Everyone’s talking about AI. And why not? For the first time, it feels like we have magic at our fingertips. With a few API calls, any developer can spin up a chatbot, a content generator, or an image creator that would have seemed like science fiction just a few years ago. I’ve seen it firsthand in my portfolio – companies like OpenAI, Anthropic, and Scale AI are building the foundational models that power this revolution. It’s an incredible time to be building.
But it also raises a terrifying question for founders: if everyone is using the same off-the-shelf models from OpenAI, Google, or Anthropic, how do you build a business that can last? How do you compete when your core technology is, for all intents and purposes, a commodity?
I get this question constantly from the founders I angel invest in. They’re worried. They see a dozen other startups pop up with the same “AI-powered” pitch, and they wonder if they’re in a race to the bottom. They’re asking the right question. Because the truth is, most AI SaaS companies today are not defensible. They’re thin wrappers around a third-party API, and they’re going to get crushed.
But it doesn’t have to be that way. I’ve built and sold two companies, RemoteTeam and MovieLaLa, and I’ve invested in over 200 startups. I’ve seen what it takes to build a lasting business. It’s not about having a secret, proprietary model. For 99% of us, that’s a losing game. The real moat, the real defensibility, comes from somewhere else entirely. It comes from a framework I’ve used to evaluate every AI company I’ve invested in. It’s a framework built on three pillars: proprietary data, unique workflows, and brand.
The Trap of the “Better Model”
Before we get into the framework, let’s talk about the trap. The trap is thinking you can win by having a slightly better model. You can’t.
First, the big players are in a different league. They have armies of PhDs, massive compute budgets, and years of research behind them. You are not going to out-research them. It’s like trying to build a better search engine than Google in your garage. Good luck.
Second, even if you did manage to build a model that’s 5% more accurate for a specific task, how long will that advantage last? A month? A week? The state-of-the-art is moving so fast that any marginal improvement is quickly erased. Your competitors will just switch to the new best model, and your advantage is gone.
This is a lesson I learned the hard way. At MovieLaLa, we were trying to predict what movies people would want to watch. We spent a huge amount of time and energy building our own recommendation engine. It was good, but it was a constant battle to keep it ahead of the curve. When we were acquired by Gfycat, I saw how a massive dataset could create a much more powerful and defensible position than a slightly better algorithm.
So, if you can’t win on the model, how do you win? You win by changing the game. You win by building a moat around your business that has nothing to do with the model itself.
Pillar 1: The Unfair Advantage of Proprietary Data
This is the single most important pillar. If you have a source of proprietary data that your competitors can’t access, you have a real, durable advantage. The model is the engine, but the data is the fuel. And if you have a unique source of fuel, you can build a much more powerful engine over time.
Think about it. The off-the-shelf models are trained on the public internet. They’re great for general-purpose tasks, but they don’t know anything about your specific customers or your specific industry. That’s your opening. Your goal should be to build a product that captures unique data through its very use.
At RemoteTeam, which was acquired by Gusto, we built a platform for managing remote employees. We captured data on everything from payroll and benefits to time off and performance reviews. This data was incredibly valuable. It allowed us to build features that were tailored to the specific needs of remote teams, and it created a powerful network effect. The more customers we had, the more data we had, and the better our product became. That’s a flywheel that’s very hard for a competitor to replicate.
So how do you get proprietary data? Here are a few ways:
- Build it into the workflow: Create a product that requires users to input unique data as part of their daily work. This is the most powerful approach. Think of a vertical SaaS tool for a specific industry, like construction or legal. The data that flows through that system is your proprietary asset.
- Partnerships: Partner with companies that have unique datasets. This can be a great way to get started, but be careful about the terms of the deal. You want to make sure you have long-term, exclusive access to the data if possible.
- Manual collection: In the early days, you might need to roll up your sleeves and collect the data yourself. This is hard work, but it can be a powerful way to build a dataset that no one else has.
