I never went to business school. My education came from the trenches of Silicon Valley, building four companies, and seeing two of them through to acquisition. I learned how to build a multi-million dollar AI company not from textbooks, but from a long series of mistakes, a few lucky breaks, and a whole lot of iteration. This is the playbook they don’t teach you in a classroom.
People see the headlines – RemoteTeam acquired by Gusto, MovieLaLa acquired by Gfycat – and they think it’s a straight line to success. It’s not. It’s a chaotic, messy, and often brutal process. Especially when you’re dealing with something as hyped and misunderstood as AI.
I’ve also been fortunate enough to invest in over 200 startups, including some of the foundational companies in the AI space like Anthropic, OpenAI, Scale AI, and Hugging Face. I’ve seen what works and what crashes and burns from both sides of the table – as an operator and an investor.
Everyone is talking about AI, but most of the conversation is noise. It’s focused on the tech, the models, the algorithms. That’s the easy part. The hard part, the part that actually matters, is finding product-market fit. It's about finding a real, painful problem and solving it for a specific group of people in a way they are willing to pay for.
So, here are my seven hard-won lessons for finding AI product-market fit.
1. Solve a Real Problem, Not a “Cool” AI Problem
This sounds obvious, but it’s the most common mistake I see. Founders fall in love with a cool piece of technology and then go looking for a problem to solve with it. That’s backward. You need to start with the customer and their pain.
With MovieLaLa, my first company, we didn’t start with a grand vision of an AI-powered movie marketing platform. We started with a simple problem: my co-founder and I were movie buffs, and we were frustrated with how hard it was to discover new movies we’d love. The recommendation engines of the time were terrible. They’d recommend Titanic to me every week.
We started by building a simple app that let you track movies you wanted to see and get recommendations from friends. We were the first users. We built something we wanted to use. Only later did we start incorporating more sophisticated algorithms and machine learning to improve the recommendations. The AI was a tool to solve the problem, not the reason for the company’s existence.
Too many founders are building solutions in search of a problem. They’re so focused on building a “generative AI for X” that they forget to ask if anyone actually wants “X”. Don’t be that founder. Fall in love with the problem, not the solution.
2. Nail the User Experience Before You Scale the AI
AI products are often built by engineers, for engineers. They have a million features, a confusing interface, and a steep learning curve. This is a recipe for disaster.
Your product needs to be so simple and intuitive that your mom could use it. The AI should be invisible. It should feel like magic. The user shouldn’t have to think about the complex algorithms running in the background. They should just get the benefit.
Think about Google Search. It’s one of the most sophisticated AI products in the world, but the interface is a simple text box. That’s the goal. You want to hide the complexity and just deliver the value.
At RemoteTeam, we were building a platform to help companies manage their remote teams. We could have built all sorts of fancy AI-powered features for predicting employee churn or optimizing team performance. But we started with the basics: payroll, compliance, and time off. We made it incredibly simple to onboard a new employee in a different country and pay them in their local currency. We nailed the core user experience first. Then, we started layering in more advanced features.
Don’t let your product become a glorified tech demo. Focus on the user and their journey. Make it seamless, intuitive, and even joyful to use. The AI is just a means to an end.
3. Your Data is Your Moat, Not Your Model
Everyone is obsessed with having the best model. They’re chasing the latest and greatest large language model (LLM) and trying to eke out a few extra percentage points on a benchmark. This is a fool’s errand.
The reality is that the models are becoming commoditized. What was state-of-the-art six months ago is now open source. You can’t build a sustainable competitive advantage on a model alone.
Your real moat is your data. A unique, proprietary dataset that you can use to fine-tune a model for a specific task is far more valuable than the model itself. This is why I invested in Scale AI. They understood early on that the bottleneck in AI is not the algorithms, but the high-quality labeled data needed to train them.
At MovieLaLa, our data was our most valuable asset. We had data on millions of users, their movie preferences, and their social connections. We used this data to build a recommendation engine that was far more accurate than anything else on the market. That’s what made us valuable. That’s why Gfycat acquired us.
So, stop worrying about which model you’re using and start thinking about your data strategy. How are you going to collect a unique dataset? How are you going to use it to create a product that is 10x better than the competition? Your data is your kingdom. Protect it.
4. Don’t Be Afraid to Start with a “Wizard of Oz” MVP
Founders often think they need to build a fully automated AI system from day one. This is a huge mistake. It’s slow, expensive, and you’ll probably build the wrong thing.
