell the listeners a bit about who we're talking about. So I'm Maayan, I'm VP of Career Training at Masterschool. I come from a technology background, I started as an iOS developer, iOS 2, old-school, something like 14 years ago. I realized I wanted to find a way to merge my two areas of interest, learning and technology. And I did a dual degree at Stanford University, an MBA and an MA in Education, and paved my way very, very quickly into the world of EdTech, educational technology. I started in product at a company in the Valley, in EdTech, called BloomBoard. There we built micro-credentialing products for teachers' professional development. It sounds much better in English, professional, professional training. And something like 6 years ago I met someone on... a woman who pinged me on LinkedIn, she said, like, too many people told us we need to talk, so let's talk. I said, interesting, and that's how I met Yael Ramati, who was my co-founder in a startup we founded called TailorED, which basically built personalized lesson plans, learning paths, learning paths personalized for students. A startup we opened in 2017. One thing led to another, we raised more than five million dollars, we reached almost five million users through an API, through students and so on, YC alumni of 2019. The startup unfortunately shut down, TailorED closed about a year ago, and today I joined Masterschool too, sorry, a year and a half, and I joined Masterschool a year ago.

Eyal David: [02:26] So we'll still touch on Masterschool and dive in too, but tell us a bit about TailorED. Why, basically, what did you do well, and why, in your opinion, did the company close?

TailorED: what they built, the pivots and why it shut down

Maayan Yavne: [02:35] Ah, a question, a question. So basically the insight with which Yael met me that evening, we met at the University of, in Palo Alto, and she said: how is it that a teacher still doesn't know what the most effective learning experience is for each student? There's tons of content, there's tons of data, how do we still not know how to solve this problem? And that's broadly the problem we came to solve. Not only for the teacher, we really wanted to solve it at the ecosystem level. Meaning, how what works for each student can actually drive decision-making within the learning ecosystem. What did we actually build? We built an artificial intelligence engine, when it was still hard, that basically scanned existing content, understood a student's needs on one hand, scanned existing content, and basically built a personalized learning path for each student. The personalization here wasn't about what pace to learn at, or adaptive in one way or another, it focused on how to recommend content, on how to learn it. Meaning, we looked at students' needs beyond just their skills or their knowledge, and we looked at additional needs like social-emotional ones, like confidence, motivation, ability to focus, even social ability, and based on that we basically built the content, each one adapted to their needs. What does that actually mean? Say a student with low self-confidence would get a worksheet structured from easy to hard, because research has shown that's what helps the student most. From there we started with the initial product for teachers, which builds lesson plans, and from there it evolved into a Netflix-style library with a recommendation engine, the ability to split students into groups within a classroom, and how to run a class with differentiated instruction. From there we actually won a prize in the AI Challenge competition in Israel. We managed to show an increase in the success rate in five-unit math, and that's how we expanded to about 70 schools in Israel. We were accepted to Y Combinator, and then we also expanded subjects, we started with math, English, science, and the first pivot we actually made, we started releasing an API for content creators. So basically we came to publishers and said, here's how we're going to start doing Personalization as a Service. Because one of our big difficulties was bringing high-quality content to our platform. So at first we started with partnerships, we actually released APIs for content developers who could either bring their content to us or use the personalization API on their own content. After a few years we basically decided to make another pivot, with all the difficulty, and we left the K12 world and went to corporate training, where we took the same technology and put it on top of, basically we brought personalization of learning paths for employees inside an organization. And there we said, okay, how are we going to bring personalization into an organization? Everyone hates the LMS, let's get rid of the LMS, let's bring learning to where employees are. And we basically built a product that sits on top of Slack or Teams and runs the employees' learning path where they are, which is in communication tools. So what, in your opinion, didn't work there? So broadly, the main difficulty in my view was the fit between the product and the business model, and a bit of a mismatch with the funding model. Why do I say that? Because we basically built a product for K12. In K12 you ultimately sell to schools or districts, a bit bigger, or sometimes even at the state level, say in New York. These are very, very long sales processes, I'm talking about a 24-month sales cycle, with relatively small contracts, usually something like up to 5,000 dollars per school. It's not simple for a startup to wait that long and work that long to see so little revenue in the end, that's one problem. So the model of selling to schools really doesn't fit an early-stage startup. Second, we chose a bottom-up model, which was very much, everyone had lots of hopes that a bottom-up B2C2B model would work in education. So what did we do? We said, okay, we'll solve teachers' problems. Because that's the best way to actually reach impact on students. We'll solve teachers' problems, that's why we built lesson planning. And then the teachers will be so excited, and they'll be champions inside the organization, we won't need to build a standard B2B model, we can do bottom-up, champions inside the organization, the teacher will bring it to the school, and that's how we'll conquer a school. And we'll get in with them through the teachers. With teachers we discovered several things. First of all, it's very, very different: however much a teacher loves your product, it doesn't mean he's a champion inside the organization. The structures of public bodies like schools are very, very different from private companies. Broadly there's a very flat hierarchy there, there are two segments, there are teachers and there are administrators. And a teacher doesn't always want to advance to become an administrator. And not always, because we had teachers who bought our product on the premium model and had no interest at all in the product being bought at the school level. They were very, very happy to spend the 10 dollars a month and use it, and they used it every day. And there are teachers who used the product very little but were crazy champions for bringing it to the school. Why? Because it advanced them professionally, because it gave them more presence, and that way they can grow their reputation within the district. And on the other hand, administrators, and I'm going to generalize, but not harshly, some do and some don't, but they don't always listen to teachers. Because what matters to an administrator doesn't always matter to teachers. Example: for a teacher, saving time is very, very important. But administrators have other problems, there are budgets, there are targets, they work in a public body.

