An MBA is great for learning how to talk about business. Building a product that 50,000 people actually use? That’s a different education entirely. That’s what happened when I built an AI-powered personalized learning platform. We scaled it to 50,000 students, and the lessons I learned in the trenches were more valuable than any case study I ever read in a classroom.
People throw around terms like “personalized learning” and “AI in education” like they’re magic bullets. They’re not. They are incredibly powerful tools, but they are just that—tools. And like any tool, you can use them to build something amazing or you can completely mess it up. I’m here to share the real story, the stuff they don’t teach you in school. Here are the seven biggest lessons I learned.
1. Personalization is a Data Problem, Not an Algorithm Problem
Everyone gets obsessed with the algorithm. They want to build the most complex, sophisticated, multi-layered neural network to predict what a student needs to learn next. I get it. It’s the sexy part of the problem. But here’s the truth: your fancy algorithm is useless without good data. And not just a lot of data, but the right data.
When we first started, we were tracking everything. Every click, every answer, every second spent on a page. We had mountains of data. But it wasn’t telling us what we really needed to know. It was just noise. We were drowning in data but starving for insights.
The turning point was when we stopped thinking about data as just a record of what happened and started thinking about it as a conversation with the user. We started asking better questions. Instead of just tracking if a student got an answer right or wrong, we started tracking how they got it wrong. Was it a conceptual error? A calculation mistake? A simple typo? We built in feedback loops where students could tell us why they were struggling. That qualitative data, combined with the quantitative data, was gold. It allowed us to move from just predicting the next question to actually understanding the student’s learning process.
2. Engagement is a Product of Agency, Not Gamification
Gamification was all the rage when we were building our platform. Points, badges, leaderboards—we had them all. And for a while, it worked. Engagement went up. But it was a sugar high. The moment we turned off the rewards, engagement plummeted. We had created a system where students were learning for the points, not for the sake of learning.
We had to go back to the drawing board. We started talking to students, and we found something interesting. The students who were most engaged weren’t the ones with the most badges. They were the ones who felt like they were in control of their own learning. They were the ones who could set their own goals, choose their own learning paths, and see their own progress in a meaningful way.
So we killed the leaderboard. We kept some of the badges, but we made them more meaningful. Instead of getting a badge for answering 10 questions in a row, you got a badge for mastering a new concept. We gave students more control over their learning journey. We let them choose what they wanted to learn next. And you know what happened? Engagement went up, and it stayed up. We had given them agency, and that was more powerful than any game we could have designed.
3. The “AI” Should Be Invisible
This might sound counterintuitive, but the best AI is the AI you don’t even notice. When we first launched, we were so proud of our AI. We had a little chatbot that would pop up and say things like, “Our algorithm has determined that you should work on this skill next.” Students hated it. It was creepy and intrusive. It felt like Big Brother was watching them.
We learned that the AI should be a silent partner. It should be working in the background, making the learning experience better without calling attention to itself. The recommendations should feel natural, like they are coming from a helpful teacher, not a machine. The interface should be simple and intuitive. The student should never have to think about the technology. They should just be focused on learning.
We got rid of the chatbot and replaced it with a simple, personalized dashboard. The dashboard showed the student their progress, their strengths and weaknesses, and a few recommended next steps. It was all powered by the same AI, but it was presented in a much more human-friendly way. The feedback was immediate and positive. Students loved it. They felt like they were in control, even though the AI was still guiding them.
4. Teachers are Your Most Important Users
In EdTech, it’s easy to get so focused on the student experience that you forget about the teachers. That’s a huge mistake. Teachers are the ones who are going to be using your platform in the classroom every day. If they don’t like it, they won’t use it. And if they don’t use it, it doesn’t matter how much the students love it.
We made this mistake early on. We built a platform that was great for students, but it was a pain for teachers to use. The dashboard was confusing, it was hard to track student progress, and it didn’t integrate with their existing gradebook. We had to do a major overhaul of our teacher-facing tools. We spent months working with teachers, understanding their workflows, and building a product that would make their lives easier.
It was a lot of work, but it was worth it. When we launched the new teacher dashboard, our adoption rates skyrocketed. Teachers became our biggest advocates. They were the ones who were telling other teachers about our platform. They were the ones who were helping us to improve it. We learned that if you want to build a successful EdTech product, you have to build it for teachers first.
5. The Ethical Tightrope is Real
When you are collecting data on students, you have a huge responsibility. You have to be transparent about what data you are collecting and how you are using it. You have to protect their privacy. And you have to be constantly vigilant about bias in your algorithms.
This is not something you can just bolt on at the end. It has to be part of your DNA from day one. We had a lot of long, hard conversations about the ethical implications of what we were building. We put in place strict data privacy policies. We had our algorithms audited for bias. And we were transparent with students and teachers about how our platform worked.
It’s a tightrope walk. On the one hand, you want to use data to create a more personalized and effective learning experience. On the other hand, you have to be incredibly careful not to cross any ethical lines. There are no easy answers here. It’s something that you have to be constantly thinking about and working on.
6. Scalability is a Feature, Not an Afterthought
When you are building a product, it’s easy to get caught up in the here and now. You are so focused on building the next feature that you don’t think about what will happen when you have 10,000 users, or 50,000, or 100,000. That’s a recipe for disaster.
We hit a wall when we got to about 10,000 users. Our database was slow, our servers were crashing, and our code was a mess. We had to spend months re-architecting our entire platform to make it scalable. It was a painful and expensive process. We learned the hard way that scalability is a feature. It’s something you have to design for from the very beginning.
This doesn’t mean you have to build a system that can handle a million users on day one. But you do have to be thinking about it. You have to make smart choices about your technology stack. You have to write clean, modular code. And you have to have a plan for how you are going to scale when the time comes.
7. The Mission is What Matters
Building a company is hard. There are going to be times when you want to give up. The only thing that will keep you going is a deep, unwavering belief in your mission. For us, that mission was to make education more accessible and effective for everyone.
That was our North Star. It was the thing that we came back to every time we had a tough decision to make. It was the thing that kept us going when we were working late nights and weekends. And it was the thing that made it all worthwhile.
When you are building a company, you are going to be pulled in a million different directions. You are going to have investors telling you to do one thing, customers telling you to do another, and your own team telling you to do a third. The only way to stay sane is to have a clear mission and to stick to it. That’s the only way to build something that is truly great.
The Real Graduation
Scaling an EdTech platform to 50,000 users was one of the hardest things I’ve ever done. It was also one of the most rewarding. I learned more about business, technology, and people than I ever could have in a classroom. It was my real MBA.
So if you are thinking about building something, my advice to you is this: just do it. Don’t wait for permission. Don’t wait until you have all the answers. Just start. You will learn more than you ever thought possible. And you might just build something that changes the world.
Frequently Asked Questions
How do I know which items apply to my situation?
Start by honestly assessing where your biggest bottleneck is right now. The items that address that specific constraint will give you the highest return on your time and energy.
Are these recommendations still relevant in 2026?
Absolutely. While specific tools and tactics change, the underlying principles remain consistent. I update my thinking regularly based on what I'm seeing in the market and across my portfolio companies.
Can I implement all of these at once?
I'd strongly recommend against it. Pick the 2-3 items that resonate most with your current situation and focus there. Trying to do everything simultaneously is a recipe for doing nothing well.
How were these items selected?
Each item on this list comes from direct experience, either from building my own companies or from patterns I've observed across the 200+ startups I've invested in. I prioritize practical, actionable items over theoretical concepts.