 Picture your brain learning, new memories created, memories erased, new concepts established, dopamine released. Your brain's hundred billion neurons forming hundred trillion connections. How would you capture that in a model? How would you explain it to a computer? This question lies at the heart of the education revolution. Because if you can understand how people learn, you can also partialize content to them, and the impact of partialized education is extraordinary. In one study by the Bill and Melinda Gates Foundation, students using partialized techniques had 50% better learning outcomes in a year, meaning that over a 12-year period due to compounding, these students would learn a hundred times more. Imagine if we could empower every single student globally to learn a hundred times more. Imagine the revolution, the renaissance that this could catalyze. So you might be asking, why has this not been done before? Why was this not done five or ten years ago? In fact, attempts have been made for almost a century. Because since Gutenberg invented the printing press, little has changed in how we share and consume knowledge. We still, to a very large degree, rely on a one-size-fits-all model. A model where we present the very same content in the very same way to all students. However, Sydney Pussy invented this teaching machine. Essentially, it was an automaton with a paper drum displaying multiple choice questions. Allowing students to tackle questions at their own pace. An automaton even had this feature, where if you would get an answer correctly, you were rewarded with sweets. However, despite of its sugar-coated bait, Sydney Pussy's teaching machine did not live up to the hype. The subsequent introduction of the Macintosh promised smarter ways to personalize. Suddenly, you could introduce a vast number of rules and heuristics. You could tell the computer if a student does X, do Y. Unfortunately, the intricacies of the human brain, the intricacies of how we learn, was simply too complex to capture in rules. So the Macintosh followed the trend of most such technologies before it. We now live in a very special time, a rare point in history. With the confluence of fields is enabling personalization to a whole new extent and at the whole new scale. These fields are online education, allowing us to collect immense data sets and neural networks, allowing us to learn directly from that data. At the intersection, we have the next great education revolution, the next printing press. Because the power of neural networks lies in their ability to learn directly from raw data, meaning that in difference from the Macintosh and Sydney Pussy's teaching machine, it allows computers to learn by themselves. You give the neural network a set of inputs and then you tell it the expected output. If it gets the output incorrectly, it readjusts its calculations iteratively until it can accurately predict the output. So you might ask it, for example, to predict how a student would answer a certain question. You give it a set of inputs and a damn tries to predict how a student would answer that question. If it fills, it readjusts and refines hundreds of millions of parameters until it can accurately predict the student's answer. So you might all be old enough to remember the game of Atari and Breakout. So we quite recently saw this game directly from pixel inputs, meaning that the pixels on the screen and the running score was the only data that the algorithm received. So it had to figure out for itself what the rules of the game were and how to score points. Initially, its attempts were random. However, over time, it figured out an optimal strategy, digging a tunnel behind the bricks with superhuman accuracy, allowing the ball to stick there and eliminate all of the bricks. So imagine if we could develop a similar system for education. If we could develop a system which would, with superhuman accuracy, shoot the optimal explanation at each student, they would find that tunnel behind the bricks that keeps the students engaged and curious, optimizing their learning in real time. At the company I founded, that's the vision that me and my team of researchers and engineers are realizing. We combined the MANS data sets from online education with neural networks to personalize education for every single student globally. Neural networks allows us transforming education for over 70 million students. We recommend content based on how they learn, what they've already mastered, and what works best for similar students. Neural networks allows us to understand whether you're engaged, confused, or simply just bored. Neural networks allows us to identify factors or indicators of successful learning outcomes which was previously impossible to capture. In essence, we give our neural network a set of inputs, how long it took for you to answer certain questions, how many passes you took throughout the lessons, and the number of attempts you made. Using these inputs, it tries to predict future interactions. The strategy layer then calculates which future outcome that would maximize your learning outcome over time. So in a nutshell, it predicts the future and then it selects the future that it likes the best. So let's take an example, your learning Spanish through an app. In real time, the neural network would receive historical data. It would then make a set of predictions such as what you would answer, how you would answer certain questions, how long it would take you, and which content you would find the most engaging. The strategy layer would then calculate which pieces of content that are most likely to maximize your engagement over time and recommend it to you. So in real time, based on your preferences, your abilities, and your strengths and weaknesses, you get partialised recommendations. SANA optimizes the next step for you in real time until you've successfully mastered the whole course. By providing our platform, our ambition is to transform every single learning product into a superhuman teaching system. A system which finds your knowledge gaps and implements strategies to address them. A system which models your learning style and makes learning any concept infinitely more engaging. Completely content agnostic, our technology is, in itself, truly universal and can be applied to any subject. We work with products for doctors, allowing a doctor in London to learn and respond to any problem. We work with doctors in London to learn and retain knowledge more efficiently. For schools, allowing a student from New Delhi to Sydney to succeed in STEM subjects like maths. And for programming, allowing any student globally to learn to program in a partialised way. Together we're driving forward the future of medicine, technology, communication and innovation. Fundamentally improving the function of an entire five trillion education industry that directly affects billions of people's lives every single day. Thank you.