Arvind Jain & Rajan Anandan: Scaling Rubrik, Glean, and the Future of AI

SHOW NOTES

• A company is all about its people [01:39] 
• Adapting to challenging times [03:34] 
• Team alignment matters: Lessons from Glean & Rubrik [05:46] 
• Building fast is important but quality is key [08:52] 
• Product-led growth vs top-down sales [11:30] 
• Boosting sales while creating a category [14:55] 
• Riding the trends and using tech to iterate [19:04] 
• Rethinking moats as learnings from customers [24:30 ] 
• Audience Q&A excerpts [27:31]

TRANSCRIPT

Rajan Anandan: Arvind is the Founder and CEO of Glean. He founded Glean in 2019 to make it easy to find information in companies. But prior to Glean, Arvind co-founded another extraordinary company called Rubrik. So, you co-founded Rubrik in 2014. That’s a company that’s on its way to being public, and Glean [was founded] in 2019. Two extraordinary companies, both within a period of a decade. This is really rare and I wanted to just get right into it. What are the lessons? What did you learn? Like what’s most important?

Arvind Jain: One thing that I learned building a company, it feels hard, but I think if you keep working, things start to happen. So I think, whoever is actually having second thoughts and feeling, “Oh! this is too hard,” I think things will get easier maybe. Because sometimes people say things never get easier, but I feel things happen if we persist and if we keep working. So that’s one learning that I had because I went through the same journey when I started Rubrik for the first six months, for the first nine months, I kept thinking, “Did I make a big mistake?”

Rajan: Why is that? What happened in the first six months? 

Arvind: Well, partly nothing happens! You don’t get customers because you don’t yet have a product. You know nobody’s going to actually pick up your phone. Like you don’t try to talk to them. You’re a nobody. They’re not interested in spending time with you. 

Rajan: You go from working on a billion-user product to…

A company is all about its people [01:39]

 Arvind: Yeah, right. And I have some really interesting learnings to share from there. But I think…that’s the number one learning – these days, whenever somebody asks me for advice, that’s the one thing I tell them, “Don’t give up. Don’t lose hope, just persist. And, if you keep working on a problem, ultimately, the solution arrives, and things work.” 

The second thing I would say is that the company is all about people. And you have to build a really good team. I’ve seen entrepreneurs who say that, “Yes, I have this great idea. This is the business problem I want to solve.” And I asked them, “What’s your team, who are you working with?” And they say, “Oh yeah, I have three people who are part-time in Pune working on this with me.” And I say, “That doesn’t work.” Like, you have to actually have folks really world-class. 

Initially, your company is not about your idea. It’s actually about the people, and building a great team is hard. And again, you have to work very hard to… I  spent, for example, in the first year of Glean, I was a recruiter. That was my main role. I was a recruiter for the company. I didn’t actually get to write code because there were other people who do that better than me and they wrote it. And my job was to recruit. I’m on LinkedIn connecting with all the engineers spending most of my day on that – trying to convince them about why they should actually come and join Glean. So that’s one lesson. 

Second, the journey is hard. My second year was actually when my role at the company changed and I was the SDR (sales development representative) for our company. That was the thing that I spent the most time on in the second year of our company because now we had a product that we were ready to actually go and…

Adapting to challenging times [03:34] 

Rajan: But, you didn’t come from a revenue background, right? And at Rubrik, Arvind, you were more towards building the product and the tech. That’s right. Now all of a sudden you had to…So talk a little bit about that because a lot of our founders, it’s something that they really wish they didn’t have to do. 

Arvind: Yeah. So, I’m an engineer by training and I think I still feel and identify, more as an engineer, not as a CEO. And so that was the first time I had to do it. I’ve actually…and initially, I felt that this is so hard and this is something that I just absolutely don’t want to do. Like I’m actually pleading with people, trying to get [or say] give me half an hour of their time so that I can share with them what our product is. 

