 Welcome everyone to this keynote session today of Agile India 2021. We are really excited to introduce Aditya Rasti today. She's coming to us from Los Angeles and she's the founder of the EdTech company in Baib. Look, we're really looking forward to what you have to say to us today. Fantastic. Thanks Matt. Hi everybody. It's awesome to be here. Matt tells me he's joining us from New Zealand. I'm in LA and most of the work is in India and I'm sure you guys are tuning in from everywhere. So it's a small world like they say. Anyway, I'm here to talk to you about our little story. I run an EdTech company called Mbaib. It's an artificial intelligence platform for education now almost like nine years in the making because building anything in AI requires collection of data and it's been quite the journey for us to get to this point. So I'm going to be focusing my talk on sharing a little bit about how we execute and I know that we have a bit of a curious name for the session today. And I'm going to sort of start by talking you through it, tell you a little bit about Mbaib and then get into just four memorable stories along the way that will give you an insight into how we execute as a team. And I'll perform most of the time. So let's get into it. What is what is Octopus execution right so we try to do a bit of an illustration for you. So typically, you know, teams are run in a very functional way and at least most companies that I've come across you have like engineering and you have, and you always have a product person that will sort of bind everything together. And I got it in by so an octopus basically is one organism with nine brains and three hearts and eight arms, and each arm is an autonomous engine in itself because it also has a mini brain so it's empowered to do its thing. And it gets around does a lot of stuff. So thinking that you know our nine teams, and there's just some examples out there at front end back end, you know, content in content you have question bank you have media you have, you know, the knowledge graph team and then you have, you know, a bunch of other things that we do. We all work together. So every day at 130 in the afternoon there's a stand up which is literally cross functional for the whole company every problem solve for the eight channels that you're looking at right now. It's like government to the students to parents to even rural India to NGOs you know how do you solve for their problems in education and how does one single product platform serve it. If you look at the three hearts you know they're we really really live by our values as a company and three of our values that are very sort of heart like our closely net, which basically means you know working extremely closely together. And problem solving not as you know I raised the giro ticket on you so have you solved it, but actually sort of collaborating to get to the outcome of the problem statement that you're working on. Empathically strong you know we're going to come about you know talk about a story around that is like really really listening to stakeholders around us, and making sure that we leave with empathy and not sort of just, you know, functional execution. And then vision led, this is the first value I mean in fact if you enter in vibes office this is what you would see. This is a this is like a bit of a, you know, like a poetic thing which has been on the door of our office ever since we set it up. Enter this this space if your heart beats for education if you dream of a new sky, if you will not accept the status quo if you want your dent in the universe if you will not rest till you get there. What does this really mean it means that we don't just want to solve for yet another meat to ed tech company, you know I build video conferencing for teachers I build recorded content for students. We wanted to shift the game. Now, to take you guys to what that game really means I've built a bit of a schematic which I don't expect you guys to go through in detail but just to give you a sense of what imbibe does, you know and how you're different. So, if you look at it, you know, education is 40% of consumer expenditure in general, I mean everybody feels and especially in countries like India, doing better in learning is a way to outperform the demographic space that you're in so a lot of sort of people especially like, you know, I mean, lower middle class middle class people would just spend a lot of money on education saying that this is what's going to take us to the next level. Unfortunately, you know if you really get into statistics right only like 9% of Indians are able to spend about 2000 rupees per month on school, and yet sort of most of tech companies have an average fee of about $2,000 and they pedal inputs which is take some videos take questions, you know, solve it out, blah blah blah. What you want to do is you want to take that entire set of basic inputs and convert them to advanced inputs and then transcend those inputs into outcomes. So we say that just like an MRI machine makes a doctor a better doctor, we want our product to make teachers better teachers, and think of it like this that every student is able to study through the mind and is guided through the mind of a million teachers, and the teacher is able to become as strong as the heuristic experience of the mind of a million teachers. So, and the end game of this is actually to build a product that we've just launched called achieve, which allows you to build personalized AI journeys