Two Unicorns Built in 12 Years. The Principles Have Never Changed | Glean, Arvind Jain
In a Nutshell
Arvind Jain, founder of Glean—a powerful enterprise AI like ChatGPT connected to company data—built two unicorns (Rubrik and Glean) by solving universal productivity issues like information overload, starting from personal frustrations at Rubrik in 2019. Despite early rejections and search fatigue, strong conviction, internal validation with 20 design partners, and a free beta led to seamless paid adoption in 2022, fueled by word-of-mouth. Core principles: leverage existing tech like GPT-4/Claude instead of reinventing, focus engineering on enterprise context, move fast with agility, and maintain unshakeable belief in the problem despite naysayers.
These notes were generated by AI and may contain inaccuracies.
For a startup, you have one weapon always, which is that you have focus and the ability to move fast, and we have to use those as our primary weapons to stay ahead. It's also strategy: we don't want to reinvent the wheel; we want to make use of all the great technologies available and focus on the rest where others are not wanting to spend time. Customers are looking for solutions to problems; they're not looking to buy great technology. This allows us to focus and dedicate our engineering time on problems that others are not investing in. Focus, investment, speed, agility, and making the right strategic choices—such as not training our own models and benefiting from the industry's innovation—help us stay ahead in this market.
My name is Arvind Jain. I'm the founder and CEO of Glean. Glean is an enterprise AI company. Think of it as a more powerful version of ChatGPT that's connected to all of your company's data and knowledge. You can ask it any questions or give it any task, and Glean will use all of that enterprise context as well as the world's knowledge to answer those questions or complete those tasks. We started in early 2019, making us one of the first enterprise GenAI companies in the world. We have so much opportunity to transform how people work and fundamentally change the nature of how humans are going to be working.
I'm an engineer by training. I studied computer science in college and worked at five different startups over the last 25 years. I worked at Google when it was a privately held company, then we started Rubrik in 2014. I've been in the tech industry for quite some time. I've been an engineer, a programmer—that's my core identity. I never set out to start these two companies, Rubrik and Glean; these were things that happened as I was working through. Primarily, I've always thought of myself as a product builder. I love working on technology and taking complex problems and figuring out how software can solve them.
The journey for Glean started the same way. I was already a founder of a successful company when I started Glean. Rubrik had pretty good success; we were growing fast. As we grew, our company started to struggle on the productivity front. All of our business metrics—how much code we were writing, how much business we were doing—started to look much worse on a per person basis. When we tried to understand what was happening, our employees complained the most that they couldn't find information within the company and couldn't find the right people to help them with their tasks.
This is not a unique problem. Every company has knowledge and data, and every single person struggles with it, dropping their productivity. It was obvious: here's an opportunity to help every single person in every single company become more productive by reducing their frustration. That motivates me. When people complain they can't find information inside the company, we had to solve that problem. I was super excited to build this product, so that led me to create Glean. It was not that I was trying to start a company or be an entrepreneur; I wanted to solve a technical problem and build a product like Glean, so maybe I should start a company.
Search as a product is quite hard to build. You have to bring a lot of data together in one place. Data is in different formats and types: images, tickets, documents. You have to build a deeper understanding of all those different types of data and understand activity and patterns inside the business. It's a pretty complex piece of technology, and doing a good job is super important. Most search systems in the past in enterprises were systems that people didn't like. There was fatigue in that category; enterprises that tried buying search products had all universally failed. We had to battle the perception that enterprise search products are all really bad.
We were building this product and wanted feedback. I would go on LinkedIn and connect with industry leaders to see if they would benefit from the product. Most of the time, nobody responded. I couldn't manage to get even 30 minutes of a stranger's time because I hadn't built the product yet and couldn't show a demo. Being a seller, a business development representative, sending cold outreach through email and LinkedIn was a humbling experience. It helped me develop as a person, building stamina, mental strength, and fortitude. Building a business is never easy. A lot of times you give up because of struggle, negativity, or lack of support.
For me, I had surprisingly strong conviction. I knew how hard it is to find information at work and how many hours I spent. I talked to a lot of people, and everybody agreed—there's not a single person who says it was easy to find information or get answers at work. I knew people needed it, but leaders were not used to buying it. Once we build a good product, the world will come to it. We knew we had only one chance to succeed and had to build a product with very high quality. People's expectation from search technology is what Google delivered: ask a question and instantly get right documents. We were willing to wait, spend the time, make the investment, and only launch when we had a Google-quality search experience.
The problem we're solving may not generate immediate business success, but everybody understood it's an important, high-impact problem. Conviction came from using the product ourselves; we couldn't live without Glean after building the first version. We offered the product for free for the first 2 years. We had 20 design partners, and people raved about it. There were situations where security teams wanted it turned off because it connects to many systems, but there was a revolt—people wouldn't let it disappear. These were powerful signs that we built something useful.
In 2022, investors came to us; we didn't have to convince them. By then, everybody got to use the beta for free, and we had word of mouth. People figured out we built a good product. We told customers we were going GA, so everybody could buy it. They had the choice, and each one bought it. Expectations were set for a long time that we'd start charging, so it went smoothly.
For a startup, your weapons are focus and the ability to move fast. Our engineering philosophy is we don't want to reinvent the wheel. We make use of all the great technology available and innovate where others are not spending time. You don't get to say you're just going to build cool technology and hope people buy it. Many companies are building foundational language models like those from Google, OpenAI, and Anthropic. Instead of making that investment, we leverage them. Enterprise customers love that we are not building models; they want solutions to problems, not great technology.
To make AI work in the enterprise, these models don't know your enterprise or how your company works. To use this technology, you have to bring in enterprise context for any task. Making use of GPT-4, Claude, and Gemini allows us to focus engineering time on problems others are not investing in.
Glean is now over a thousand people. We're no longer a small startup; we're a fairly large company investing more in making AI work in the enterprise than any other company. We are a pure play enterprise AI company with no other product lines or legacy products. All attention goes into making AI effective in the enterprise. That focus, investment, speed, agility, and strategic choices like not training our own models while benefiting from industry's innovation help us stay ahead.
The number one thing for any aspiring founder is belief and conviction. You start with an idea because you thought about a problem that exists. Talk to a good number of people to confirm it impacts many. Once done, stay true to that idea. You'll talk to investors who may be lukewarm, or employees who don't believe as much—that's okay. Keep the conviction because the problem exists for a good reason and still does. Don't become the person who kills your own idea.
Keep EO in your library
Save the videos and channels worth coming back to, and find them again in one place.
