My Journey with AI: From Grad School to Experimentation Machine

Startups are unique among business enterprises in that they are, at their core, experimentation machines.

The problem is that experiments are resource-intensive—both in time, money, and human capital—and resources are not something a startup has to spare.

The crux of the startup challenge is to experiment, learn, and iterate quickly enough to find product-market fit before going broke.

That’s what makes AI a game changer for startups.

AI will help startups accelerate the experimentation process at every stage:

• Identify the most promising experiments to run

• Design better experiments, faster

• Collect and analyze results more efficiently

• Incorporate feedback automatically into the next iteration

With the right tools, startup teams will have more cycles to experiment and learn, giving them a better chance at hitting pay dirt.

I have been fascinated by AI for decades. Here is my personal journey with AI, and why I am so bullish on its potential:

My AI Journey: From College to Experimentation Machine

My fascination with AI began as a computer science major at Harvard College in the late 1980s. That era was marked by the birth of personal computers and the emergence of telecommunications protocols that enabled the widespread use of email, eventually providing the underpinnings for the Internet. It was a time of rapid technological advancement, and I was captivated by the potential of these new tools.

I was a science fiction buff (Isaac Asimov, in particular) and loved to imagine a world of robots and cognitive aids. I took graduate courses in neural networks, natural language processing, and computer vision in parallel to my undergraduate computer science work. My undergraduate computer science thesis was titled “Coordinating Natural Language and Graphics in the Automatic Generation of Explanations of Causal Business Models.”, which is a fancy way of saying I applied early AI concepts to an application designed to “generate sentences in one of the languages natural to human beings” to explain business decisions. Sound familiar?

After completing my degree, I decided to pursue a business career rather than continue down the path of computer science research. But I always kept sight of the transformative potential of AI. As an entrepreneur, I was an executive at two software companies that used early AI techniques—one to personalize ecommerce and the other to assist in college savings. As a young VC investor at Flybridge decades ago, I led the seed rounds in a few early pioneering AI startups, including one that used machine learning techniques to provide fairer, transparent underwriting of consumer loans. I extolled the virtues of pursuing machine learning to founders (equating it to the advice of embracing “plastics” from the classic movie The Graduate). When ChatGPT debuted in 2022, I knew it would be a turning point. The ability of gen AI tools to rapidly (and rather cost-effectively) generate and analyze content would open up new possibilities for startup founders.

Early in 2023, I began experimenting with gen AI in the “Launching Tech Ventures” course I teach at Harvard Business School. With the help of a few former students, I created an AI chatbot or faculty co-pilot called ChatLTV, which my class used to access information about the course and explore general startup topics. The experience of integrating AI into my teaching reinforced my belief in its potential to reshape the startup landscape.

At Flybridge, my team and I decided to focus exclusively on investing in AI-enabled startups. This focus reflects my deep conviction that AI will catalyze a new era of startup innovation and growth.

This journey ultimately led me to write my new book The Experimentation Machine: Finding Product-Market Fit in the Age of AI. My goal is to share the strategies I’ve learned so far in using AI to build successful ventures—across my students and portfolio companies.

Building Startups in the Age of AI

The path to product-market fit will always be uncertain and require creativity, strategic insight, and relentless experimentation.

However, generative AI promises to transform this process for founders, allowing them to innovate, validate, and iterate faster and more efficiently than ever before. 

In my Experimentation Machine newsletter, I am going to share a lot of stories and case studies about founders using AI to build their startups more effectively and efficiently. Here’s what you can expect in future posts:

• Case studies of founders navigating the journey to product-market fit

• How to design experiments to test your core startup hypotheses with AI

• AI-powered go-to-market strategies

• Updates about my book, including chances to read early drafts.

Subscribe to The Experimentation Machine

Thanks for reading, and happy building!

Best,

Jeff