unlimited, you know, number of journeys that are powered through data, and that self select exactly what each child needs to do to improve in their context as per their stated goal. It's a promise of output or outcomes. It's not a promise of take more content and building this and building this through 15 years of, you know, student learning has been quite, you know, the interesting challenge and now we're seeing with our latest product achieve. You know what started earlier on as a giant medical company and built an AI stack that was fragmented. Now we're looking at a single platform with about, you know, almost half a billion hours of, of, you know, learning data, and about two billion questions with all of the stitch together into a single backbone and an ML and an AI layer that serves all of these advanced inputs. And, you know, we are gaining traction and it's just a very, very exciting time to be at imbibed. So let's get into the stories. Right, so I'm just going to talk to you about four different points in execution where we face problems in the company and we wanted to solve them. So if you looked at the previous slide right it talked about deep AI and like, you know, wanting to build a very solid, solid layer. And I was reading somewhere about this thing where, you know, the lady who set up Google image search sat and tagged 12 million images and that's how that really happened. So I was like, you know, this is about 2019, just after the reliance investment into imbibed. We're actually part of, we're a subsidiary of geo platforms. I was talking to my, you know, chief data scientist and I was like, you know, we're not really cutting edge, we have to be cutting edge because if we really want to solve for education through EI and ML we need to be able to build fundamental platforms that look at, you know, computer vision for education, because if you cannot let a machine read content the way a student reads it or a teacher reads it, you're not going to be effective in terms of making suggestions of content to a student or to a teacher. Second, you know, the machine needs to be able to understand and imbibe what that content truly stands for. So natural language processing became a very, very important thing for us to focus on. So two years ago I was like, we were obviously in the middle of like a major launch revamp saying, let's, you know, acquire seven companies put them together. And now, you know, build this whole thing out and everything that we built, you know, for personalized improvement in the medical and the it space is not going to go from first grade to the last exam you'll ever take and job entrance exams. So I said, I said, I told him, I said, don't do not come out till you've done something good with it, you know, build some good semantic models build similarity scores, you know, run, you know, run all of your fancy natural language algorithms on on knowledge infused graphs which is all of the data that we've collected. And somebody's verifying Microsoft outlook apologies for that. So, so he went away and he did it and as a result when he came back, you know, and I'm disturbed. This is again, you know, we were all working we were dying, you know, of like, sometimes like people issues there was, he was not disturbed. And, you know, now we've actually as a result of that many resurface we've been able to build a doubt resolution system that has image processing with greater than 95% accuracy that is able to solve, you know, almost like 95% of questions we've done so far with both an image and text with real time doubt resolution so here where at take one and at take two would have a cost of 45 rupees per question sold. We're basically solving it real time we're detecting images we're being able to look at, you know, what the question really is we're even being able to decide for learning content, and soon we'll be launching a feature where you choose the book and the engine is going to ask questions open ended questions regarding how much you understood the theory of that book and made this you know you can read more about that natural language platform on imbibe.com slash AI has become a backbone of a lot more research to come. So, so lesson from this is basically that you know it if you're starving, you know, for something truly disruptive, you have to make the time and space to make it happen. And this has happened over and over at imbibe because when you want to really solve hard problems first principles you have to make investments like these and you have to make your investors shareholders, and your team understand. Let's come to the next example intellectually fear so this is actually like, you know, very, very important. We all think they're data driven but correlation is not causation and I feel like you know we don't really go to the end the tail of the fish born and really sort of slice it and understand what what it truly means. So here's an example we've got a problem statement called adaptive practice. Now adaptive practice is one of the most like. I'm thinking of how to say this really politely, but it's a very facetious word for a lot of people, right, because they say adaptive practice but it's not truly adaptive because either they don't have the content to truly serve adaptively or they just collect three slices of data about the what they call it adaptive right so adaptive is as adaptive is adaptive can get. At imbibe it's quite adaptive so we have about you know 100 tags and each question and we capture like, you