A much better approach is to start with a “Wizard of Oz” MVP. This is where you have a human performing the task behind the scenes, but the user thinks it’s an automated system. It’s a great way to validate your idea and get feedback from real users before you write a single line of code.
I’ve seen this work time and time again. A founder has an idea for an AI-powered personal assistant. Instead of spending a year and a million dollars building it, they just have a human responding to user requests via text message. They can learn what users are asking for, how they’re phrasing their requests, and what features are most valuable. Then, they can start to automate the most common tasks.
It’s not about tricking your users. It’s about being lean and iterating quickly. It’s about learning by doing. You can’t find product-market fit in a vacuum. You need to get your product in front of users as quickly as possible and see how they react. A Wizard of Oz MVP is the fastest way to do that.
5. Distribution is Everything
You could have the best product in the world, but if no one knows about it, you’re dead. Distribution is the single most important factor for a startup’s success. And it’s often the most neglected.
Founders are so focused on building the product that they forget they also need to build a machine for acquiring customers. You need to figure out your go-to-market strategy from day one. Who is your target customer? Where do they hang out online? How are you going to reach them?
With MovieLaLa, we had a built-in viral loop. Users would invite their friends to the app to get better recommendations. This was our primary growth engine. We also did a lot of content marketing, writing blog posts about movies and the film industry. We became a trusted voice for movie lovers.
There is no one-size-fits-all solution for distribution. You need to experiment with different channels and see what works for your product and your audience. It could be SEO, content marketing, paid ads, sales, or partnerships. The important thing is to have a plan and to execute on it relentlessly.
6. Build a Remote-First Team from Day One
The war for talent is real, especially in the AI space. If you’re only hiring in Silicon Valley, you’re fishing in a very small and very expensive pond. The best people are distributed all over the world.
I’ve been a huge advocate for remote work for over a decade. I built RemoteTeam as a 100% remote company from day one. It was our superpower. We were able to hire the best people, regardless of where they lived. This gave us a huge advantage over our competitors who were stuck in the Bay Area bubble.
Building a remote-first culture is not easy. You need to be intentional about communication, collaboration, and building a sense of community. But the benefits are enormous. You get access to a global talent pool, you have lower overhead, and you can build a more diverse and inclusive team.
In the age of AI, remote work is more important than ever. The tools for remote collaboration are getting better every day. There’s no excuse not to embrace it. If you want to build a world-class AI company, you need to be a world-class remote company.
7. Stay Lean and Iterate Quickly
The AI landscape is moving at a dizzying pace. What’s cutting-edge today will be obsolete tomorrow. The only way to survive is to stay lean and iterate quickly.
You need to be constantly shipping, learning, and adapting. Don’t spend a year in stealth mode building the “perfect” product. Get a minimum viable product (MVP) out the door as quickly as possible and start getting feedback from users.
This is the core philosophy of the Lean Startup, and it’s more relevant than ever in the age of AI. You need to be in a constant cycle of build, measure, learn. Every line of code you write should be a hypothesis that you’re testing. Is this feature valuable to users? Does this change improve the user experience? You need to be ruthless about cutting what’s not working and doubling down on what is.
This requires a certain mindset. You need to be comfortable with uncertainty and ambiguity. You need to be willing to be wrong. You need to be obsessed with your users and their problems.
Building an AI company is a marathon, not a sprint. There will be ups and downs. You will make mistakes. The key is to learn from them and to keep moving forward.
I’m still learning every day. The AI space is the most exciting place to be right now, and I’m incredibly optimistic about the future. But it’s not going to be easy. It’s going to take a lot of hard work, a little bit of luck, and a relentless focus on the customer.
Now, go build something amazing.
Frequently Asked Questions
What if I disagree with some of the advice?
Good. That means you're thinking critically, which is exactly what a good founder should do. Take what resonates, test it, and discard what doesn't work for your specific situation. No advice is universal.
Who is this guide designed for?
This guide is written for founders and operators who want practical, actionable advice rather than theoretical frameworks. Whether you're just starting out or scaling an existing business, the principles here apply across stages.
Is this guide based on real experience?
Every recommendation in this guide comes from direct experience, either from building and selling my own companies, or from patterns I've observed across 200+ angel investments. I don't write about things I haven't personally tested.
How should I work through this guide?
Don't try to absorb everything in one sitting. Read through once to get the big picture, then go back and work through each section as it becomes relevant to your current challenges. Bookmark it and return to it regularly.