Eyal David: [08:57] And let's say, saving teachers time isn't the number one thing. So you got all kinds of insights, and they also steered you toward pivots, which is probably hard too. So you have a lot of understanding of this industry, and we'd like to hear a bit about the trends in the industry. Tell us a bit about the EdTech industry.

EdTech's four segments and the revolutions in content

Maayan Yavne: [09:16] Yes, sure. So the EdTech industry is a fascinating industry that's really going through crazy changes, because if you want to talk about a super industry, it really is a super-conservative industry, but, but it really is going through crazy changes in recent years. But first let's do a light segmentation, to help. So EdTech is basically divided into four main industries: there's early childhood learning, which is broadly 0 to 6; there's K12, which is formal education, which is compulsory, K is kindergarten, 12 is twelfth grade; there's higher ed, which is basically higher education; and there's, let's call it, lifelong learning, or training, or corporate training, but it's adult learning, the whole topic of learning for adults. Those are basically the four big segments. Now, several very significant changes have happened in recent years. I'll start first with the world of content. Because content really is one of the main drivers in this industry. Let's start with how we all learned. Even in Israel almost not, but in the United States we learned from textbooks. When I grew up, everyone had McGraw-Hill, Pearson, and each had a textbook, you'd buy the textbook for the course. And that's how you'd learn. All the IP, all the knowledge, was with the publishers, the big companies. Broadly, two very, very interesting processes happened. One is digitization, meaning the revolution, the digitalization of content, where not everyone jumped, jumped, managed very quickly to close the gap. That really opened a door for small players, small digital-content players, who managed to produce digital content very, very quickly. We already know that the textbook publishers actually failed there. At first, until not long ago, they only had PDFs, let's put it that way, and only today are interactive digital products starting. So that was the first opening that really caused a big boom in types of content, and the medium in which you learn also changes the type of content. If once you knew a page, now it's interactive, it can move and so on. And second, in the United States several laws or changes happened at the federal level around standardizing content. Basically, to break the textbooks' monopoly, it was decided at the federal level what the standards are, what students actually need to learn. And it really went through laws, very, very, it's numbered and really at a very, very precise level. And then they basically passed laws that teachers need to teach according to the standards, and teachers are also measured on how much they, or how they actually need to cover the standards. Which caused a crazy rush of all the teachers to search for content by standards, and the content had to be, as they say, aligned to standards, standards-aligned. That was a gap of something like six years, exactly during that period when the big companies, until they moved, until they really managed to move all their content to standardization. Also a crazy boom of all kinds of players, where the first thing they did was simply produce content that's aligned to the standards. So really from 2010 to 2020 simply a crazy boom, a market that's going, simply growing, going through very large fragmentation. The teacher's role already starts to change: instead of opening the textbook and teaching the textbook they've known for 15 years, suddenly they find themselves, Sunday evening, starting to search for content. And you say 2020 not by chance. Right, because in 2020 the very, very significant part happened, Covid, which basically forced... there was, let's call it, the digital divide. Meaning, there were the teachers who made the move to the digital world, or at least brought in digital tools, and there were those who didn't, and everyone lived in peace with that. Whoever did it did, and whoever didn't didn't. And Covid was really the final push. It forced, for, we got it relatively little in Israel, but say in the US there were at least two years remote, where basically all learning became digital, which really forced everyone into digital learning, which opened up crazy budgets, and also finally pushed the whole industry forward,

