I’m going to tell you something most entrepreneurs won’t. My first EdTech startup was a complete and utter failure. Five years of my life, gone. We raised money, built a product, and even had some early traction. But in the end, it all came crashing down. And the reason it failed is probably not what you think.
Everyone is talking about AI in education as if it’s some kind of magic wand. Wave it over a textbook, and suddenly every student is a genius. I bought into that hype. I thought I could build an AI-powered platform that would revolutionize learning. I was wrong. So, here’s the real story, the one you won’t read in the headlines. This is what I learned from my five-year-long, multi-million dollar mistake.
The Allure of the AI Revolution
Back in 2018, the air in Silicon Valley was thick with the promise of AI. It felt like we were on the cusp of a new technological era, and I wanted to be a part of it. I’d had two successful exits, one in remote work and another in media, but education was always a passion of mine. It seemed like the perfect place to apply the power of AI. The idea was simple: create a personalized learning platform that could adapt to each student’s individual needs. We would use sophisticated algorithms to analyze their performance, identify their weaknesses, and deliver customized content to help them improve. It was a beautiful vision, and I was completely captivated by it.
We assembled a brilliant team of engineers and data scientists. We spent months, which turned into years, building our platform. We called it ‘Cognify’. The tech was impressive. We had a sleek user interface, a powerful recommendation engine, and a dashboard that provided teachers with a firehose of data. We were sure we had a winner on our hands. We were so focused on the ‘what’ – the technology – that we barely stopped to think about the ‘how’. How would this actually work in a real classroom? How would teachers use it? And most importantly, how would it actually help students learn?
Lesson 1: The Human Element is Non-Negotiable
Our first major mistake was believing that technology could replace the human element in education. We were so enamored with our own creation that we forgot about the people who would actually be using it. We’d show up to schools with our fancy presentations and our impressive demos, and we’d talk about how our AI was going to change everything. The teachers would nod along, but I could see the skepticism in their eyes. They’d ask questions like, “But how will I have time to look at all this data?” or “What if a student is having a bad day? Will the AI know that?”
I remember one teacher, a veteran with over 20 years of experience, who pulled me aside after a presentation. She said, “Your platform is very impressive, but you’re forgetting something important. A student is not a data point. A student is a person. They have hopes, fears, and dreams. They have good days and bad days. Your AI can’t see that. Only a human can.”
At the time, I dismissed her concerns as resistance to change. I thought she just didn’t ‘get it’. But she was right. We were so focused on optimizing for engagement metrics and completion rates that we lost sight of the bigger picture. We were trying to solve a human problem with a purely technological solution. And it was never going to work.
Here’s a concrete example. Our platform would flag students who were struggling with a particular concept. It would then recommend a series of exercises and videos to help them catch up. On paper, it was a great idea. But in reality, it was a disaster. The students felt like they were being punished for not understanding something. They would get stuck in a loop of remedial exercises, and their confidence would plummet. They started to hate learning. We had created a system that was designed to help them, but it was actually making things worse.
Lesson 2: The 'Minimum Viable AI' is More Powerful Than a Bloated Platform
Our second mistake was trying to do too much. We were so excited about the possibilities of AI that we just kept adding features. We had a personalized learning path generator, a real-time progress tracker, a gamified rewards system, a social learning module, and a dozen other things. We thought that the more features we had, the more valuable our platform would be. We were wrong again.
All those features made our platform bloated, confusing, and difficult to use. Teachers were overwhelmed. Students were distracted. And the core value proposition – personalized learning – was getting lost in the noise. We were so busy building new features that we didn't have time to perfect the ones we already had. The result was a platform that was a mile wide and an inch deep.
I remember a meeting with a potential investor. He was a seasoned EdTech veteran, and he asked me a simple question: “What is the one thing that your platform does better than anyone else?” I started to list off all our features, and he cut me off. “No,” he said, “I asked for one thing. Not ten.” I didn’t have a good answer. And that’s when I knew we were in trouble.
If I were to do it all over again, I would focus on what I now call the ‘Minimum Viable AI’. Instead of trying to build a platform that does everything, I would focus on doing one thing exceptionally well. For example, instead of trying to create a fully autonomous personalized learning system, I would build a tool that helps teachers identify at-risk students and provides them with actionable insights. The AI would do the heavy lifting in the background, but the teacher would still be in control. It’s not about replacing the teacher; it’s about augmenting their abilities.
Lesson 3: Pedagogy, Not Technology, Drives Learning
This was the hardest lesson for me to learn. As a technologist, I was convinced that technology was the answer to everything. I believed that if we just built a powerful enough AI, we could solve all the problems in education. I was so focused on the technology that I completely ignored the pedagogy – the art and science of teaching.
We didn’t have a single educator on our founding team. We had engineers, data scientists, and product managers, but no one who had ever actually taught in a classroom. We thought we could just read a few books on education and figure it out. It was the height of arrogance, and it was our downfall.
We made so many basic pedagogical mistakes. For example, we assumed that all students learn in the same way. We created a one-size-fits-all learning path that was supposed to work for everyone. But of course, it didn’t. Some students learn best by reading, others by watching videos, and still others by doing hands-on projects. Our platform didn’t account for any of that.
We also failed to understand the importance of intrinsic motivation. We tried to motivate students with points, badges, and leaderboards, but it was all superficial. The students would game the system to earn rewards, but they weren’t actually learning anything. They were just going through the motions. We had created a system that was optimized for engagement, not for learning.
It wasn’t until we were on the brink of collapse that I finally started talking to real educators. I spent weeks visiting schools, observing classrooms, and interviewing teachers. And I learned more in those few weeks than I had in the previous five years. I learned that education is not a technology problem. It’s a human problem. And it requires a human solution.
The Phoenix from the Ashes
So, what happened to Cognify? It died a slow and painful death. We ran out of money, and we had to shut it down. It was one of the most difficult experiences of my life. But it was also one of the most valuable.
My failure with Cognify taught me that AI is not a silver bullet. It’s a powerful tool, but it’s just a tool. And like any tool, it can be used for good or for ill. The key is to start with the human element, not the technology. It’s about understanding the needs of teachers and students and then designing solutions that meet those needs.
My new venture in the EdTech space is built on this philosophy. We’re still using AI, but we’re using it in a much more thoughtful and deliberate way. We’re not trying to replace teachers; we’re trying to empower them. We’re not trying to create a one-size-fits-all solution; we’re creating a flexible platform that can be adapted to the unique needs of each classroom.
It’s still early days, but the results have been promising. We’re seeing higher levels of student engagement, improved learning outcomes, and, most importantly, happier teachers. It’s a long road ahead, but I’m confident that this time, we’re on the right path. My five-year failure was a painful but necessary lesson. It taught me that in the world of education, the heart is just as important as the algorithm.
Frequently Asked Questions
How has this view evolved over time?
My thinking on most topics has changed significantly over the years. Early in my career, I held many conventional views that experience proved wrong. I try to update my beliefs when the evidence changes.
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.