Arvind, to start out with, maybe talk a little bit about your journey, your founding journey, your entrepreneurial journey. You've done many, many successful startups over your career. Like I think once first time you met, you said you just every startup you've been at has been very successful. So, just say a bit more about your journey.
Well, first of all, thank you so much for giving me this chance to talk to you here. I really appreciate the opportunity. I'll just introduce myself, my background. So, currently I'm the founder and CEO of Glean, which is an enterprise AI company. Actually, the first gen AI company for the enterprise that brought transformers to the enterprise. We started the company in 2019. Before that, I was doing another startup. We started that in early 2014. And then I spent a large part of my career at Google. And then a few more startups as you were saying before that. So that's my background. Thanks for having me here.
Great. Thanks, Arvind. So, Arvind, from the Rubrik journey, what lessons from Rubrik, which is the data protection company, shaped your approach when you started Glean?
Well, I think one of the things I learned from Google and Rubrik was that as an engineer, before my hesitation in starting a new project or a company would always be that building the product is easy and selling it and convincing people to buy it, building the business is the hard part. Because as an engineer, whatever you know how to do, you feel that's the easy part. But then, over the years I've actually learned that the most difficult part of building great companies is building great products. And if you build great products, then everything else sort of follows. Of course, you have to work hard to build the business and make all the right decisions. And that has allowed me to feel more free as an engineer to actually jump into more ventures.
Makes sense. So, Arvind, is there an emotionally defining moment that comes to mind in your various startup journeys? You know, it's never the zero to one like everybody looks at Glean now and it's one of the most successful AI companies, the fastest growing AI company, etc. But either a heartbreak or was there an emotionally defining moment where you didn't think we would make it and then here you are? You know, everybody always thinks of success as this overnight thing. It's never this linear and straightforward. It absolutely isn't. I mean, startup journeys are very, very hard and difficult. For me, I have an engineer mindset, which is constant questioning and doubt, and that doesn't make it easy when you're trying to start a company because you're constantly doubting your own ideas. And there are all these companies that have been around for decades and are worldwide brands, and how do you really beat them? So I think these doubts have always been part of journeys and there are always big ups and downs.
For us at Glean, for example, when we started, we were an enterprise search company. It was a fascinating problem for me as a user, as a frustrated user who could never find anything I needed when I was doing my work inside my company. It was a great product to build but a terrible business to build because there had never been any success in enterprise search. When we started, most people, I would go and pitch the product to them and they would say, 'Yeah, it makes sense. I have the same pain myself. I know my team has it, but this is not the number one thing I'm worried about.' So we had lots of rejection. And we would put these questions in my mind: are we on the right thing? But sometimes you just have to go back to the starting moment: why did you even start this company and has anything changed? Did the problem actually disappear? When we started, we knew that everybody in the world faced this problem. They couldn't find information. They couldn't get answers to their questions. And that problem is still the same today. Everybody has it. So we just keep going, and that's how you go through the ups and downs.
Perfect. Well, that's great. And what have you unlearned over the years? So as a founder, when you go through multiple startups, I'm sure there are some things you've said, look, this worked the last time. It'll work the next time. But are there certain things where you said, I will not do it at this startup that I did at the previous startup?
Yeah, well, I think first let's be a little bit more human. Unfortunately, most of the good things you learn you forget and then you make the same mistakes again. And I think that's a bigger problem for me. I also keep making the same mistakes over and over again. But in terms of things we explicitly chose to stay away from, I think one of the things which after my Rubrik experience, I felt that it was important to, as a technology company, I really wanted to build a product that everybody in our company uses very actively every day. And it was an important thing for me because I felt that sometimes when you don't build for yourself and when you build for others, there's a distance between how attached you are to that product and how aligned you are to the mission. So that's one of the things that was important for us when we started Glean. Everybody in our company absolutely loves the product because they use it every single day. They all have ideas. The product doesn't work all the time. In fact, it does a lot of things which are not great. But then we keep getting all those ideas from everybody in our company and you can have this collective innovation that happens across hundreds or thousands of people. So that's one thing that I love, and that was one of the things which I wanted to explicitly change from my previous startup when we started this.