Eyal David: [13:39] across all four segments, not only in K12. Okay, that's one trend that happened? Yes. What other trends do you identify in this world?

AI as the third revolution: content as a commodity and personalization

Maayan Yavne: [13:46] So now there's no escaping talking about AI, because, I mean, I think AI did, is going to do, is doing, because it's not even in the past, we're really living it now, the changes, among the very, very significant changes, and it continues the world of content. Meaning, at first content was someone's IP, say in textbooks, it belonged, you want to learn something, you need to buy it. After that we moved to a digitization process, where it still belongs to someone, and then it was also premium. So you could, and there was also Google, so say, there was accessible content, not always high-quality, not always tailored, and you buy, you basically, there was content and you could learn, but you had to pay for higher-quality content. And finding content was a problem, especially, say, that was one of our problems at TailorED: we built a personalization model, but on existing content, and there wasn't enough content. And that's where we got stuck. So everything was fine when we got to areas like math and English, but if we wanted to reach some somewhat niche topic, we were very blocked in terms of existing content. And here comes AI and makes a third revolution in content. It's a revolution where you no longer need to buy it, it doesn't belong to anyone, it's a commodity, like I get water and internet and electricity, and today I have content. And that changes all the rules of the game. It's also high-quality content, not just content? Yes, it's not just content, that's it. And then now, it's not, now there's content, and not only is it content, it's personal. It solved the whole personalization problem. In any given situation I can choose what I want to learn, how I want to learn it, in my context, and what matters to me, tailored to my learning objectives. It really levels the... levels the playing field for everyone now. And what does that mean for teachers? Because AI has another aspect,

Eyal David: [15:39] right? There's the whole co-pilot aspect, which basically helps me practice, like, train better, focus better.

Teachers and AI: a tool for the teacher, not a replacement

Maayan Yavne: [15:47] How do you think it lands with teachers? It's funny, it really, I think it's in the conversation, it really continues the line I also saw before AI. I see all kinds of conversations where we'd come into a school to sell, and there were the teachers who loved it, and there were the teachers who were like, wait, this replaces me. Like in every industry. Yes, wait, this replaces me, and we're like, no way. And because of that, by the way, we made the very, very important product decision: no, like, not to replace the teacher, it's a tool for the teacher. That's why we always built it as a tool for the teacher. That's a mistake, by the way, that many made in the 2010s. They said, come on, we'll fix education, let's build products for students.

Eyal David: [16:22] And they forgot that the moment a teacher closes the door... Exactly, moving the cheese in these cases, no, that's exactly the person you want. So I really connect with that. So how, in your opinion, does AI now enhance the learning experience itself? How can teachers actually use it?

Maayan Yavne: [16:37] Totally. I think that when you look at AI and the teacher's role with AI, it basically splits into two parts. One part is teaching them to use AI. These are new skills that today we all need to learn, from students to the regular employee. The productivity tool: how do you actually need to use AI to function in today's world? That's one thing that changes the whole learning experience. And on the other hand, for the teacher the shift is from what to teach, like, from what we learn to how we learn. Meaning, okay, everyone now has a private tutor. What are the additional parts the teacher brings to the table? And that comes through in evaluation, in assessment. Meaning the teacher in the context of evaluation: if everyone has the tutor and is learning, how do I really assess that the learning sank in? And alongside that, of course, there's the social-emotional. Meaning, how we support. In the end we always say that learning is a social experience. However much you try to isolate it, learning is ultimately an interactive thing.

