Here is our article on Medium about EdTech and how education will change in the near future 👇🏻
I have three children. The youngest is just 2yo and currently considers throwing food around a form of creative self expression. The next one just turned recently 6yo and questions almost everything , a habit I’ve tried to keep as I think is fantastic. And then I have a 7yo who is way more reflective and is starting to enter this kind of fun “pre-teenage” stage, with new feelings coming out constantly and a humor that changes from 0 to 100 without clear explanations.
As I guess happens to any of you in a similar situation, I think about their future a lot, not in an anxious way but in the sense that anyone paying attention to what’s happening in education technology right now cannot help but feel that something truly big is coming. And this something is not coming in decades, but in our opinion it will happen along the next five to ten years, literally within the actual school careers of children who are now under ten years old.
As a parent who also watches this change and this space professionally, and who actually enjoys history in general, I figure out things that can make anyone stop and think. To go to a historical example, I think it’s worth sitting with one potentially uncomfortable fact. France’s baccalauréat (still one of the most used university entrance qualifications in the world ) was created by Napoleon back in 1808. This standardized school model that so many western countries run on was designed literally during the Industrial Revolution with a very specific purpose: to produce reliable, consistent workers for factories and administrative roles. It was a system that basically told kids: Sit still. Follow instructions. Reproduce what you were told. Pass the test.
Even in our days, nobody hesitates that his model worked extraordinarily well for over a century, and it is actually right now, in 2026, producing students who are being evaluated for jobs that don’t exist yet, by testing skills that are increasingly automated, using methods designed for a world that ended roughly fifteen years ago.
Even the World Economic Forum has put it plainly quite recently , stating that most secondary education systems are still built to optimize for standardized metrics that reward memorization, individual performance and technical accuracy , which are precisely the skills being automated fastest.
The scary thing is that no single child alive today will spend their working life doing the thing their school system was primarily designed to train them for, and that is the baseline from which everything interesting follows, and a crazy number that stopped me when I first read it is that according to a 2025 Microsoft report, 86% of education organizations worldwide now use generative AI , being the highest adoption rate of any industry. A sector that historically has been one of the slowest to adopt new technology has, in the space of about three years, become the leading industry for AI integration. Insane.
But in any case, it’s also true that the scale at which this is already happening is easy to underestimate. Khan Academy’s AI tutor, Khanmigo, went from roughly 68.000 users in partner school districts in 2024 to more than 700.000 in 2024/25, expanding from 45 to more than 380 district partners in a single academic year. This is definitely not a pilot program scaling tentatively but a technology crossing the adoption threshold and accelerating really fast. Khan Academy ran a strong 6 month testing program from October 2025 to April 2026 measuring Khanmigo’s effectiveness, iterating on outcomes in real time in ways no textbook publisher has ever been able to do.
What makes Khanmigo specially interesting is its design philosophy, because it was actually and deliberately built to never give students direct answers. Instead, his system uses the Socratic method of guiding learners through questions and hints until they reach the answer themselves. This is not a trivial design choice, but is the plain and clear difference between a tool that replaces thinking and a tool that trains it, and it is the model that the most thoughtful edtech companies are converging on.
The global AI in education market sit at approx 7 billion USD in 2025, with not exaggerated projections putting it at close to 136 billion by 2035, growing at a compound annual growth rate of over 34%. To put that another way, we can say that the market is expected to be roughly twenty times its current size within the school careers of children who are starting primary school right now.
Think about what private tutoring does that a classroom of thirty children cannot. It adjusts pace to the individual, it identifies the specific misconception a child has and addresses exactly that, rather than moving the whole class forward regardless. It responds to engagement , noticing when a student’s attention has wandered and changing approach accordingly. It celebrates progress in the specific areas where that student needed to grow, not just the areas the curriculum deemed important this term.
Private tutoring that does these things has historically cost big sums of money per hour. It has been, overwhelmingly, the preserve of families who can afford it and a mechanism that compounds educational inequality rather than reducing it, but AI changes this equation structurally. Khanmigo, for example, costs just 4 USD per month, offering a meaningful fraction of the value of private tutoring at roughly 1% of the price. Scaled across an entire school system, that access gap (which has distorted educational outcomes for generations ) begins to close.
