Everyone thinks they’re going to get rich in the AI gold rush. They’re wrong. Most are just selling shovels in a race to the bottom, burning cash on GPU servers and fancy models while their margins evaporate.
I’ve seen it happen more times than I can count. I’m Sahin Boydas. I’ve built and sold two companies, RemoteTeam to Gusto and MovieLaLa to Gfycat. I’ve also been fortunate enough to be an early investor in over 200 startups, including some of the foundational companies in the AI space like OpenAI, Anthropic, Scale AI, and Hugging Face. I wrote a book about this stuff, "Becoming Top 1%".
I’m not here to sell you a dream. I’m here to give you a dose of reality. A great product is table stakes in vertical SaaS. You need a go-to-market strategy that is ruthlessly tailored to your specific industry. This is my step-by-step guide to acquiring your first 100 customers and actually building a profitable AI SaaS business, not just a cool tech demo.
The Brutal Economics of AI
Let's get one thing straight: running AI is expensive. Really expensive. The public sees these magical APIs that spit out text or images and assumes the cost is near zero. It’s not. Not even close.
When I was looking at the early models from OpenAI and Anthropic, the costs were astronomical. We’re talking about training runs that cost millions of dollars. And that’s just to get the model built. Then you have inference costs – the cost of actually running the model for your users. Every time a customer makes a query, you’re paying for compute. It’s a meter that’s always running.
I remember a pitch from a startup that wanted to build an AI-powered code completion tool. Great idea. But when I dug into their numbers, they were planning to charge a flat $10 per month. Their average user was making thousands of API calls a day. Their cost per user was going to be closer to $50 a month. They were planning to lose $40 on every single customer. They were mesmerized by the tech and completely ignored the business model. They didn’t last six months.
This is the most common mistake I see. Founders fall in love with the technology and forget about the economics. You can’t just slap a subscription fee on an AI product and hope for the best. You have to understand your unit economics from day one.
Pricing Your AI Product: The Meter is Your Friend
So how do you avoid this trap? You have to get your pricing right. And for most AI SaaS businesses, that means usage-based pricing.
Flat-rate subscriptions are a death sentence in AI. You’re exposing yourself to unlimited downside. If a user goes viral or builds a power-user workflow on top of your product, you could be on the hook for tens of thousands of dollars in compute costs with no corresponding increase in revenue. It’s a recipe for disaster.
Usage-based pricing aligns your revenue directly with your costs. The more a customer uses, the more they pay. It’s fair, it’s transparent, and it protects your margins. It also allows you to capture more value from your power users.
I pushed the team at one of my portfolio companies, a legal tech startup using AI to analyze contracts, to switch to usage-based pricing. They were hesitant at first. They were worried about customer pushback. But I told them, “Your best customers, the ones who are getting the most value from your product, will be happy to pay more. And your tire-kickers, the ones who are just playing around with the free tier, won’t be a drain on your resources.”
They made the switch. And what happened? Their revenue went up by 40% in the first quarter. Their churn went down. And their power users started giving them even more feedback, because now they had a vested interest in the product’s success.
There are a few different ways to structure usage-based pricing for AI:
- Per-API call: This is the simplest model. You charge a small fee for every API call. This is great for products with a clear, transactional value proposition.
- Per-unit of compute: This is a more sophisticated model. You charge based on the amount of GPU time or other resources a customer consumes. This is a good option for more complex products where the cost of a query can vary significantly.
- Value-based pricing: This is the holy grail. You charge based on the value your product creates for the customer. For example, if you’re an AI-powered ad optimization tool, you could charge a percentage of the ad spend you manage or the revenue you generate. This is the hardest model to implement, but it’s also the most profitable.
Don’t overcomplicate it to start. Pick a simple, transparent model and iterate from there. The key is to have a meter running. Always have a meter running.
Your Go-to-Market is Your Moat
In the world of AI, your model is not your moat. Let me say that again. Your model is not your moat. The big players—OpenAI, Google, Anthropic—will always have better models than you. They have more data, more engineers, and more money. You can’t compete on model quality alone.