Eyal David: [17:48] Sometimes digitally interactive, but interactive. So it's interesting what it means that you're suddenly interactive, but if I can put you in a chat with four bots and you won't know it, then maybe that's actually a key to success. I think there's also another thing, like you said about assessment: assessment is also important for the student himself, or whoever is learning, because, just saying, ADHD, right? Give me microdosing, I don't need all this goodness, as someone who went through the education system and thinks that way. Okay, cool, so you give assessment as one place where it can really help, you also talked about content. What other areas, in your opinion, are real opportunities in this industry? We're already really, really... no, you can see it from the solutions, from the startups or the big companies that are already starting to bring it in. The first thing, the first wave, everyone jumped on content creation.

The opportunities: content, personalization, tutoring, assessment and compliance

Maayan Yavne: [18:36] You can really see it in all the players, everyone suddenly with AI, now you generate as much content as you want, in whatever form you want, and that was the first stage. The second stage is the personalization stage, meaning, okay, whatever content you want now, but at whatever pace you want, content recommendations and so on. The third stage is basically tutoring, what we talked about, now it's really personalized in the experience, in your experience. Meaning, in your world, tailored to your needs and so on. And the fourth stage is the world of assessment, which in my eyes is the opportunity part. If I look at all of these, we see the changes in the field of evaluation, which is where I think it starts to get really interesting. Because it's an opportunity that's created not thanks to AI, but because of AI. Because thanks to AI we now have content and learning experiences and so on, but because of AI two problems are created: how do we assess learning? And we all experience this today, anyone interviewing today knows they give a home assignment. How do we know if it was created by AI, and is it good that they use it? On one hand you want them to use AI tools, on the other you don't really understand and can't assess the knowledge. And on the other hand there's the matter of compliance, which is a very, very significant factor in the world, in the learning worlds, especially in places that work with government bodies. How do you create compliance and safety rules for individuals? That's a problem that this whole world will now need to solve. No, it's huge, I mean, I'll give an example, it really blocked entry to the market. Meaning, the first wave of content creators, the first thing they tried was basically to sell all these tools to schools, and it was blocked at first by a kind of statement of: we haven't formed our AI protocol yet, so we basically can't buy any AI until then. So like in the world where they did it with crypto and blockchain, where the first thing was building compliance or regulation tools, here too there's going to be a very big wave of tools that check their compliance. And that will be very significant for traditional education bodies.

Eyal David: [21:03] Right, wow, I wouldn't have thought of that. We're also kind of dancing around school all the time here, when basically the significant implication will probably again be on the world of work. And we're talking about classrooms, we're talking about teachers, but suddenly my way to get in now and learn some, I was going to say new profession, but it's actually skills, a collection of skills, right? That's also how you described it in the preliminary conversation, much simpler now, much faster. The question really is how good the quality of it is, how well I know how to operate on my own.

A world built on skills, not roles

Maayan Yavne: [21:35] I think it really breaks apart the... the more the world becomes multidisciplinary, let's call it, the more we basically see that once there was one skill set, one role, one skill set, and once you'd specialize in it. Today roles are becoming more multidisciplinary, and not only that, the roles of three years from now don't exist, those of three to five years from now don't exist yet today.