But the deeper change is what becomes possible when personalization moves beyond tutoring into the design of education itself. Today’s most ambitious edtech platforms don’t just adjust pace but instead they identify learning styles. They notice that one child engages more deeply when math problems are framed around football rather than abstract numbers. They track which type of explanation worked and which didn’t, and adjust automatically. They connect topics across disciplines in ways a single teacher managing thirty different children cannot.
The OECD’s Learning Compass 2030 framework describes the direction this is heading. They describe education systems that equip students not just to acquire knowledge, but to think critically, act judiciously and navigate real world challenges with a genuine opinion. Not what to think but HOW to think.
But despite of all of this, and looking at the old system, we should also remember that the things that made great teachers great were never primarily about information delivery. A great teacher noticed when a child was struggling emotionally, not just academically. They created the psychological safety to make mistakes without shame. They modeled intellectual curiosity by being genuinely excited about ideas, and they connected with children as humans first, as students second , creating a relationship that was often what made learning feel safe enough to actually happen.
No AI system is good at empathy. Machines are genuinely bad at navigating messy, unstructured social situations, at inspiring a child who has given up and at reading the room in a way that changes what happens next. Emotional awareness, ethical reasoning or collaborative problem solving all remain distinctly human capabilities, and they are, not coincidentally, exactly the capabilities that will be most economically valuable in a world where AI handles the rest.
Project based learning, which the Harvard Graduate School of Education has shown increases student engagement and produces deeper understanding across disciplines, is not primarily a technology story but is much more of a pedagogy story. A lemonade stand, as one education researcher memorably put it, teaches entrepreneurship, customer service and resilience better than any classroom worksheet. These things need to be protected, amplified and placed at the center of what schools do , not squeezed to the margins as AI handles the academic content more efficiently.
The children who will thrive in 2040 will not be the ones who know the most facts. They will be the ones who can connect disparate ideas, understand and motivate people, navigate uncertainty with confidence and create things that didn’t exist before. The curriculum of the future needs to build for that outcome, and the technology can help get there only if it is used to augment human development rather than replace it.
And from the business perspective, for anyone building or investing in education technology, you all must know that the gap between what is now technically possible and what is actually deployed in schools is enormous, and that gap represents one of the most significant market opportunities of the next decade.
The edtech market is projected to exceed the gigantic figure of 1 trillion by 2030. Gartner projects that 40% of enterprise applications will be integrated with task specific AI agents by the end of 2026 , and education is not exempt from that trajectory. Microsoft committed more than 4 billion to AI education initiatives in July 2025 alone, targeting schools, community colleges and nonprofits through its newly launched “Microsoft Elevate Academy”.
But the companies that will matter most in this space are not the ones that bolt AI onto existing systems and call it innovation. They are the ones that understand pedagogy deeply enough to use AI in service of genuine learning outcomes. The ones that know the difference between a tool that tests comprehension and a tool that builds it, and the ones that can serve both the child in a well funded private school in Madrid and the child in an underserved public school in rural India, because equitable access is not just an ethical imperative, it is simply the market.
Specialized education technology companies (he ones that sit at the intersection of learning science, data engineering and product design ) are positioned to do something genuinely important here. The technology is mature enough to deploy. The institutional willingness to adopt has arrived faster than anyone expected. The regulatory environment is beginning to develop frameworks. What remains is the translation layer, basically the work of turning powerful general purpose AI into education products that produce outcomes that parents, teachers, regulators and most importantly children can trust.
My 2yo daughter is currently just learning that cause and effect is a real thing . She will start school around 2028 and she will likely be in the workforce by around 2045. The jobs she will do in 2045, with big probability do not yet exist, the technology she will use routinely was not imaginable five years ago, the skills that will matter for her (creativity, adaptability, critical thinking, emotional intelligence or the ability to collaborate across cultures and disciplines ) are not well measured by any standardized test currently in widespread use.
The education system that serves her best will be one that uses the most powerful tools available to personalize her learning, free her teachers to do the human things only humans can do and keep genuine curiosity at the center of everything.
That whole system is being built right now. It is not finished and is not evenly distributed, but the pieces are moving faster than most people realize.
And we should realise that parents have always wanted the same thing: an education that sees their child as an individual, not a student number in a big classroom.
And for the first time in the history of mass education, technology is making that actually possible at scale, something that is worth getting excited about.
Top comments (0)