know, more than like a million parameters and every student and I could you know and you know dimensionally we can we can talk about it. So now figuring out that when you match that data collected with the content available how do you make sure that you serve content in the right way. So you also add in the, the, you know, an important sort of factor in this in this journey which is related to accessibility and engineering. Then you also understand that if you're solving for edtech for Bharat you're dealing with a range of devices a range of Internet connections. So there are many, many, many factors in a regression equation, which would give you the, you know, how much a person would end up practicing and how much would they end up improving. So, so a lot of cross functional problem solving happens, you know, so we've got the content team that will produce the questions that will supply the tags there's the data engineering team that will decide. All right. So what is the data that we're collecting from a session and what derived tags can be added on top so give you one example. And if you take it back to, you know, the way teachers typically teach in the classroom right so they're like all right who can solve this this question. And that question is a question that's in the mind of a teacher, you know, after a lot of understanding students and and their own content because they know this. This is a question that either nobody can solve or everybody can solve. So how do you actually get a machine to understand that so there's there's a parameter that we came up with called a question discrimination factor. And when you're intellectually fierce you always start with benchmarking so we looked at what Pearson is doing in that space and we've come up with a model that is 10x better than that. And that is just one derived parameter that also goes into the decision of what question to serve the student next. The other vector over here is also you know what device you're serving it on whether the UX is fine and the question loaded the CDN is working. And what we do is we've actually instrumented the crap out of everything that is being sent forward and we have these amazing brainstorming sessions where we talk from everything about academic intuition to, you know, content serving latency to uptime to whatever in the same room. So it's really fun because you have DevOps and all of that. So when we relaunched our product and the new avatar, you know, just give you some sense of like results from this kind of a brainstorming exercise across intellectual problem solving, like 4th to 10th October so we're just doing like test runs in the market we've not really started a lot of promotions. We have been nominated and Google play Google play users choice 2021 awards results are coming out soon fingers crossed category best for personal growth. So what here happened here was that, you know, we started with like a first time retention of 2.3% obviously when we pushed out our beta. It wasn't like the best, but the be amped up, you know, marketing and we also solve for retention simultaneously and retention right now is actually strengthening at about 6% there are some we have a lot of tongs in the fire to take it up further. The point is, if you're able to take a problem statement like retention or problem statement like does my adaptive practice truly add value. If you're able to numerically instrument the factors that, you know, that actually cause, not just correlate that number to go up or go down and, you know, split it up and then execute, you know, between teams while each team understands what the other team is up to and why. The result, you know, becomes that the whole is much, much greater than the sum of the parts. So this is just one example, you know, there are countless other examples of you know execution which works I mean sometimes people are like oh when traffic goes up time spent will go down. Not necessarily so not when you can put a finger on it. The next story is about frugally grounded. This is all of these things that I'm reading out to you guys are, you know, our values that we live by. I just found it really funny that Amazon has about 14 values and now I kind of understand because each of these has come from like some of the other lesson that we've learned executing over the last nine years. So here's what was happening right be wanted to build for Bharat. Okay. And while building for Bharat we also wanted to make sure that. So this is actually my reminder for the convention speech because I actually started early. So, so, so the question was, you know, we built this product and it's amazing and it's doing really well in English, you know how do you sort of launch it and multiple vernacular languages because in five is a product is like committed to inclusive growth. So we, you know, we wanted to solve the Hindi and empty problem and we went to a bunch of companies to asking for, you know, how much would it cost to translate our question bank because we were determined not to build a product that is just like a, you know, like a, like a skin of Hindi or like dubbing. We wanted to actually go deep redo our videos redo the text on screen and you know re render our content. We also wanted to make sure that our, you know, our friends and in the Hindi hinterland are able to get questions which are done right we did not want to use images because we wanted it still to be machine readable and establish parity in data between an English question and a