How I Build an AI Strategy (The Counterintuitive Guide)
Everyone’s talking about AI. And most of them are getting it wrong. They see the flashy demos from OpenAI and think the whole game is about building the next superhuman intelligence. I’ve invested in OpenAI, Anthropic, Scale AI, and Hugging Face, and I can tell you—that’s not where the real action is for 99% of businesses.
The real action is in the “boring” businesses. The ones that make things, move things, and sell things. The companies with decades of data and processes that everyone thinks are outdated. If you’re running one of those businesses, you’re not behind on AI. You’re in the perfect position to dominate.
I’ve seen it happen. I saw it when I was building my own companies, and I see it now in the 200+ investments I’ve made. The founders who win with AI aren’t the ones with the most complex algorithms. They’re the ones who are obsessed with solving a real-world problem.
At my first company, MovieLaLa, we were trying to predict which movies would be box office hits. The studios had these incredibly complex models, but they were slow and expensive. We took a different approach. We looked at social media data—how many people were talking about a movie, what they were saying, and how that changed over time. It was a simple model, but it was fast and it was accurate. We ended up getting acquired by Gfycat because we could do something the big guys couldn’t.
Your Biggest Disadvantage is Your Greatest Strength
I talk to founders and CEOs every day who tell me, “Sahin, we can’t do AI. We’re a traditional business. We don’t have the data or the talent.” That’s where they’re wrong. That’s your biggest advantage.
Startups in Silicon Valley have a tough time with AI. They have to invent a problem, then build a product, then find customers, and then figure out how to make AI work. They’re starting from zero. You’re starting from a hundred. You already have customers, you already have processes, and you already have problems. You just need to connect the dots.
Think about it. A manufacturing company has decades of data on machine performance, supply chain logistics, and quality control. A retail business has years of sales data, customer behavior, and inventory records. This is the goldmine. This is what AI feeds on. You don’t need to create it from scratch. You’re sitting on it.
Stop “Testing” AI. Start Implementing.
I see so many companies create these elaborate, multi-year “AI strategies.” They form committees, they hire consultants, they write white papers. It’s all a waste of time. Your AI strategy should be simple: find a problem, and solve it with AI.
This brings me to my biggest pet peeve: the “AI pilot.” People love to “test” AI. They set up a small-scale experiment with no real stakes, and then they’re surprised when it doesn’t go anywhere. I believe a successful pilot isn’t a test. It’s the first step of a full implementation.
When I’m looking at a company, or advising one of my portfolio founders, I have 5 criteria to make sure an AI project is set up for success from day one. If you can’t check all five of these boxes, you’re not ready.
1. It Solves a Real, Painful Problem.
Don’t start with the technology. Start with the pain. What’s the most inefficient, frustrating, or costly part of your business? What keeps you up at night? Is it customer churn? Is it supply chain delays? Is it the mind-numbing manual data entry that your team spends hours on every week?
I once worked with a company that did industrial equipment maintenance. Their biggest problem was unexpected machine failures. It was costing them millions in downtime and repairs. They didn’t need a fancy AI. They needed a simple predictive model that could tell them when a machine was likely to fail. We built it in two months. It saved them $10 million in the first year.
Another example: one of my portfolio companies is in the e-commerce space. They were struggling with customer support. They were getting thousands of tickets a day, and their team couldn’t keep up. We didn’t try to build a chatbot that could solve every problem. We started with the most common one: “Where is my order?” We built a simple AI that could look up the order status and give the customer an update. It handled 30% of their support tickets overnight. That’s a real, painful problem solved.
2. The Data is Already There.
This is the big one. So many AI projects fail because they’re based on data that doesn’t exist. Don’t fall into that trap. Look for problems where the data is a natural byproduct of your existing operations. Sensor readings, transaction logs, customer support tickets, sales records—this is the stuff you want.
If you have to launch a massive, multi-year project to collect and clean your data before you can even start with AI, you’ve already lost. The best AI projects use the data you already have, in the format it’s already in.
I remember talking to a founder who wanted to build an AI to predict employee churn. Great idea, but they had no data. They wanted to start by surveying their employees. I told them to stop. The data was already there, in the HR system. How often were people getting promoted? How long were they staying in their roles? How did their performance reviews look? That’s the data that predicts churn, not a survey.
3. You Can Measure the Impact in Dollars.
This is non-negotiable. If you can’t put a dollar value on the problem you’re solving, don’t bother. “Improving efficiency” is not a goal. “Saving $500,000 a year in overtime costs” is a goal. “Reducing customer churn by 2%” is a goal. “Automating 10,000 hours of manual data entry” is a goal.
When you can measure the impact in dollars, you can make a clear business case for the project. You can get buy-in from your team and your board. And you can prove that the project was a success.
When we built the predictive maintenance model for that industrial equipment company, we knew exactly how much downtime was costing them per hour. So when we showed them that we could reduce downtime by 20%, the math was simple. It was a no-brainer.
4. It Has a Clear Owner Who Feels the Pain.
An AI project can’t be owned by the IT department. It has to be owned by the person who feels the pain of the problem you’re solving. The head of sales should own the AI project that’s supposed to improve lead scoring. The head of operations should own the project that’s supposed to optimize the supply chain.
When the owner of the project is the one who lives with the problem every day, they’re going to be motivated to make it work. They’re going to be your champion. They’re going to be the one who drives it forward.
I’ve seen too many AI projects die on the vine because they were handed off to a data science team that didn’t understand the business context. They built a beautiful model that was 99% accurate, but it didn’t solve the right problem. The business owner has to be in the driver’s seat.
5. It Can Be Implemented in Under 90 Days.
This is my rule of thumb. If you can’t get a meaningful version of your AI solution up and running in under 90 days, it’s too complex. You’re trying to boil the ocean. Start with something small, something manageable, something you can get a quick win with.
That first win is everything. It builds momentum. It gets people excited. It proves that AI can work in your business. Once you have that first win, you can build on it. You can tackle bigger and bigger problems. But you have to start with that first, fast win.
At RemoteTeam, which was acquired by Gusto, we wanted to help companies manage their remote employees. We could have tried to build an all-in-one platform that did everything. Instead, we started with one simple problem: international payroll. It was a huge pain for companies, and we knew we could solve it. We built a simple solution in a few months, and it took off. That gave us the foundation to build the rest of the platform.
The Counterintuitive Path to AI
So, what’s the counterintuitive guide to building an AI strategy? It’s this: stop trying to build an AI strategy. Stop writing documents and forming committees. Stop looking for the most advanced, cutting-edge technology.
Instead, look for the most painful, expensive, and annoying problems in your business. Find the ones where you already have the data, where you can measure the impact in dollars, and where you have a clear owner who is desperate for a solution. Then, find the simplest possible AI tool that can solve that problem, and implement it in under 90 days.
That’s it. That’s the whole strategy. It’s not sexy, but it works. I’ve built my career on it, and I’ve seen it work for dozens of companies. It can work for you, too. The age of AI is here, but it’s not what you think. It’s not about replacing humans with robots. It’s about augmenting your team, solving real problems, and building a better business. Now go find your first win.
Frequently Asked Questions
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 an ai strategy (the counterintuitive guide).?
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.
How long does it take to build an ai strategy (the counterintuitive guide).?
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.