Product skills, follow the money and selling to B2G

Eyal David: [22:02] I called that skills, right? Yes, so it's no longer a role. What's a role? It's a collection of skills. If a role is a collection of skills, then we say, okay, we need to break down the skills, and then how do we teach those skills. And AI gives this flexibility, the ability to translate very, very quickly, to do reverse engineering: what's the market need, the skills required, learn it, and then apply it back. Or the collection of skills most relevant to several jobs that will be relevant ten years from now. Exactly, and those are the skills we basically need to focus on. We don't know the skills, we need to learn them. So it's like, as you say, it's basically two markets now heading into a big revolution: the job market and the training market, the education market, and both are now going to dance some dance, and it's interesting how it will end. Totally. I have a question: looking at this industry and previous roles, what, in your eyes, are the most important product skills in a product person? Interesting. Since we already talked about skills. Yes, the connection between, and I'm talking, so I think product is actually growing stronger. The need for product skills is only growing with the changes, with the change in technology we're seeing. Because let's look: today we have AI tools that help us with development. Okay, so maybe you don't need a developer, but throughout the history of product there was the business that says, here's the need; there was the PM who says, okay, this is what we build; and there was someone who developed it, and then someone who checked whether it really met the needs. So now let's say AI tools take over, and I don't know if they'll make development redundant, but they'll really be the development process, so now we're going to develop everything. So what basically grows is the need to define what to build on one hand, and on the other hand we have the need to check whether it's right or not. And those are two product capabilities that already exist today and are only getting sharper. So if I look at the skills a product person needs, in the end, in my eyes it always starts with... I don't have to say it in English, but, Driving, Driving Towards Clarity. That's always the number one skill set: driving toward clarity. Constantly getting to the truth of what really needs to be here. You get a project and you say, what's the clarity, what needs to happen to define this project, and how do you first define the problem and then the solutions. That's one thing. The second thing is strategic communication. I don't see AI solving that soon, and it's a need that's only rising. On the contrary, the more AI enters our lives and separates us, not necessarily separates, but I'm just saying, stands between us and interaction with other people, the ability to communicate strategically in the process of driving toward clarity is very, very significant, and structured thinking. I see as something very significant for a PM the ability to break down, as I said, from a problem, from a need to a problem to a solution, and to close the loop. Interesting. So surprisingly, yes, I connect with all of it, because surprisingly these are also really the answers we hear from hundreds of industries from other perspectives. But yes, I really agree with you. I think what you basically connected here is several things that are very product-y, but we gave here several things you can start learning, but it's basically a business view, a commercial view, a get-real view in which we really understand what's needed here, which is maybe also a bit of experience, right? I don't know how you even learn it, to strip and strip and strip until you're really left with that good scope, and then also to make sure, as you say, systemically, to close the loop and see that this thing is really converging correctly, to continue from there. It's really like that, and AI tools, and what you can see is not only in product, I really see it in other roles. Let's take the data analyst role, just as an example. A data analyst today does number crunching, and now, the more our AI tools get stronger, they can do it for us. But what hasn't changed is saying what we ask, meaning, what's the question we want to know on one hand, and on the other hand, after he's done the data analysis, is it accurate? Because our AI tools still aren't accurate. So the same in product: we can use lots of tools in product to make ourselves much more efficient, to write our requirements, analysis, competitive analysis, all the tools will basically be more effective, but in the end there's the checking part, whether it's really right, and the judgment that comes from experience, because product is many times an apprenticeship role. There's another thing here, that there are so many more problems that will be solved that you need to be very precise about the solution you bring or the specific problem you bring, and that's basically something you're describing. Yes, totally precise about the problem; a solution can come, but that's what driving toward clarity is about. Meaning, wait, wait, are we aligned on the problem? What's the problem? What's the need? Specifically, that you communicate and align everybody on the problem, and then once it's defined you can basically move forward, and then check whether the solution we got, in whatever form, actually really addresses the need or problem we started with. And if we really look for a second at the market you live in, EdTech, what's basically the crucial thing, the thing you need to track, that you learned is what makes a company succeed or