Hindi question. So what happened was imbibe's data science team, you know, we acquired a company called online theory, you know, again very frugally, and the founders came on board and the challenge I gave them is let's build a Hindi and empty engine. So what I need to understand and why I call this ox octopus execution is Hindi and empty translation or translation as a product as a company in itself and and if you look at imbibe it's a platform and a platform that has an ecosystem of apis and solutions that could individually become billion dollar companies in themselves. So the execution is extremely extremely broad, and it's also extremely extremely deep. So, so they went around that challenge and they had two levers to bring down the cost first was to make sure that they built in the enemy with online theories label data, and we beat Google's blue score. And the second thing that they did was instead of outsourcing to a single supplier where overheads are managed, we actually set up like a crowd sourced way of doing translation. And the result was that we were able to bring down the cost per word to 1.5 rupees which was like 50% lower than the average market costs that we got. We beat Google's blue score for Ed and we ended up generating 36,000 hours of labeled data for voice and video which is now going to become a voice querying engine that we're going to be building. So, you know, I mean, sometimes people and investors ask you to prioritize that, you know, only do this much aren't you doing too much. I feel like if you have your finger on the pulse and you're clear about why you're doing stuff. It's fun to out execute if you're able to build a template to do so. So let's come to the final story, empathically strong. So, empathy is a very interesting word right I mean it, it, a lot of people say, you know, empathy is sort of, you know, work life balance and whatnot, I feel like when your vision led. And a vibe especially like people are encouraged to make their own choices about how they want to work or be or do whatever. But one thing is non negotiable that we live and die for the student. And remember when I set up this company. The first thought that was in my head was, like, why is there information asymmetry that exists. Basic very, very, you know, chicken shit information asymmetry that that holds people back from being their best self so I'll give you an example of my days back in the day. So I decided to go for the IT entrance exam and we used to be a desperate company now we're a lot about learning and and prep, and balancing that but it's a story for another day. So what happened to me was, I basically hated organic chemistry and I decided not to study it. And I was recently alright at physics, and I ended up getting a rank of about 5000 and IT. And the only reason for that is nobody came and told me, you know, tap, tap, tap on the shoulder. Hey, do you know that you need to study organic because no one's going to de scope the syllabus for you and here's the weightage. Sounds lame. Tier B city, but reasonably well off, you know, parents and, you know, can complain. So why, why was that exposure not existent or how was that missed. And just like that, there are many, many other examples of information asymmetry or just simple, simple things that hold people back, like something as ridiculous as addition of improper fractions. So let me tell you about what that what that story is all about. So I was in a classroom, and, and I was observing a teacher teach. That's been a lot of, you know, how invite was built, and that's where the first principles come come from. I was observing that teacher at the end of the class a student came with a doubt, you know, to the teacher saying sir, can you please help me with this. And the teacher, you know, just dismissed him and said, don't send me hope I got, and he just did this he handbaved you know and said, you're not you're not going to be able to do this. So I went up to him later I said so please don't mind me saying so but the guy's paying you like $2,000 a year like why are you discouraging him like why wouldn't you improve him. So he said, Oh, you know, I can't because he doesn't even understand fifth grade concepts, and he's trying to crack it I said what exactly what what fifth grade concept is he not getting he says, he doesn't understand how to add improper fractions he thinks one by two plus one by three is two by five. I was doing my math around what that is. But it is definitely not two by five. So I was like, Alright, so but then why don't you just teach him that he says, How many kids do I end up teaching, you know, one by two plus one by three is two by five and this is not the only problem there are many many problems. So then I realized I was like holy shit. What's happening in education is that you basically have this this massive concepts that you have to navigate that is slotted into yours. Every year when you pass, you know, you pass with what you have. Nobody can take you back to fix what you left behind. And in, and it's like a recursive problem because every successive year, your debt keeps piling up kind of like technical debt and engineering. And as a result of that, like, unless and until somebody goes back and fix you basically setting yourself up for diminishing returns have no matter how much effort you put in this to study. So, just to give you an insight around, you know how empathy works at imbibe like I couldn't get that thing out of my head and we went back and be