Makes sense. So Arvind, for those people that may have not heard of Glean, could you just describe what Glean does? You started with enterprise search, but where is it now? What's your vision for the company and the product?
Yeah, so when we started, our vision was very simple. The way Glean works is that you bring Glean inside your company, connect it with all of your enterprise systems, systems like Google Drive, SharePoint, Slack, Jira, Confluence, Workday. Basically, all the systems where work happens on a day-to-day basis. You connect all of those systems to Glean. Now we go and understand all the information, data, knowledge inside your company. We also understand who your people are, build a deeper understanding of how work happens inside your enterprise. And with all of that learning, we give you a few different products. You can go to Glean and ask questions and we'll surface the right information back to you. As the AI models have gotten better, we can now not just surface information, but actually answer those questions and engage in a conversation. As people started to use Glean two years back, we started to get a lot of demands from our customers who said, 'Look, you've connected all of our company's data and knowledge and information, brought it in one place, and you're making sense of it. We actually want to use this underlying platform directly ourselves to build more applications.' So that's when our agent journey started, and now Glean is an enterprise AI agent platform. You can come to Glean and build agents across many different use cases, for different teams and departments. Since we're connected to all of the information, we're connected to HR systems, engineering systems, sales systems. So you can actually start to build agents for different use cases across all of these different teams like legal, finance, HR, engineering, sales, support, all in this one horizontal AI platform that we have built.
That's great. How many people here work in AI startups or even just use AI in general at work? Quite a few. So Arvind, for the group, there is always this debate. Writer, one of our portfolio companies, ended up building their own model, a 7 billion to 70 billion parameter model for corporate data. But you have been using third-party models. So this whole area of should I build my own, should I just use OpenAI or Anthropic? How do you decide and how should startups think about that decision?
Yeah, well, I don't think any startup should start with building models. I think one thing should be clear: the foundation models are generally more capable than any specialized model you're going to build. Specialized models could be better in the sense that they could be cheaper and as good, and the foundation model may be a little bit faster, but not in terms of quality or reasoning. For that, foundation models are actually the best. So from our point of view, at Glean, our model is to not reinvent. We want to leverage and utilize all the innovation happening in the industry and do things that others are not focused on. So when it comes to LLMs, we actually use third-party models because they are all really good and we have the ability to bring them into our product. In fact, our customers love us because we're able to bring all the models, all the model innovation from across the industry, whether it's from OpenAI, Google, Anthropic, or X, and bring the best of those models to our customers in a safe and secure way. But then there are some models that we build out internally ourselves because one of the ways to make AI work in the enterprise is that you have to deeply understand the business, the people, the processes, and how work happens inside a company. For those, you need to build semantic models using transformers, where you train those models on the specific data inside that enterprise. But that's our model: don't reinvent, use the innovation that's happening out there. And that's my answer to any other startup as well. When you start, you don't have to worry about cost or speed. The number one thing you have to worry about when building an AI application is to make sure it works correctly, and to make it work correctly, use the best model that's out there.
Makes sense. So then, Arvind, the enterprise search market, it's become really hot all of a sudden. So what's your take on what would it take to win in the enterprise search market?
Yeah, well, let me first tell an interesting story. When I was starting the company in early 2019, I think I was talking to some investors in late 2018 as well. I was coming off from a startup that was actually quite successful. It was a multi-billion dollar outcome. So investors in the Valley, their model is that if they see somebody who had success one time, they want to back them. So people wanted to back me to start another company, but there were quite a few who said that all their excitement would suddenly disappear when I told them that I was going to work on enterprise search. Because that was sort of the worst investment that anybody thought they could make. It was a market that nobody wanted to be in.
Yeah, we are not very intelligent as an investor community. So it's actually an interesting thing. That's how far people wanted to be from enterprise search at the time.