fail? Yes, now you've put me in a bind, because everyone is going to think I'll talk about student success, learning success, and I'm going to sound very cynical. But the truth is that, like every other industry, the way to succeed in the learning world, even more than other industries, is follow the money. That's always true, in my view. But who actually holds the money in this industry? Yes. Because a significant percentage, if I sum up the four segments we said, early childhood learning, K-12, higher education and adult learning, I'd say three of them are broadly funded by government organizations. And adult learning falls under that? Could be, yes. Training, there's vocational training, so a lot of the programs are funded. Partially there's of course private learning within corporate training, but there's a not-small percentage of vocational training that's funded by government organizations. That's it. So let's say maybe what that actually means. It basically means, at least as I understand it, that there's a body or several bodies with several agendas that aren't necessarily the good of the student, like the good of the teacher or making the system more efficient. You need to satisfy the agendas, you mustn't move the cheese too much, as we said, as you basically discovered. What other things do you identify there that are a big no-no, that you discovered along the way? I think we're talking, like, it's classic B2G. They don't think in a year or in five years, they think in decades. I remember I once called a district and they said, wow, this is a really interesting product. We, it was in 2020, I said, we'll bring it in in 2024. I don't know if we'll even still exist then. So it's the same mindset, they don't think about next year or the coming months, they think in decades and long-term impact. So you need to understand that, like, it's different timelines, with different agendas too. When you talk with simple bodies, many times they're public bodies, elected officials, who have complex motives, let's call it, because on one hand... so the last thing I'll say: not everyone has the agenda of learning success, though I think in the end that's the intention. But there are lots and lots of stakeholders here with lots and lots of, like, other motives. And the way they work, they work in a very regulated way, and anyone who has worked with B2G knows it. The date the budget is released, the budget line, and when it closes, those are very, very important things that actually dictate the seasonality of your market. I'll just give an example: in K12 there's a very, very defined pilot season, it's in Q2, because basically the budgets open, and all the purchases happen in the summer to start the school year. And if somehow you missed the train, that's it, a whole year you can't make sales. Like, you can, but it's not at the same need. Second, if you build a product and there's no budget line defined for it, aligned with that government body's alignment, you can be the best product in the world, it won't help you. And another part is that whoever buys your product often has very, very different needs and agendas from whoever uses your product. And in the learning world, or specifically in the K12 world, because that's what I know well, I call it the starting four. You have the teacher, you have the student, you have the parent, and you have the administrator. The administrator is the buyer, the teacher is sometimes a user, he's like the gateway, the student is often a user, and the parent is a kind of secondary stakeholder who has lots and lots of motives for the process to succeed. And in the product you need to take everything into account for it to succeed. That's it, you need to streamline it for the teacher as much as possible, and for the administrator. So that's, for example, the mistake we made in the startup: we said only the teacher, we'll do B2C2B, we'll build an excellent product for teachers, and as I said earlier, obviously it'll spread like fire and obviously the administrator will need to buy. What turned out? We made several mistakes. Because we were a relatively unique platform, it was artificial intelligence still at a relatively early period, we were a platform that combined both platform and content. It didn't fit into anyone's budget. Wait, just, what are you? Are you an LMS, are you math, are you English? We said no, no, we're math and English and personalization and, does everything. We don't know who to refer you to, because they didn't even know where to buy it from, so in the end they even placed us as a content product. And we weren't a content product, and we were also relatively narrow, so if you go head to head with a very broad textbook catalog, it doesn't work. So that's one mistake we made: we didn't have a budget line through which they could actually purchase us. Second, we hardly developed tools for administrators. And in the end what matters to administrators is analytics. Proving how well they're really doing their job. So because we didn't invest in that, and we put learning metrics that matter to teachers, but not necessarily the metrics administrators are actually measured by, it interested them less. So even when we did studies and showed how, in a randomized controlled trial, we raised learning rates by 84% in math, in the end it's not the most... he doesn't have a dashboard that comes and says, wow, wow, like, the school is really number one in the whole district. Let's give an example of someone who did do it. There was, I don't remember which regulation, but they started really putting emphasis at the federal level on, let's call it, school climate. So it says, we basically