brainstormed as a team saying, Wait, you know, we set out in 2012, you know, September to really build something that that removes information asymmetry for education that uses data. And this was like, we were the first company to look at data and education actually. So, so that is the case, you know, how do we sort of this there seems to be something here which is a very fundamental truth, kind of like, you know, I mean I feel like a lot of platforms are based on very elegant problems like something like hey, where are my old friends was Facebook and LinkedIn is like who's in the next office, you know. Google is like, where can I find what I need. So for imbibe you're like you know, can the mainstay of our data platform basically be, you know, all learning is connected. So, so fix what you need to fix. And all learning is connected. So then the question was how do you actually manifest that connection. So we just sat about doing something very simple in 2014 I sat with, you know, a bunch of teachers and said how do you teach what's the problem. And we built sequences for every chapter, simple two dimensional connections, and we connected them back to previous years. That effort has now become, you know, our knowledge graph 50% average improvement is what it delivers, you know, through the algorithms that run off of it. 1,000 concepts 1.85 lakh or 185,000 competencies and 130 1000 topics. And if you what you see on the left inside is not a galaxy it's actually k 12 learning. So it's one big dense, you know, like galaxy of concepts and you know somebody just came and drew lines of chalk in the middle and said, you know you shall be grade one grade to grade three. So that's how it was happening the government of India came and said, you know, let's do the new education policy and let's talk about great less learning. So this seven years of investment and understanding and there were so many insights that came out of it further so for example, you know, what started as a two dimensional graph became a three dimensional graph because depth of a concept is another thing that came in. But you know if you're very careful about listening to insights or listening to things that your customers say or listening to things that your team says, and really sort of, you know, structuring it in a prioritized kind of a way and and putting an honest effort things will come out. There's not always, you know, not a wild goose chase because the knowledge graph was a wild goose chase for a bit it just felt right in my gut to get this done. So again had to take investors shareholders everybody along to continue to fund this activity to create base data. I feel like a lot of, you know, startups in India have data as retrospective thought, it's not part of the core engine of how they build their companies and therefore, you know, often more often than not, you know, when you execute the first time around you don't build in the hooks that you have a continuous first mover advantage because of the information you store and how you learn from that information. So I feel like that's another, you know, really important thing that is worth understanding. So, empathy is also about, you know, how you sort of think about execution 360 degrees. So there are a lot of little nuggets like you know the way we baked it, you know, baked it in into our products so for example, for a student you know we've taken like 1500 books from the market aligned or learning content to it because for us. It's very important to collect as much data as we can so be like how do you make the consumption absolutely easy and intuitive and not sort of aligned to, you know, and not sort of aligned to just what do you call, you know, my way of learning because that's what I'm saying. For investors like, you know, delivering, we have a huge vision. So obviously we have to make sure that we're very frugal in how we deliver it. We have to make sure that our platform is super scalable that it is not just something which is an in the only wonder because education is a scalable problem for teachers, you know, everybody is building like they're so it's interesting like the interest in the market is vernacular and about the same percentage is sitting in the rural India and the public sector. People are trying to recycle the same content that did for the consumer market for you know the 10 to 20 million privileged Indian audiences. And they're trying to sort of push that down teachers' throats and they don't understand something very basic that teachers like their own presence. And that was for example, we took all of our video content and be extracted assets like 50,000 3D assets and you know we also created versions of our content which had anchors and no anchors so that the teacher can be the star of the classroom. And we're making engines where teachers can build powerful lessons with minimal or no effort, and then at the same time be at the center of their class building that social emotional connect so that more and more teachers can become amazing teachers, and more and more people can become teachers even if they're not because imbibed is all of the hard work to understand the student and to build and develop the content. So for our people, we're flat. I don't understand hierarchy I don't understand designations most of the time. The, our youngest VP is 23. And there is about 40% of our media content has been built by people who are specially able to, you know, and in fact a case study of that was featured in Satya Mave Jayate