“There’s the myth of the self-made man, but in reality, there are so many actors and factors that contribute to our success.”
– Sunit Gala, 2023
As someone who interviews a handful of high-level entrepreneurs on a monthly basis, you catch on pretty early to the different types of energies and personalities of who’s sitting in front of you. Sunit was unlike what you would expect from a Bay Area Ph.D. in “computer stuff” (as he put it), and a serial tech entrepreneur with several exits under his belt. He talked about the subject of artificial intelligence in animated vigor, the excitement bouncing off like sparks. However, he was simultaneously reserved, pausing sporadically to ask if I wanted him to slow down or stop talking.
In this riveting 80-minute interview, Sunit uncovers his insights on artificial intelligence, the past and where it’s going, and the riveting story of how he helped scale Oracle’s product department from $5 million to $100 million in revenue in one year. In the second half of the interview, we take a close look at how Sunit went from an average C+ student in college to discovering his ‘gift’ for mathematics only in graduate school—which he got into by mere chance.
Sunit Gala is a serial tech entrepreneur and engineer who’s sold his companies to Siemens, FTT, and Iksula. He has over 1000 academic citations to date for his work in AI and big data. Today he is focused on leveraging his expertise in startup scaling to help other tech and artificial intelligence companies make their breakthroughs on the market.
Interviewer: Can you give us a run-through of the AI landscape?
Sunit: AI has been around since at least the fifties and for a long time the focus was on logic or logical reasoning which subsequently led to semantics and neural networks. From the seventies, until 2009 or 2010, the neural network space was relatively static with no major breakthroughs. However, Deep Mind changed that and is the most important breakthrough in AI in the last 40 to 50 years. People are taking large volumes of data and applying neural networks to do curve fitting on steroids. Most neural network and machine learning algorithms are similar, both use supervised learning but Deep Mind has introduced unsupervised learning at scale in what is called deep reinforcement learning.
Interviewer: What is a neural network?
Sunit: A neural network is crucial to understand, but its importance can be lost in the noise. Here’s one simple way to understand it. Essentially, it takes a table of data as input, where each column represents a variable and each row contains the data values for all the columns. For example, if there are only two columns X and Y, then X represents one variable and Y represents another variable; and each row represents a point on the XY axis. By feeding a large number of rows as input, a neural network can effectively and quickly draw an equation that connects these dots, even when dealing with a thousand variables.
Interviewer: When you created this AI machine for lawyers, how long did it take?
Sunit: There’s no “exact” time frame for when the model reaches maturity. It can be months or years. First, a large corpus of text needs to be ingested by the engine (in this case, relevant law books from all 50 states), and then someone has to curate the dictionary to provide the context, and has to tell the system that when a user needs more information, where else can they go? It is a fairly labor-intensive task as somebody has to curate dictionaries, terms, and terminology. They keep testing and playing with different scenarios and use cases to see if it is behaving reasonably, and as expected. The “maturity” phase you’re referring to is when the machine can be reasonably trusted and accepted by the users.
Interviewer: Do you see any dangers in terms of accuracy when it comes to AI and how people might become overly reliant on Google to find a quick answer, even though it’s not always correct? Do you see any problems with that?
Sunit: Like any tool, the value is extracted by the user. The same tool in the hand of a different user ends up becoming a weapon. ChatGPT can be a very good starting point to get some quick research done, but you have to be extremely skeptical of the results, especially if the topic is arcane or unusual.
Interviewer: Where do you see it all going? How do you see that transition happening?
Sunit: Ever since the industrial revolution, the sentiment has been fear-based, as in, “This technology is going to go kill millions of jobs and everybody’s going to starve to death.” How often have we heard that?
The only case when people lose their jobs and are unable to get better ones is due to certain political policies, such as ones where we allow jobs to be exported to cheaper countries (ie. Mexico, China..etc) without upskilling or other alternatives. The opposite is true when you introduce technology to an existing, productive activity. Whenever you are in a productive activity and you are simply bringing in new technology, your productivity, and concomitant economic output will increase. The saved time allows you to upskill at a faster rate. The nature of progress is upward bound.
Interviewer: Can you tell us about your journey into entrepreneurship? How did your background or past affect your present career?
Sunit: My journey to entrepreneurship has been shaped by a variety of factors. I believe that no individual can achieve success alone. Throughout my life, many people have supported me, including my parents who nurtured me, my teachers who provided me with an education, and my family. Although I did not enjoy doing homework, I still learned and developed skills through education.