Pillar 2: Your Workflow is Your Moat
Even with the same underlying model, the way you present it to the user can be a huge differentiator. Most AI tools are just a text box. You type in a prompt, you get a response. That’s not a workflow, that’s a toy. A real business solves a real problem. And that means building a unique workflow around the AI that guides the user to a valuable outcome.
Think about the difference between a blank page and a template. A blank page is intimidating. A template gives you a starting point and a structure. A great AI SaaS product is like a smart template. It understands the user’s goal and provides a workflow that helps them achieve it.
This is where so many AI companies go wrong. They’re so focused on the technology that they forget about the user. They build a tool that can do a million things, but they don’t show the user how to do any of them.
I’m an investor in a company called Copy.ai. They’re a great example of a company that has built a strong workflow moat. They use the same underlying models as everyone else, but they’ve built a product that’s tailored to the specific needs of marketers. They have templates for everything from writing ad copy to brainstorming blog post ideas. They’ve taken the raw power of the AI and channeled it into a set of specific, valuable workflows. That’s why they’re winning.
Your UI is a key part of this. A great UI can make a complex workflow feel simple and intuitive. It can guide the user, provide feedback, and make the whole experience feel magical. Don’t just throw a text box on a page and call it a day. Think about the entire user journey, from the moment they sign up to the moment they get their desired outcome. Every step of that journey is an opportunity to build a defensible workflow.
Pillar 3: In a Sea of Sameness, Brand is Everything
When every product looks and feels the same, brand is the only thing that can make you stand out. A strong brand is more than just a logo or a tagline. It’s a story. It’s a point of view. It’s a promise to your customers.
In the AI space, where there’s so much hype and confusion, a strong brand can be a beacon of trust. It can signal to your customers that you’re not just another fly-by-night startup. You’re a company that’s here to stay, and you’re committed to solving their problems.
Building a brand is not easy. It takes time and consistency. But it’s one of the most valuable investments you can make. Here are a few things to focus on:
- Have a strong opinion: Don’t be afraid to take a stand. Have a clear point of view on your industry and your customer’s problems. This is what will attract your tribe.
- Build in public: Share your journey. Talk about your successes and your failures. This will make your brand feel more human and relatable.
- Create a community: Build a community around your product. This could be a Slack group, a forum, or a series of events. A strong community can be a powerful source of feedback, support, and evangelism.
I wrote a book called “Becoming Top 1%” because I have a strong opinion about what it takes to succeed. That book is part of my brand. It’s a way for me to share my story and my point of view with the world. It’s not just about selling books, it’s about building a connection with people who share my values.
The Road Ahead
Building a defensible AI SaaS business is not about having the best model. It’s about being smart, strategic, and focused on the customer. It’s about building a moat that’s made of proprietary data, unique workflows, and a brand that people trust.
This is not a race to the bottom. It’s a race to the top. The companies that win will be the ones that understand that AI is not a product, it’s an ingredient. The real magic happens when you combine that ingredient with a deep understanding of your customer’s needs and a relentless focus on building a product that they love.
So, stop worrying about the models. Start worrying about your data, your workflow, and your brand. That’s how you build a company that lasts. That’s how you win.
Frequently Asked Questions
How long does it take to build a defensible ai saas when everyone has access to the same models?
The timeline varies depending on your starting point and resources. For most founders, expect 2-4 weeks for initial setup and 2-3 months to see meaningful results. I've seen teams move faster when they focus on one thing at a time rather than trying to do everything at once.
What tools do I need to get started?
Start with the basics. You don't need expensive software or fancy tools. A spreadsheet, a note-taking app, and direct access to your customers will get you further than any enterprise platform. Add tools only when you hit a specific bottleneck.
What are the most common mistakes when building a defensible ai saas when everyone has access to the same models?
The biggest mistake I see is overcomplicating things early on. Start with the simplest version that works, get real feedback, and iterate from there. Another common trap is copying what worked for someone else without understanding the context behind their decisions.