as well. So this, you know, the, the organization at the end of the day is an organism I'm just going to cycle back to the very beginning, and, you know, bring back the octopus right. A lot of people feel like. So this is reminder for my thyroid medicine now you guys know a lot about me. The organization at the end of the day is an organism, and it's it's living and breathing and learning and adapting to its environment. And I feel like, you know, if you're true to your purpose and your vision led a lot of the blueprint sort of kind of self defines itself provided you have the right kind of tools that you use to sort of, you know, process and respond to the the feedback that you're getting. And our values are our tools. And I feel like they've constantly led us in the right direction and hopefully you guys will hear a lot more from us from the market, you know, in the months to come. So this is it. This is, you know, four examples of octopus execution. There is, you know, one was about setting aside things for future priorities being vision led the second story we talked about was being intellectually fierce and cross and the problem solving third was frugally grounded, you know, we could have just very easily gone and outsourced the contract for the value that we were getting at our investors that have been convinced because there was no other, you know, competing code, but we cut down the cost by 50% because we realize the level of adaptation of the index that we wanted to do. And, you know, for 11 languages, it would just have been completely non scalable. Because again, you know, investors also want profitability. And then at the end, you know, we talked about the Knowledge Graph, which is the heart of our IP that also came from deep empathy and listening to our customers and their interactions with and our stakeholders and their interactions with each other. So this is it. I'm open for questions now so I'm going to stop sharing, and I'm back on the screen. So I see we've got one, we've got one question for you at the moment, DT. Thanks, DT for this wonderful insight into imbibe. Can you share something that you feel is beyond what you consider as this is the cutting edge the way you see it now. We'd love to see what we can expect five to 10 years in the field of education from now, casting a vision for the future. So I really hope that we are not in the metaverse. Let me just say that. Because I believe in playgrounds and schools and hugs and you know, kids sort of engaging with everybody. But you know that that aside, I feel like one of the things that we're very, very committed to to our, you know, our vision for data and what we're doing is being the the instrument to abolish the word syllabus from education. So and I feel like we're very close to it because a lot of people are making noises about there are too many, you know, curriculums and learning but so far no one has had the the data or the syllabi, you know, to be able to the data and the, you know, and and sort of this great list learning backbone to be able to truly go to governments and get them to commit to that kind of an agenda. We're also a global innovator with the World Economic Forum and I can tell you that there are many, many conversations underway right now to figure out a unified quote unquote syllabus for the world, which doesn't have grades. What we believe is that, and it's again like you know the stuff that we're doing and which is why we built the data platform in the first place. The second thing is about bringing sort of course skilling to early childhood on and sort of, and sort of blur the lines between skilling and schooling, because I mean, at the end of the day there are certain foundational skills like quant for example, that Lincoln very nicely with certain functional skills like for example data science, you know, there are concepts like angular momentum which if you get that you're probably a great abstract thinkers who's stopping you from learning Python and going for it. And I feel like it's, you know, it there will be a time where skilling will stop being gimmicky for kids and it'll become a lot more coherent. So, so I've seen these two things and the third thing obviously is like with deeper penetration of high end devices I feel like a lot more imagination is going to be ignited for kids. And a quick story again from a classroom in Rajasthan where we work with tribal students. We got them to improve in the syllabus by atomic in an atomic structure by about 48%, but when we actually ended up having a conversation, you know about it with the person who scored the highest marks. We turned out that it turned out that she didn't believe atoms and molecules really existed and I at that point of time I showed them YouTube and they were like, what, because they've never seen YouTube. And there are classrooms like that still today in the world. So, so, so this is what I believe I feel like zero syllabi, you know, no line between skilling and learning in school. And so lifelong learning and that mean I hate segments and that tech in general because you know I'm a proponent of continuous learning and all learning being connected. And the third thing is, you know, a lot more, a lot more fertile imagination, I feel, because you'll have kids really seeing a lot more things come to life with deeper device and connectivity proliferation. Wonderful. Next question we have for you is from Sanjay Gupta. Many of us are an adult learning. Is there any learning you have that