Hello and welcome back to Equity Tech French's flagship podcast about the business of startups. I'm Rebecca Balon and this is the episode where we bring on an industry expert to help us explore a trend in the tech world and dive deep. Enterprise AI is shifting fast from chat bots that answer questions to systems that actually do work across an organization. One of the companies at the forefront of that shift is Glean, which started as an enterprise search product and is now positioning itself as a full on AI work assistant. Glean has raised funding at multi-billion dollar valuation as the market heats up around who will own the AI layer. And Glean CEO and founder Arvind Jain is here with me at WebSummit Qatar to break it down. Arvind, welcome to the show. Or should I say a mosaken given where we are.
Thank you so much for having me.
Thank you for coming. This is so fun. Is this your first time to Doha?
That's right. First time.
It's fabulous. I mean, this is a beautiful place and also great weather, too. So,
Oh my gosh, it's so nice. Where are you normally based?
San Francisco. Okay. I'm in New York and it is minus and it's freezing and it's been so nice here in the sun. So we're really happy to be here. Arvind, talk us a little bit through your background and what Glean's core pitch is now because it has shifted a little bit over the years.
Yeah. So we started Glean in early 2019 so we're now seven years old. Before that I was one of the founders of Rubrik, which is an enterprise data security company. We started that in early 2014 and then I had a long career before that at Google as one of the early search engineers. For Glean itself, as you were pointing out, we started out with the mission of building a Google for your work life. That was what we wanted to do: to make it easy for people to find information that they need, to get answers to questions that they had while they were working. And so make it easy for them to find stuff inside the company. Transformers played a big role in us building that product, so inadvertently we became the first enterprise generative AI company in the world. But over the years, as the language models continue to get better, it helped us evolve. Today you can think of us as number one, an enterprise AI assistant. So think of it just like ChatGPT or Microsoft Copilot or Claude, but something that knows a little bit more about you and your company. So we can do a lot more in terms of helping you with getting answers to questions that require you to tap into your internal company's data and knowledge. The second thing that we also do is over the years to build this really good search product, this AI assistant product, we had to do a lot of sort of platform level work, connecting with your enterprise systems, understanding how your business works. So today that platform is actually also coming in very handy. We let our customers use that directly to go and build AI agents and really bring AI into each one of their business processes. So the company has evolved from a search and AI assistant, which is an end user product for individuals, to an AI platform that now drives AI transformation within a business.
And how much of what you learned from how employees are looking up information about their companies and the way that they work and the information that they need, how much of that has played into this kind of more of an agentic platform service that you're offering?
A big part of building a good AI agent platform is fundamentally you're trying to get agents to do the work that today humans complete in your enterprise. For agents to be able to do that, you have to build a few pieces of technology because the AI models themselves they don't really understand anything about your business. They don't know who the different people are, they don't know what kind of work you do, what kind of products you build. So you somehow have to connect the power of the reasoning power, the generation power of models with the context inside your company. You have to learn from the work that people do, the kind of questions they ask or the kind of tasks they perform. And with those learnings, by bringing that collective human intelligence together in one place, now you're able to bring the right context to agents so that they can sort of perform like humans do in your enterprise. So it's quite critical like you know the layer that we built initially building a good search product required us to deeply understand people and how they work and what their preferences are. All of that is now becoming foundational in terms of building high quality agents.
Yeah. And you know a lot of companies are trying to do what you're doing. They're trying to get into that enterprise AI agentic space. I guess would you say that you have a leg up in so far as you were already kind of there? You're already embedded in a lot of company infrastructure?
100%. And so we are considered the pioneers in enterprise AI. We were the first ones to sort of really holistically think about how your business works. You know, we built integrations to many enterprise systems which is fundamental to making AI work in the enterprise. Search today also is a core component of any AI platform that you're going to buy. So we do have a pretty big lead in terms of technology. We've been at this problem for seven years. Most other companies probably are in this journey for the last one or two years, so that gives us a strong foundation to go and continue to bring our platform to the largest enterprises in the world, and that's what we're doing. We have a very healthy, fast growing business.
Yeah. Well, I mean, it shows in your valuation jump, right? So earlier, sorry, that was last year, last summer, you raised $150 million Series F round led by Wellington that valued the company at $7.2 billion. And that just a year earlier from your last raise, I believe you were at a $4.6 billion valuation. So that's a pretty big jump in a short amount of time. What changed in the business or in the market?
Yeah. Well, first, I mean, I think these numbers will look quite ordinary in the new world of AI. And for us, we haven't as a company, we don't fundraise all the time. We don't do four rounds of funding in one year. We're not building models. We don't have a need for a lot of capital. Even raising capital once a year is actually quite aggressive in our opinion.
What changes have you seen in the past couple of years in the enterprise AI space? Do you feel like it's a total land grab at the moment?