But over time, you know, I started to actually enjoy it, enjoyed first just as a challenge. I think I just did that as something that’s hard. I have this sort of drive within me to go and try to solve it. But second, I think there’s a lot that you learn from talking to people, going through these processes when I would get time from folks, there’s an art for how you talk to customers. Like, initially as an engineer, I was like, I would talk to them and I would just be excited about telling them about our product, not really listening to them about what they need. I was actually more interested in telling them what I had and over time I learned those skills.

So I think I would say that as an engineer I think it’s also great to ultimately acquire those skills to go and understand what’s the right sales process. Like, how do you actually convince customers to buy? To buy products from you, these experiences help you. You’re not going to do that for a long time as if you’re an engineer or a product person – ultimately you’ll hire sales leaders and folks who are going to do those things for you, but you’ll be able to lead them much better if you actually go through some of those experiences yourself early on. So for me, it was great learning. Like now I feel like I do know a little bit about how to sell.

Team alignment matters: Lessons from Glean & Rubrik [05:46] 

Rajan: Great. Awesome. Those are a bunch of good lessons. Now let’s focus on Glean. Where did the idea for Glean come from? Tell us a little bit about it. 

Arvind: So the idea for Glean came from Rubrik. We were actually quite fortunate at Rubrik. We grew fast as a company. We were about, more than a 1000 people in four years, which is incredible. But as we had that hyper-growth, we were growing fast because we had a huge market and we had a good product. And so we wanted to do a land grab. And so we grew fast and as we reached that scale, we realized that work was not happening anymore in the company. People were struggling. Productivity was low. An engineer, who was writing 300 lines of code a day before, now was writing 50 lines of code. 

Same for salespeople, like they’re not able to sell as much. So we’re going through a big struggle and some of it is natural. As you grow, and become a large company, things start to slow down, there are more discussions and more processes, but we felt there was more to it.

And so we would ask people… part of building a company is always learning from your team about what things are broken and how we can fix them. So in our pulse surveys, we would ask people, I mean one of our questions was, “What can we do better? What are the ways, what are things coming in the way of you doing good work?” And the largest complaint that we would see from people was people saying that, “Hey, I can’t find anything in this company. I don’t know where to go and look for information, but I need that. And I also don’t know who to go and ask for help when I need help,” and so this is a big problem. 

And as you mentioned I’ve worked at Google for a long time. I’m a search engineer. So when people are saying, we can’t find things, the first thing that comes to my mind is, “Hey, we should actually have a search product.” I was not surprised because we’ve built, in this modern SaaS era, we had 300 different SaaS applications for the business…

Rajan: Is that the core of the problem, that companies have hundreds of these SaaS applications in a silo?

Arvind: Yeah, there are the data silos that a large number of applications create. But there’s also just simply, there’s a lot of knowledge. Knowledge is not something that you really keep up to date, you don’t curate it. You don’t delete out-of-date stale information and [it] just keeps accumulating like this. 

The knowledge growth is exponential in companies. So, ultimately you land in this space where there’s so much information fragmented across so many places and it’s hard to find those things. So this was the problem that people were facing. I was facing it. You know, I had an assistant like, like whenever I needed something, I would just ask her, “Hey, go find this for me, or, bring the person who works on this, so that I can talk to them.” So I had that luxury to actually have some assistance. She would do it, and she would have all the struggle, but I understood the problem. So we tried to buy a product for this. We said let’s go buy a search engine that can connect all of our company data. And we realized that there’s nothing to buy. There’s no product. 

Rajan: Interesting. 

Arvind: And so, that’s what got me excited. And then ultimately, we decided to build Glean.

Building fast is important but quality is key [08:52]

Rajan: Got it. Now, when you started Glean, actually both Rubrik and Glean, you spent almost two years building the product. In the modern era of startups, which is, to get something out there, iterate, and process not at all what you did. So maybe talk a little bit about that, Arvind, this idea of building very deep products, taking the time. Maybe talk a little bit about that, but I think it’s so important for founders, especially building, like deep products. 

