Six months ago, my team and I fell into a rabbit hole. We had a simple question: why are so many companies lighting money on fire when it comes to AI? We’d seen it in our portfolio companies, we’d heard it from fellow investors, and frankly, I’d lived it myself in my own startups. There’s this massive gap between the promise of AI and the grim reality of corporate adoption.
So, we decided to do something about it. We spent half a year analyzing the data from over 500 enterprise AI projects. Not just surface-level stuff. We got our hands dirty. We looked at internal strategy docs, project post-mortems, and even raw budget allocations. The results were, to put it mildly, shocking.
It turns out, the number one reason corporate AI strategies fail has almost nothing to do with the technology. It’s not about the algorithms, the models, or the compute. It’s something far more fundamental, and it’s costing companies billions.
The Great AI Charade
Here’s the thing. Most companies are playing a game of pretend. They’re buying “AI” like it’s a new software license. A VP reads a report, the board gets nervous about falling behind, and a mandate comes down from on high: “We need an AI strategy!”
What happens next is a predictable and painful dance. A budget is assigned. A team is formed, often staffed with people who are brilliant but have no real context on the business’s core problems. They spend months evaluating vendors, running proof-of-concepts, and generating beautiful charts that show… well, not much of anything that matters.
I saw this firsthand a few years ago. A major e-commerce company, a household name, spent over $10 million on a fancy AI-powered recommendation engine. The goal was to increase the average order value. After 18 months, they rolled it out. The result? A 0.2% lift. The executive who championed the project declared victory. But when you looked at the actual numbers, the cost of maintaining the system completely wiped out any gains. It was a net loss. A very expensive one.
They didn’t have a business problem. They had a technology solution looking for a home. Our data showed this pattern over and over again. We found that a staggering 78% of the failed projects we analyzed started with a specific technology in mind, rather than a specific, well-defined business problem.
You’re Solving the Wrong Problem
The fundamental flaw is this: companies are trying to adopt AI, not solve problems with it.
Think about that for a second. It’s a subtle but critical distinction. When you try to “adopt AI,” you start with the tech. You ask, “What can we do with a large language model?” or “How can we use computer vision?” This approach is a recipe for disaster. It leads to science projects, not business results.
I remember back at RemoteTeam, before the acquisition by Gusto, we were struggling with customer churn. We had a ton of data, and the knee-jerk reaction was, “Let’s build a predictive AI model to see who will churn!” I was right there, leading the charge. I honestly had no idea what I was doing at first. We spent two months trying to build a complex model. It was a mess. The predictions were barely better than a coin flip.
Then, our head of customer success, a woman who had spent a decade just talking to customers, walked into my office. She said, “Look, I can tell you why people are churning. We’re not onboarding them correctly, and our support response time for payroll issues is too slow.”
It was a gut punch. The problem wasn’t a prediction problem. It was an operational one. We scrapped the AI project and put all those resources into fixing our onboarding flow and hiring more support engineers. Our churn rate dropped by 30% in one quarter. We didn’t need a fancy algorithm; we needed to listen to our customers and fix our broken processes. We later used simpler automation to help the support team, but only after we understood the real issue. For more on this, you can read my post on the true cost of ignoring customer feedback.
The AI ROI Litmus Test
So how do you avoid this trap? You need to stop talking about AI. Seriously. Ban the term from your strategy meetings for a month. Instead, focus obsessively on identifying the most painful, expensive, and inefficient problems in your business.
Here’s a simple litmus test I now use for any AI project, whether it’s in one of my portfolio companies or a new idea I’m exploring:
Can you define the problem without using the word “AI”? If you can’t, you don’t have a real business problem. “We need an AI to improve marketing” is not a problem. “We’re spending $2 million a year on ad campaigns with a low conversion rate, and we don’t know which channels are effective” is a problem.
Is the problem measurable in terms of dollars, hours, or customer satisfaction? You have to know what success looks like. If you can’t quantify the impact, you can’t justify the investment. Don’t accept vague goals like “enhance efficiency.” Demand specific targets like “reduce manual data entry by 500 hours per week.”
This is the exact opposite of how most companies operate. They get excited about the tech, and the business case is a flimsy justification bolted on at the end. You need to flip that script. Start with the pain. Start with the ROI. The technology is the last piece of the puzzle, not the first.
I’ve invested in over 200 companies, including some you might have heard of like Anthropic and Scale AI. The successful ones all have one thing in common: they are relentlessly focused on solving a real-world problem. The AI is just a tool they use to do it. It’s a means to an end, not the end itself. If you want to dig deeper into how to evaluate these kinds of investments, check out my thoughts on how I became a top angel investor.
Your AI Strategy Is Not an AI Strategy
Here’s the bottom line. Stop trying to have an “AI strategy.” It’s a meaningless buzzword. Instead, have a business strategy that identifies the critical problems you need to solve to win in your market. Then, and only then, should you ask if AI can be a part of the solution.
Maybe the answer is yes. Maybe it’s a simple automation script. Maybe it’s hiring more people. The tool doesn’t matter as much as the problem you’re pointing it at.
The companies that are getting real, tangible returns from AI aren’t the ones with the biggest AI budgets or the most PhDs. They’re the ones who are most deeply in touch with their own business. They’re the ones who are honest about their weaknesses and focused on fixing them.
So, take a hard look at your own initiatives. Are you playing the great AI charade, or are you solving real problems? The answer will determine whether your AI journey ends in a blaze of glory or just a pile of wasted cash.
Frequently Asked Questions
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
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.
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.