It is indeed a land grab. When we started the company, we were bringing a good AI powered solution to businesses by bringing a really good assistant that could help people with their work. But for the longest time, we had to convince the enterprises that the product we built was useful, that it would make people more productive. And so we were in this what I call the category creation mode. We were evangelizing AI technologies and transformers, and we had to acquire the customers that we would actually get were the ones who were willing to be at the leading edge. But in the last three years, or more in the last two and a half years, that has completely changed. Today enterprises are very eager to embrace AI, to bring it into their enterprise in a very deep way. They want to take every piece of work that happens inside the company and they want to see if there's an opportunity for AI to make those business processes better, more efficient. So we have completely transformed from where we were, we had to evangelize, to where now everybody is looking for a product that we have. But with that also comes competition, and there are a lot of other companies also offering similar products or at least talking about it.
Yeah, I mean we've got Microsoft Copilot, Google Gemini for Workspace, Salesforce Einstein, there's Notion, Atlassian. I mean I just listed quite a few, but who is Glean really competing with on a day to day basis?
Yeah. Well, I mean, you named a few, and I think from that perspective, you can pick the top 10 or 15 software companies in the world. Now they're all AI companies. They all are actually trying to build products and solutions like ours, right? But I think one of the interesting things with how Glean has actually built its technology is that while these big players compete with us, we actually think of them more as partners as opposed to competitors. Today, at Glean, we don't build the large foundation models.
So who do you use for the chatbot purposes?
Yeah, for chatbot or even for our agent platform, we actually use the typical big, most popular language models like Gemini, Claude, GPT, but we also use open source models. And that's actually one of the key strengths of Glean is that since we're not a foundation model company, enterprises actually like us. They like to work with us because we bring the entire innovation that's happening out there because model innovation is not restricted to one company anymore. You're getting a lot of great capabilities across different models. They're all good at doing different things. So with our platform, since you can actually enjoy all the innovation that's happening in open source as well as these closed source models, that's actually something that enterprises prefer a lot. And this is what I meant by that we are not necessarily competing with those model companies. In fact, our product gets better because we're able to leverage the innovation that they are actually making in the market.
Mhm. Yeah. So I mean that kind of answers my next question, which is what is defensible here? If big tech is threatening to bundle AI products that companies are already paying for, is it the trust? Is it the workflow depth? Is it the data access?
Yeah, so when you think about the AI architecture inside your enterprise, there are a lot of things that you have to actually build before you can truly make AI ubiquitous inside the enterprise. Some of those things are, first of course, model access. For you to build all these agents that are going to replace some of the work that humans do, they all need the reasoning and the intelligence of these models. So you need to make sure that you have a model access layer. The second thing that you need is these models don't really understand much about your business. So you have to actually build integrations to common enterprise systems so that when you go inside a company, you can understand how this company works, who the different people are, what's the business, what are the products you build, how does work actually happen. That layer integrating with all of your enterprise systems is something that we are uniquely really good at. And the third thing which is also very important is that enterprises are actually quite worried about AI. If you start to build and train models that connect with all of your internal know how, your documents inside the company, it actually creates a big security risk because within the company a lot of information is not accessible by all employees for the right reasons. So you can't just simply take all the information, train a model, and then let people get answers to whatever questions they have. You have to understand people and their roles, understand what information can be accessed by whom. And so you have to build this permissions aware governance layer and a retrieval layer that is able to, given any task, bring the right information but knowing who's asking that question so that it filters that information based on their access rights. So those are some of the key technologies that Glean has built over the years.
So is it kind of like a wrapper or it's like an additional layer that, you know, if someone asks a question and maybe they don't have the pay grade to see the answer to it, the system might initially generate it but then, similar to how if you're trying to ask Gemini to generate something with Disney it'll block you, is it like that where it will actually understand what knowledge, what data you can access and it will never let you use anything?
Yeah, where it will actually understand you know what knowledge, what data you can access and it will never let you use anything. It won't even, when it invokes the models, it won't even provide that input to the model, anything that you didn't have access to. Because if you do that once, then it's hard to sort of tease apart. You can't build a wrapper after the fact, you have to actually filter information that you actually put into an AI model based on your access rights. So those are some of the key technologies that we build, and that's how we actually complement all the other work that the rest of the industry is doing. But I do want to actually answer this question to you about how does AI actually get consumed in an enterprise. Productivity players like Microsoft and Google, they're going to actually bundle AI in their current productivity suites. People are going to engage with AI in fact in every workflow, every application that you have inside the company. It's not going to be us 100%. I think it would be foolish for Glean to say that whenever somebody wants to use AI they have to actually come to our product and do that. That doesn't make sense. So with that said, then what is our role in this space? We are sort of, think of us as the platform layer that sits underneath these individual UI experiences that gets delivered to people in their email systems, in their document systems, CRM. What we do is we're building an enterprise wide intelligence layer. As I mentioned, we connect with your internal systems, we figure out the governance of all of that information inside the company, and we're sort of bringing the collective intelligence, human intelligence of your enterprise in one place. We are making deep observations inside the company on how work happens inside your business. And then our idea is to actually make all of this knowledge, this intelligence available to those AI experiences that may be delivered by others. So we serve as a middleware platform behind the scenes and bring the right information, the right data to all those verticalized AI experiences that you're going to be building.