Arvind: Yeah. From a desire perspective, I wish we’d also built a product in three months and brought it to customers. So sometimes, it becomes hard to do that, so I would not advise anyone to take more time, [and] to go and build the product. If you can actually bring, if you can build something which adds value, the faster you can and you know that you can bring it to your early users, your partners, the better it is. But it’s okay. Sometimes, you also have to be careful and balanced. You don’t actually want to take something which is not good enough, even if you’re taking it to your closest friends who are going to be your customers or design partners. You don’t want to underwhelm them on Day One. That’s my philosophy. And so you have to be careful. You want to get feedback early, but you want to actually always tell your customers that you’re doing something special, something good. First impressions, as people say, are the most important. But I think when the challenge is larger, like for us, we took more than a year in both cases to actually bring our product to our customers, there are different complexities. 

In Rubrik, for example, it was a complicated system, with hardware with the software layer running on it. And the building, a lot of integrations with different enterprise systems. So we try hard to actually figure out what is the MVP (minimum viable product). We never made our MVP unbounded. We tried to keep it small. It’s just another, it’s unfortunate or, you can say that, but it still took us so long to actually do it. 

Product-led growth vs top-down sales [11:30]

Rajan: Got it. Now let’s maybe shift gears a little bit to GTM (go-to-market). So you could have gone down the PLG (product-led growth) path but you decided to go down the enterprise-sales path. Why did you pick the enterprise-sales path? That was obviously also the harder part for you because you didn’t come from [there]. 

Arvind: Yeah. I think PLG is really appealing but if you think about the stages of the company, you don’t have to think about PLG in my opinion, while you’re building the product. The first year, it doesn’t matter whether your sales motion is going to be PLG or top-down. What you are looking for is the first 10 to 15 customers who are going to be using a product and you’re going to actually be spending time with them directly. So it wasn’t a relevant question [for our growth]. So we didn’t think too much about it in the first 10 to 15 months. But then as you started to learn from them, we realized that PLG would not work for us. So it was just a…

Rajan: And why is that? Just say more. 

Arvind: So for us, our product is a search engine. It’s like Google or ChatGPT, inside your company. So for it to be useful, even to one user in your company, it has to connect with all of your company information. Because I have access to a lot of information inside my company. If I want this one place where I can go and find it, anything, I have to actually do that. So first of all, it’s, so it’s infrastructure-heavy. To actually add value to one user, you have to index all of your company information. So that was that. So that was one fundamental issue. Like, it’s a costly system to build and you can only make it work financially if you have a large number of users using it within a company. 

And second thing is, for us, we are connecting all of your enterprise systems. And so there are security considerations, where we need permissions from the administrators within the company who will give to us so that we can connect to those different systems. So that combined [with the fact that] that there are actually quite a few search companies that started around the same time as we did. And all of them followed the PLG model because that was the only model that was fashionable. That was the only one that everybody wanted to do. But it didn’t work for those reasons. And so all of those…

Rajan: The product just doesn’t work right? And you can’t, as a developer or a salesperson, download it, and it’s not connected to your system, so how do you..? What do you do?   

Arvind: Yeah, and then it’s painful. You can always ask users to connect, but it’s painful and people won’t do it. And, then once you’re not connected then you will have this complaint that I can’t find things here. So it was, so I think people did try PLG in this area and it didn’t work for them. I think for us, it’s working, but we are still like…it’s a question that we always have every year, “Can we get some momentum because PLG is an amazing lead generation, capable, channel that you can have.” 

Rajan: I see your eyes light up when you say PLG. 

Arvind: Yeah. So it’s something that we would like to do and are still trying to figure it out. 

Boosting sales while creating a category [14:55]   

Rajan: When did you know you had product-market fit at Glean? And what was that moment? That week when it’s like, “Whoa, okay? Do you think we have it?”

