Investing in AI before it became a mainstream buzzword taught me the immense value of focusing on fundamentals over hype, the critical importance of a data-first approach, and the power of visionary founders who see the world not as it is, but as it could be. These early bets underscored that true disruption often happens quietly before it suddenly becomes obvious to everyone.
The Pre-Hype AI Landscape: Seeing the Signal in the Noise
I started investing in AI startups long before "generative AI" was a term you heard in every coffee shop in Silicon Valley. Back then, the field was different. It wasn't about flashy demos or large language models writing poetry; it was about solving specific, deeply technical problems. The lessons from investing in AI before it was cool were forged in a quieter, more focused era. My first forays involved companies that were using machine learning for things like fraud detection or optimizing complex supply chains. These weren't the kind of ventures that made headlines, but they were building the foundational layers of the AI revolution we see today.
One of the core principles I developed was to look for founders who were obsessed with a problem, not a technology. The most promising entrepreneurs I met were the ones who had spent years, sometimes decades, in a particular industry and saw how machine intelligence could fundamentally reshape it. They spoke the language of their customers, not just the language of algorithms. This focus on problem-solving over pure tech novelty became a cornerstone of my investment thesis and has paid dividends time and time again. It taught me to filter out the noise of fleeting trends and focus on the signal of sustainable value creation.
Another key insight was the importance of proprietary data. In the early days, you couldn't just plug into a pre-trained model. The most successful AI companies were the ones building unique datasets that gave them a competitive moat. I learned to ask the tough questions: How are you acquiring your data? Is it defensible? Does it create a feedback loop where your product gets smarter with each new user? This data-centric view is one of the most critical investing in AI before it was cool insights that remains just as relevant today. You can have the best algorithm in the world, but without a strong data strategy, you're building on sand.
Lesson 1: The Founder's Vision is Everything
If there's one thing I've learned from backing over 200 startups, it's that the founder is the single most important variable. This is exponentially true in deep tech fields like AI. When I was investing in AI before it was cool, I wasn't just betting on a technology; I was betting on a founder's ability to see a future that others couldn't. These were the people who could connect the dots between a nascent algorithm and a billion-dollar market opportunity. They had a conviction that was almost irrational, and it was that conviction that carried them through the long, arduous journey of building something truly new.
I remember meeting the founders of a computer vision company back in 2015. Their tech was impressive, but what sold me was their vision for how it would revolutionize retail analytics. They weren't just building an algorithm to recognize objects; they were building a system to understand human behavior in physical spaces. That's the kind of expansive thinking that separates a good investment from a great one. For anyone looking to build or invest in AI, my advice is simple: find the founders who are not just technologists but also storytellers and visionaries. They are the ones who will create the future. For more on identifying top talent, check out my thoughts on the traits of successful startup founders.
Lesson 2: Data Isn't Just a Resource, It's the Product
Another of the core what I learned investing in AI before it was cool takeaways is that the business model of many successful AI companies is not just selling software, but using data to create an ever-improving product. The best AI startups create a virtuous cycle: more users lead to more data, which makes the AI model smarter, which in turn attracts more users. This data flywheel is an incredibly powerful competitive advantage that is very difficult for competitors to replicate.
I always look for companies that have a clear strategy for data acquisition and a plan to build a proprietary dataset. This could be through user-generated content, partnerships, or by creating a tool that people are willing to use in exchange for their data. For example, an early investment of mine was in a company that provided a free tool for developers. The tool was incredibly useful, and in the background, it was collecting anonymized code data that was used to train a powerful AI model for code completion. This is a perfect example of a data-first approach that creates long-term, defensible value.
Key Insight: Don't just ask how a company's AI works. Ask how its data works. The long-term winners will be the ones who have built a data engine, not just an algorithm. This is a fundamental shift in how we should think about building and investing in technology companies.
Lesson 3: The ‘Boring’ Problems are Often the Most Profitable
Everyone today is chasing the dream of building the next ChatGPT, but some of the most successful AI investments I've made have been in companies solving decidedly “boring” problems. Think about the unsexy but critical business functions that power our economy: logistics, accounting, regulatory compliance, and manufacturing. These industries are often complex, inefficient, and ripe for AI-powered disruption. The lessons from investing in AI before it was cool often lead back to these unglamorous, yet highly valuable, opportunities.