During my first semester of graduate school, I experienced a transformative moment that changed my outlook. I asked my graduate coordinator why he accepted me given my poor academic record. He explained that he saw potential in me and believed I could excel. This motivated me to work harder. I also took a course on algorithms that semester and the professor noticed my talent for math. He offered me a research opportunity in neural networks and AI, which subsequently led me to explore semantics and distributed systems for big data applications.
After completing my Ph.D., I struggled to find a job during a deep recession and ended up at a startup working on algorithms for distributed databases. I noticed a gap in the sales process, where technical product managers lacked engineering expertise, and salespeople were not technical enough. This sparked my interest in product management and led me to become the first Java product manager at Oracle. I later started my first company which was eventually acquired by Siemens
Interviewer: Was helping scale Oracle in one year due to your engineering expertise or leadership skills?
Sunit: Scaling Oracle’s revenue in a single year was not due to my engineering expertise. My bosses placed their trust in me and allowed me to build the right team. We were a curious and hardworking bunch. with an intimate understanding of the products and domain. We eventually stumbled onto the right messaging and positioning.
At the time, Oracle had several products that were powerful but quite rough around the edges. We all understood the potential of these products but were puzzled as to why they weren’t selling. Through conversations with stakeholders, including existing and potential customers, I learned that our sales force was not adequately educated and had difficulty demonstrating the products to customers.
I believe in process and knew we needed to reposition ourselves. I talked about the life cycle of data and how we could help companies collect, report, analyze, and optimize their data, eventually reaching the goal of producing better outcomes driven by good data. By creating a story around this and improving our messaging and positioning, we were able to make the products more appealing to customers. We also got a bit lucky, as compliance became a big deal for everyone after a new law (known as Sarbanes-Oxley) was passed at the time. Additionally, my team created better demos to showcase the product’s capabilities and make them more visually appealing. These processes and collective action were what helped scale Oracle’s revenue in one year.
Interviewer: Can you give any advice to entrepreneurs who are currently having trouble fundraising?
Fundraising is often about who you know, rather than what you know. It’s a full-time job to network when you’re fundraising, which means you don’t have much time to work on the business. Additionally, different investors have different metrics, and investing is fad-driven. You never know what will be the next big thing or when a fad will end. To get funding, your positioning must be clear, the numbers must check out, and you must have some traction. Networking and talking to people are crucial, but unfortunately, there are no shortcuts to funding.
Interviewer: What is the next step for you?
Sunit: I genuinely want to make an impact. I’ve done some volunteer work with the local food bank and with Meals on Wheels. I also helped co-create what has now become the world’s largest free vaccination reminder system, Immunize India. I don’t know much about issues like world hunger, so I don’t know how to think about it or, or solve those problems. How I can be extremely effective is in helping founders find their paths. Many of my mentees have gone on to find their own companies over the years. I find that very fulfilling and impactful. This is why currently I want to find the right company to work at and help them grow.
Interviewer: Has your perspective on impact changed throughout your lifetime? Were you the type that was very success-driven, or have you always thought about impact in the back of your mind?
Sunit: No, I’ve never thought about impact at all. As a young person in grad school, I was just ridiculously curious about anything and everything. I have a lot of useless knowledge—there was neither method nor madness, but I was just doing and trying things out. I believe life has been kind to me. There’s the myth of the self-made man, but in reality, there are so many things, so many actors and factors that contribute to our success. I would recommend checking out a book called Fooled by Randomness by essayist, Nassim Nicholas Taleb.
Conclusion
Although AI appears nascent to the common eye, it seems after seven or so decades, it’s finally on the cusp of the next breakthrough. However, with that, also comes the issue of ethics and practicality if AI does reach a stage where it’s sentient or completely unsupervised. Meanwhile, as the AI war plays out between tech companies, it also paves the way for startups and workers to utilize AI to expedite menial activity and become exponentially more productive. AI does have the power to change a lot of things, and in the next couple of years we may start witnessing shifts in the labor force as AI becomes less covert knowledge, and more democratized and suited for everyday use. There’s already several AI that exists for various industries and niches, and it’ll only get more sophisticated and accurate as time progresses. As Sunit puts it, the only thing that will come out of AI is “progress” in humanity, as has been the case with all new world-changing technologies in history.