want there to be a climate where students live and feel safe, and the social-emotional state of the students, and all kinds of budgets started for raising school climate. A company like Panorama Education came and said, no problem, here's a very basic product, a survey, which is like a Google Form you send to all the students, asking them about 4-5 climate metrics, and here are excellent dashboards created, which afterwards you can come and say, look at our school climate, we measure it and it really went up, and then I'll release more budgets in schools. A dizzying success as a product. So if I go here with follow the money, then basically, okay, who's your buyer? Then, basically, what drives him? Which is really, as you said, that screenshot of "hey, how we" this, we improved, and then a basically pretty simple fit of tools for him, getting in, as we said, seasonally, like, again, according to this dance, and probably also according to that budget line. What, what takeaways do you have around follow the money? And then you need to keep doing alignment, because what do we all know? There are all the people who come and say, like, products in education, education is broken, all the products are terrible, but that's part of the reason, the products are kind of broken too. Because let's take the typical LMS that we know, everyone hates their LMS. It's not that they're bad product people, they simply sell to other people, they build for administrators. But then the result is that people buy it, and then people use it, and that's why engagement is always one of the biggest problems in learning. And everyone tries to raise engagement, but, and then the moment you've built a very, very big product that fits, which is why administrators push them forward, now you need to create, to carry it over, alignment with the teachers, and solve the teachers' problems, and solve the students' problems, so that they'll actually use it. Okay, but as you said, the skill of a good product manager is really to build the strategic plan, right? So basically, let's solve the first problem, of getting in the door, getting a budget answer, then let's solve for the teachers, then let's get the students to use it. I don't know, the parents are also there somewhere along the way. And I have a question, let's stick for a second with engagement, because it really intrigues me. So basically in today's world, of co-pilots, what we talked about now, how do you think they can come in, how would you integrate it into the learning experience? I think I'd, let's think like product managers, what are the problems we have if we look at the engagement problem. So let's take, let's do a bit of stakeholders on the... If I want to solve the engagement problem, then the stakeholders are basically the teacher and the student. Who is measured on a student's engagement? So in the K12 world very little, but let's go to a world where engagement is linked to learning outcomes, like in corporate training, or Coursera for example, or all the MOOC course models where people pay on a subscription basis. Yes, it really went well, because in the end, what is the teacher measured on, that's the next persona we're driving toward, right? Yes. But there's a really interesting area where you measure a teacher on engagement versus learning outcomes, let's call it efficacy. That's always a tension. You can be very engaging if you put a TV in the classroom, like we always remember in the nineties, very engaging, not much learning. We learned English. Yes, so that's basically the significant challenge we're trying to crack with engagement: is it because the learning process is so tedious and hard and not good that I need to show engagement in the learning model I chose? Or... and that's why I think you can't separate engagement from the learning model. And here we'll look at a co-pilot, for example. The learning model of a co-pilot today can be through questions and answers. Of conversation. Which I think is a model that on one hand is very interactive and can be engaging, but like everything in the world it's one size fits all. Not everyone knows how to write. Not everyone naturally engages, because they prefer maybe more personal learning. And that brings us back to how to learn. So I think at first, how we look at co-pilot and how we want to use it to increase engagement, at first it starts as a supplement, and that's what we really see in the market. It starts as a supplementary tutor. The last thing they do is completely replace the learning process. It starts as, you're in the standard learning process and now you have a co-pilot that's a supplementary tutor for you to fill in the gaps. Either as a question practicer, or for going deeper, all the things like we talked about, personalization. And there it's basically a tool to engage with, because if you don't get along with the standard learning model, you can go deeper and improve, and that's a process that increases engagement. I think that as we progress in technology, the co-pilot will really start to take a much more central part, and then it's the world of tutoring. Which may actually somewhat replace or make redundant frontal, interpersonal learning. And there will be a not-small challenge there. So before we run there, I'll just say that, okay, if we stick for a second with the teacher, with the persona we talked about, that persona can be very afraid of what you just described, right? Who will replace me, as you said. So we need to wrap that too in a way of, okay, it's a force multiplier, like, you're going to gain from it. Maybe, basically, if today you invest a lot in students with learning difficulties, maybe