Your moat is your go-to-market strategy. It’s how you reach, acquire, and retain customers in a specific vertical. A great product is not enough. You need a distribution advantage.
When we were building RemoteTeam, we were entering a crowded market. There were dozens of HR tools out there. But we had a unique insight: the rise of remote work was creating a new set of challenges for companies. Payroll, compliance, and team management were all different when your team was distributed across the globe. We tailored our product and our go-to-market to this specific niche. We wrote blog posts about the challenges of remote work. We sponsored remote work newsletters. We built a community of remote-first founders. We became the go-to experts in our niche. That was our moat. Gusto didn't acquire us for our code; they acquired us for our market position and our customer base.
For an AI SaaS business, this is even more critical. You need to find a niche where you can be the dominant player. Don’t try to build a general-purpose AI writing tool. Build an AI writing tool for real estate agents. Or for lawyers. Or for scientists. The more specific you are, the better. You can build a dataset and a workflow that is perfectly tailored to the needs of your customers. You can become the indispensable tool for your industry.
This is how you build a real business, not just a feature. The big AI platforms are like the cloud providers of the last decade. They provide the infrastructure. It’s up to you to build the applications on top. And the most successful applications will be the ones that solve a specific business problem for a specific set of customers.
The Grind: Acquiring Your First 100 Customers
Alright, so you have a niche, a product, and a pricing model. Now comes the hard part: getting people to actually pay you for it. The first 100 customers are the toughest. This is where you have to grind. There are no silver bullets here. It’s about direct, unscalable work.
Here’s my playbook. It’s not glamorous, but it works.
Step 1: Forget about scale. Do things that don’t scale. Paul Graham was right. Your first customers will come from hand-to-hand combat. I mean that literally. You need to be personally reaching out to potential customers and convincing them to try your product. Make a list of 100 dream customers in your niche. Find the contact information for the decision-makers. And then email them. One by one. Personalize each email. Tell them you’re the founder, you’ve built something specifically for them, and you’d love to get their feedback.
Step 2: Live in their communities. Where do your customers hang out online? Are they in specific subreddits? LinkedIn groups? Slack communities? You need to be there. Not to spam them with links to your product. But to listen. To answer questions. To become a trusted resource. When you provide value first, people will be much more receptive to hearing about what you’re building. I spent hours every day in remote work forums when we were starting RemoteTeam. I wasn’t selling. I was helping. The sales came naturally after that.
Step 3: Build a feedback engine. Your first 100 customers are not just a source of revenue. They are your most important source of feedback. You need to be talking to them constantly. What do they love? What do they hate? What features are they begging for? I set up a private Slack channel with our first 50 customers. I was in there every day, answering questions, debugging problems, and getting ideas. This direct line to our users was invaluable. It helped us build a product that people actually wanted to use.
Step 4: Close the loop. When a customer gives you feedback, and you implement it, tell them. It sounds simple, but it’s incredibly powerful. It shows that you’re listening. It makes them feel like a part of the team. It turns them into evangelists for your product. I once had a customer who suggested a small UI tweak. We pushed the change live the next day and sent him a personal email to let him know. He was so blown away that he wrote a blog post about us that drove a flood of new signups.
This is the work. It’s not about running Facebook ads or hiring a fancy PR firm. It’s about building real relationships with real people. If you can’t do that, you’re not going to make it.
Stop Chasing Rainbows
Building a profitable AI SaaS company isn't about having the most advanced algorithm or the biggest GPU cluster. It's about finding a painful, expensive problem in a specific industry and solving it better than anyone else. It's about understanding your costs and aligning them with your revenue. It's about grinding it out, one customer at a time.
Forget the hype. The AI gold rush is a myth for 99% of founders. The real opportunity isn't in building the next foundational model; it's in applying these powerful new tools to old, boring industries. The money isn't in the magic, it's in the mundane. It's in the unsexy, vertical-specific workflows that the big players will never bother with.
So, stop chasing the AI rainbow. Go find a real problem. Solve it. Charge for the value you create. That’s how you build a business that lasts. That's how you become the top 1%.
Frequently Asked Questions
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.
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.
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.