Arvind: Yeah. So two things, for me, I always fundamentally believed that we were solving an important problem. And why? Because whoever I go and talk to, they would actually say that, “Yes I’ve spent a lot of time trying to find things. It’s hard for me.” I never met a single person in any company in the world who said this was not a problem for me. Huge problem. But at the same time when I try to go and sell to them, almost everybody would say, “No, I’m not buying it.” Because, they would say that this is important, but not the most important thing for me. 

Rajan: You faced that. 

Arvind: Yeah. We faced that 100%. Everybody…see, by definition, if you think about when you try to do a category creation, you try to build something that people are not used to buying. There is that natural argument that they have, which is, “Oh yeah things are working for me without it. I know how to live without it.” So when somebody can live without it, it makes your sales very hard. 

Rajan: So how did you come over that and get over that? 

Arvind:  So we had to evangelize. First of all, our goal was, “See to actually create urgency in this market, to find buyers, you have to first find leaders who are progressive, who want to actually become heroes in their enterprise, who want to innovate.” So we worked with the world’s best companies and the CIOs (chief information officer) there and told them that, “Here’s an opportunity to do something really meaningful for their company. Yes, this is not a tool. You don’t have a budget for this. But this is an opportunity to be forward-thinking and make profits.” So that’s how you appeal. That’s the argument that you use with the best leaders and they understand it. They know that they themselves… They were also feeling the pain. And so that’s how we started to slowly evangelize and build our product. And once we have a good product… So people liked it once, and they got a lot of love. So actually, a lot of CIOs have come back and told me that, “I’ve not received this much love from my team, in my entire career,” because the pain was so big and when it actually got solved, they got those kudos.

So that’s how we started. And we’re still, I would say that we didn’t actually, I would not say that we did a fantastic job. Like, for example, I think we under-invested in marketing and so we were still basically building this awareness at some pace, we should [have] done more but then we also had macro trends, that helped us.

So there’s something like, a large company needs a great team, great people, great product, but also needs luck. And I think for us, we had multiple macro trends, when you basically keep track of them and see how they actually help you. So one of them actually is the shift to distributed work, which was driven by the pandemic. So once people started to work from home we were able to [use that]. We were in this mode where now the CIOs were thinking about, “How do I enable my people in this new model of working.” And when you’re alone, at home, there’s nobody to go and get help from, a product like Glean becomes more valuable. 

So it helped us and that gave us some tailwinds. And then, of course, AI. So as AI technologies were getting better, it gave us this opportunity to really make use of those technologies and build an amazing product. And so our product evolved from, looking like Google inside your work life, to looking more like ChatGPT inside your work life. And so that also helped us quite a bit, in terms of how everybody, all the CIOs, are trying to actually buy AI products. 

Riding the trends and using tech to iterate [19:04] 

So you, as a founder, as an entrepreneur look at what the market trends are, and what people are looking for and you have to make sure that you pitch your product to  appeal to what’s the need of the hour or what people are looking for. So everybody has to keep evolving. 

One of the things that is very interesting, that has happened this year because of AI, is people are looking for this thing called RAG, retrieval-augmented generation. So it’s a mouthful. It’s a, they’re looking for a RAG they’re not looking for a search engine. It’s really one and the same thing. If you think about it, the RAG engine is actually a search engine. But that’s a term that has basically caught attention this year. And people want to build AI applications as part of that. They know RAG is a core part of it and they want that. Now we were actually – we always called our product as a search engine for work. But that’s not interesting to people. So we’re now going to say, “We’re a RAG engine.” So you have to definitely look for what the market trends are and if you can capitalize on those, that is what gives you that extreme momentum.

Rajan: Got it. Let’s now, maybe that’s a good segue to start talking about AI. How have LLMs (large language models) changed Glean? When you started, you had BERT (bidirectional encoder representations from transformers) for semantic matching. What changed with LLMs?