this thing will come off you. Like, you need to wrap that too as a multiplier. It's not a new problem at all, we already saw it in the digital process. Meaning, that's also why you start from the teacher and what the teacher's problems are. The number one problem of every teacher, no matter where in the world, grab a teacher, what's the number one problem? Number one is they don't have time. They don't want to learn new tools. Not always, but there are, I'm saying, there are more advanced generations, and there are teachers I'm crazy about, they're really amazing. Go on Twitter and you'll see, surprisingly, but really, you want to see crazy professional learning fields of teachers, go on Twitter and see what's happening there. There's really a new generation of teachers who are very tech-savvy and really ready to bring in all these tools, but at the same time they have a very, very specific job definition: they're measured today on covering material, always. So if you can save them time in preparing content, and that's why the first thing everyone did was better content, like more content, and if you can help them with the variance of learning in the classroom, that's still the number one problem, of 1 to 30, and here it'll play a very, very significant role. That's what personalization comes to do. Meaning, we all know a private tutor is the best, it's been proven: you have a 1-to-1 ratio. In a standard classroom you have a 1-to-30 ratio. How do you basically give a 1-to-1 ratio in a 1-to-30 setting? Through personalization tools. And that basically lets the teacher move forward faster and keep the whole class engaged. And that's how you basically enable it. And of course there's this compliance where teachers need to give reports or look, and in a way where you can measure where each student is very, very quickly, you can also create very, very quick solutions as a result. That's it, and probably along the way we'll also have no choice: to engage the student and his parent we need the product to also be engaging so that he'll actually use it, otherwise he won't feed the entire machine with his, right, with his behavior? If the student doesn't want it, exactly, like someone standing... I'm not even a student, if the learner isn't interested it won't help you. But that gives us the why, why invest at all in engaging, in engagement inside the product, at the end of the day. I think the why is often very complex, because I too, like, we all immediately think about Duolingo, or right, that we think about an addictive product in learning, but many times the why in learning isn't always related to the product. Like, an example, we talked about it a bit before, but there are all the test-prep products, like the GMAT or the psychometric exam or all those exams, where the products today can be awful, it's just exercise after exercise after exercise. The motivation is very high because I want to pass the exam, the product... Okay, we talked about K12, leave it. Yes, in K12 motivation is really very, very important, and it's among the hardest, among the hardest problems, and here I really agree with you that the co-pilot, through personalization of content, can greatly increase motivation. For example, even as a teaching strategy, I'll take a little from TailorED: one of the things we checked was motivation, because motivation greatly increases the chances of success. And one of the solutions for raising motivation is using real-world exercises. Especially in math, because you can teach a triangle, and you can teach, like, a bike route or something related to something, so that's basically a way to raise motivation, and that's what gets built into a product. And that's how, say, AI tutors and co-pilots can basically do personalization through the content world in which the material is learned. And that will raise motivation. It's not that now there's, like, it could be, and there are students for whom just engaging with something extra raises their motivation. So there are many ways to raise motivation, our world of tools is just expanding really significantly. That's true, true, and I'll leave you with a thought for the end, something that makes me think now with Gen AI tools and also with Duolingo, I noticed it. The whole speech thing has come in, where you actually talk with the machine, and then suddenly, I at least think, okay, in 20 years, what value will the written word have, when basically all my interactions will be by speech? I don't know. So like, what is it, it's good, okay, you've opened that up now. It was a really successful episode, Maayan, how fun that you came, thank you very much, and you're totally coming back here. I'll ask you, so what do you wish for yourself for the next two years, Maayan, in two years?

Maayan Yavne: [45:04] I think to be true to my craft, and to grow and learn. I wish myself the acquisition of new skills that I don't even know what they are today.

Eyal David: [45:18] Maayan, we'll meet, we'll meet in another year or two and do a catch-up face to face. It was really fun, and bye for now. Thank you very much. Thank you very much, bye bye. Friends, thanks for listening. If you found this podcast valuable, you can subscribe, follow us of course for more episodes on Spotify, Apple Podcasts or any other app. And of course, if you didn't find us on some app, I'd be happy if you wrote to us. We'd really appreciate five stars on every platform and for you to follow us, so more listeners can discover us and find the podcast. You can also find previous episodes on any app or on the YouTube channel, we have links in the description. So until next time, come on, be effective, and bye bye.