is applicable for adult learning in a corporate context. So one of the things that we want to do Sanjay is that we want to actually take or several courses and their skill graph and connected with ours and build it I don't. We don't, we haven't really formally taken on the curriculum per se around adult learning. But the first thing that we are doing is now we're getting into social emotional learning and early childhood. I'm going to take a second and tell you about our early childhood lessons because I'm super excited about them. So what we're doing there is that again, you know you have Montessori and CBSC and all of these boards and we said you know what, a lot of people are not able to afford nursery education and grade one assumes prerequisites you know it's just it's just a thing all over the world so people who don't have money and they put this kids in school typically get started at day one and they're disadvantaged in grade one because they haven't done nursery education. So what we're doing is we're saying that how do you kind of build a story based learning where you connect social emotional learning and lessons in one single, you know, experience so for example, the friendship between a mouse and an elephant learning and learning about water pollution, while that's happening so you also learn about friendship. So, and my lessons basically are the intersection of skills, which in this case is social emotional and curriculum, and then figuring out the data and instrumentation around how you consume it and going super deep and interactive learning. So, when we do get end up getting around to adult learning we will probably end up going and partnering with entities who have good content in the space in a platform format, and be using our personalization algorithms to drive that. So I think that the other opportunity for us in this is to build a link between, you know, your childhood foundations and adult learning because once again if you do an accounting course and things like that. A lot of the fundamentals you know come from reading and arithmetic. If you look at grade one employability, right and advanced courses there's great custom great companies who are working on that. The next question we have here is, what's your take on replacing formal education using imbibe and parents opting for full time home education. My take is it can happen. It should happen. I think I think a lot of it is fear. I feel like you should still I mean I believe that the. One thing that we're doing that you guys should understand is that we're not just sort of like a B2C company that is only focused on student learning, you know, at home, and my business is a platform so we power students we power schools and teachers and be power parents. So we believe in facilitating the dialogue between these three and empowering them with the right amount of content and data to be able to infuse the best in the child. So, so I think, but at the same time there are so for example today, we have about 130 villages that are live on imbibe and every day we're adding hundreds more. And these are villages that don't have a school within like 10 kilometers of each other, you know, so, I mean 10 kilometers vicinity with around them so there's no, there's no option. Right so so yes it can happen. I mean, if you get started on imbibe and grade one and grade one's coming soon to grade six you can go on to grade 12. Excellent, we've got questions coming clicking fast now. There's a large population in India that is education deprived due to financial issues. How do you think that they can leverage from imbibe because they might not even have mobile devices that's from unkept. So, so so our parent company geo is solving that problem the device problem and in terms of access and financial, you know relevance you can count on us to not do, you know, to do the right thing, let's put it this way. So I gave you a stat earlier on like and you know the writings on the wall around that I mean nine, you know, 99% of the population can afford above 2000 rupees in education fee in school, and it takes a charging like $2,000, an average fee and, you know, on an individualized basis, and 50% of education companies are actually, you know, dispersing student loans to do ed tech courses, of which a further 50% of people don't know that they took a loan. So we're very committed to good practices and business and, you know, right now it's free. And it will be for some time. So, you know, we're, we're getting a lot of will take care of that. Another question here if I decide to become an architect at the age of 10, can I have all the relevant focus course which will help me. DR and cut the crap for the rest of what is not relevant. I love the cut that we want to cut the crap that's the whole point right. But the question would be that at the age of 10 have you navigated the prerequisite learning for architecture. So that mapping is you know, what you would need to do so if you were to do it in vibe let's say if you were to add let's say a skill course in architecture, we would map your fundamental skills and how you discover how much how proficient you are in them to be able to decide, you know, how you end up becoming whether you're ready to do architecture courses or not. See the whole answer and the holy grail here is tagging. Nobody wants to do that it's dirty work. Right. But the foundation of building AI is to do that dirty work and to tag and to you know map and all of that stuff. Like for the knowledge graph we have every single question we create we have 100 tags then we go back and recalibrate them. So, I mean, if I just read this line in the eyes of a product person I have like this whole Gantt chart in my head around how to get there, you know. So, so yeah, so I guess you will be able to cut the crap but it depends on what you've learned to lead to 10. Yeah, it's partly bear charting a course towards that from the age of 10. 