One of my portfolio companies built an AI platform to automate the incredibly tedious process of customs documentation for international shipping. It’s not the kind of thing that gets you on the cover of a magazine, but it solves a massive pain point for a huge industry and has become an indispensable tool for their customers. The ROI is immediate and measurable, which is a much easier sell than a speculative, futuristic vision. These “boring” problems often have clear business models and a direct path to revenue, which makes them far less risky than consumer-facing AI ventures.
When evaluating an AI startup, I often look for the following characteristics:
- High-Value, Repetitive Tasks: Is the company automating a process that is currently done by highly paid professionals and is prone to human error?
- Complex, Unstructured Data: Does the industry rely on complex documents, images, or other unstructured data that is difficult for traditional software to handle?
- Clear ROI: Can the company clearly articulate how their solution will save their customers time and money?
- Fragmented Market: Is the target market large but served by outdated, legacy software solutions?
Finding a startup that ticks these boxes is often a sign that you've found a hidden gem. For more on how I evaluate potential investments, you can read about my angel investment framework.
The Human Element: AI as an Augmentation, Not a Replacement
One of the most persistent myths about AI is that it's a job-killer. While it's true that AI will automate many tasks, the most powerful applications I've seen are the ones that augment human capabilities, not replace them. The best AI tools make experts even better at their jobs. This is a crucial investing in AI before it was cool insight that has guided my investments toward collaborative AI systems. I look for companies that are building tools that empower professionals, not just automate them away.
Think about a radiologist using an AI to help them spot anomalies in a medical scan, or a lawyer using an AI to quickly sift through thousands of legal documents. In both cases, the AI is not making the final decision, but it is providing the human expert with superpowers. This human-in-the-loop approach is not only more effective, but it is also much more likely to be adopted in industries where the stakes are high. People are naturally resistant to the idea of a black box making critical decisions, but they are very open to tools that can help them do their jobs better.
This is why I believe the future of AI is not about creating artificial general intelligence that can do everything a human can do. It's about building a suite of specialized AI tools that can help us be more creative, more productive, and more effective in our work. The real revolution will be the quiet integration of AI into the daily workflows of every professional, and the companies that enable this will be the ones that create the most value in the long run. It's a vision I've discussed in more detail when talking about the future of work and AI.
Frequently Asked Questions
What was the biggest surprise when you started investing in AI?
The biggest surprise was how long the sales cycles were, even for obviously superior technology. I learned that having a 10x better product doesn't mean you'll get a 10x faster adoption rate. In enterprise, especially in regulated industries, there's a tremendous amount of inertia. This taught me to value founders who had deep industry expertise and understood the nuances of the sales process in their specific vertical.
How has your investment thesis for AI changed over the years?
While the core principles of backing great founders and focusing on data moats have remained the same, my thesis has evolved. Initially, I was focused on very specific, narrow AI applications. Today, I'm more interested in platform-level opportunities and companies that are building the enabling infrastructure for the next generation of AI development. The rise of large language models has opened up a whole new set of possibilities that simply didn't exist a few years ago.
What's your number one piece of advice for a founder building an AI startup today?
Focus on a real-world problem that you are uniquely qualified to solve. Don't get caught up in the hype cycle or try to build the next big thing in generative AI unless you have a truly differentiated approach. The most successful AI companies will be the ones that are deeply embedded in the workflows of specific industries and are solving tangible business problems. Be a problem-solver first, and a technologist second.
Final Thoughts
Looking back, what I learned investing in AI before it was cool was a masterclass in first-principles thinking. It forced me to look beyond the surface-level hype and focus on the timeless principles of building a great business: a visionary founder, a sustainable competitive advantage, and a relentless focus on solving a real customer problem. The technology has evolved at a striking pace, but these fundamental truths remain as relevant as ever.
For entrepreneurs and investors working through the current AI gold rush, I urge you to take a step back and apply these lessons. Don't be mesmerized by the magic of the technology; be grounded in the reality of the market. The biggest opportunities won't be found in chasing the latest trend, but in the patient, disciplined work of building real solutions for real people. The AI revolution is still in its early innings, and I, for one, am more excited than ever to be a part of it.