3g has a question how the tablets mobiles availability challenge so I think you've already kind of covered that one. Your answer was Geo. Affordability of devices. How will the initiative of grade list learning aligned to exams at the end of the day that students in India have to pass to move ahead in their education careers. I think what's going to happen is the chicken and egg problem. So if we don't do exams, we become irrelevant today. If we don't agree less learning we become irrelevant tomorrow. So that's why our knowledge graph is mapped to 345 exams so you can query it from an exam basis but you can float around we don't stop you and your improvement journeys your achievement journeys that I was talking about agreed less. So you actually have two tests that you have to take in the beginning if you want to master grade 10 for example. The first test is prerequisite base so it's everything before grade 10 it's related to grade 10, right, which is invisible to you and then only grade 10. So, so I think the answer is like post building the platform which we have and then campaign and policy. And you have to sort of design for today so you can be so that's an interesting thing because you know you could build a very idealistic solution but it would not find acceptance and and uptake because the world is talking a different language. And so you've got to build both. So that's where your platform helps you because you can do whatever you want with the data right and if your data is aligned to great less learning your future proof. All right, Sunday has a question what are your thoughts on the importance of live interactions between student and teacher. I think it's extremely important. And I think that it's super important that tech companies take the responsibility to actually do something about it which is more than just a zoom conference. Right, because if you look at like all of the tech companies that are doing live teaching today, the platform adds no value so our school platform which I'm also very excited about solves two problems. The first is we have an API for student intelligence so when a teacher starts the class they'll be able to see are the kids ready for this lesson or not, and send a one click playlist for them to improve for their prerequisites. So you can actually sort of have, you know, you can search for 3d assets on the fly on your lesson you can search for you know contextual question saying, hey, I want to teach human heart, show me a 3d model right. And in fact, you know I could maybe demo it in the round table if you guys are interested in seeing it. I want to understand, you know how it works or I want an easy question on a human heart. And our platform has already got that recommended content ready and available. You can also go and create a, you know, a PowerPoint in the cloud kind of lesson using our assets, and then stream that over the lecture. So, we add the content we bring the student intelligence and you bring the emotional connect and the interaction. Right. So that's what how much we support live teaching that we read on our entire content stack to be able to support the teacher in the classroom. Wonderful. We just about a time but we'll squeeze in one more question. How do you keep discovering the problem to be solved and keep refining those problems as Chandan is asking. So I think it's it's important again to keep looking at the data that you that you collect, because a lot of times you know you would collect a lot of data but then you would not surface it and then you wouldn't talk about it. So, for example, you start with something very simple right like we have only two problems engagement and learning outcomes if you have engagement you have data that means students are committed. The content is worth it and learning outcomes are the sequence of content that you should give to these students and whether that feedback loop works or not. Okay, so we have dashboards that we've got a data center of excellence in house you got dashboards that we continuously keep looking at you know the data in our cross functional scrums and I keep pushing the team on it that's what being intellectually fierce is about. And then we come up with problem statements in these sessions and then we execute and we collect data again and then we keep going. We're going to have to wrap things up there sorry people who've got other questions there but the good news is that a DT is going to be in the room that the hangout room after this to take questions so once we finish the session, head on over there and jump in I have a feeling there's going to be quite a few people trying to get at that table so being quick. Before we go I've got a couple of things that I need to do to fill you in on some things. Thank you so much though for that it was a really really interesting session, lots of interesting insights and really interesting vision for the future of education and ways to use artificial intelligence and all that platform that you're building. It sounds like exciting times are huge. Thanks for your time ODD.