Arvind: Yeah. So, we’ve been using language models in our product since 2019. For those who don’t know this entire AI trend that we’re seeing right now, it originated in Google. And the reason these models were built initially was to make search better. That’s why these models were built in the first place. And Google had already put some of these really good models, the BERT (Bidirectional Encoder Representations from Transformers) family of models, in an open domain that we could use even in 2018 and 2019. So from day one, we’ve been using these models for semantic matching. 

The models were not good enough at that time so you couldn’t actually get them to write, for example, sentences, but they could still understand, conceptually, what a document is about, or what a piece of text is about. So you could do really good semantic matching. I could go and ask for a product manual for my car. And if a document says it’s a user guide, it would, BERT would understand that these two are the same things. So you were able to use these models behind the scenes in search, in many different ways in the core search technology. That’s how we started. And so it allowed us to actually build search, which is, which was, which is better, which didn’t have to be restricted to exact keywords and things like that.

But then, we saw the clear trend that these models were actually getting better faster than anybody could imagine. And I think it was real, at least for us not keeping up with the market as much. We saw a distinct shift where there was a moment, binary moment, where you saw that, “Oh! Now, these models can actually write and they can write coherent stuff.” And that’s amazing, in our product, because our product is all about: somebody has a question and we have to answer that question. So we saw this opportunity that now we can advance our product just the same way, as you think about Google versus ChatGPT, we could actually when somebody came and asked a question, we could go beyond just surfacing the relevant information to them. You could actually have AI read it, read that information, generate answers and give it back to users. So it’s a big enabler for us. So LLMs are actually helping us make our products better. And interestingly, there’s also the flip side of it to actually make LLMs work in the enterprise.

When ChatGPT came out, everybody wanted something that was inside their company. And so they wanted to train models for your own enterprise. And that proved to be very hard, and very difficult. Like nobody, I don’t know of any company that has big success in training large-scale models. Companies don’t have data or the data is complicated, it’s private, and there are security issues. So model training doesn’t work. So then how do you make use of this technology in the enterprise? And that’s where the RAGs come in. The only way to make AI work in your enterprise is to follow this RAG architecture today, which means that search has now also become a foundational piece in making AI work. So while search is getting better with AI, search is also a fundamental component of how AI can deliver value to the enterprise. So that actually, made our product even more useful to enterprises because now it’s no longer just an end-user product for users. It’s actually an AI platform.

Rethinking moats as learnings from customers [24:30 ] 

Rajan: Moats. Yeah, so how do you think about moats for Glean? And then also if you’re a startup building, an early-stage startup trying to build an AI company how should one think about moats?

Arvind: Yeah. So I have a different take. I don’t believe in moats. 

Rajan: Alright! Okay.

Arvind: So, what I think, and I think this may be different from the advice that you get from others, personally, to me first of all what more do we have, for example, against Google? We’ve got nothing. They have 100 times more people, all the great smart people, resources, brand, they have everything. We… I don’t think we have anything, any secret sauce in my opinion. I think all we have is a drive. And a desire to build a product and building a product is always hard and it takes time. And no matter how large you are, to actually build a good product, you have to learn from customers and those learnings happen over time. They cannot happen instantly. And that’s what I think is your moat is the fact that you actually spend time, and understand your customers. And you spent five years working on it in those five years, as we tried to build a search product we learned so much about enterprises and the challenges that people are going to run into. So anybody trying to build a search engine right now is going to take them another five years to actually go and figure those things out. So my belief is you have to look at the problem that you’re solving, you have to build a product. As long as you build a good product, work hard on it, and have a good team you will succeed. So I don’t think of “Hey, like I’m going to change my product roadmap so that I can have a bigger moat.”

Rajan: So you don’t worry about data as a moat and all this kind of stuff.

Arvind: I don’t and well, I think, partly because I guess I have a builder mindset. I’m an engineer, so I don’t actually think about strategic topics, as much, so it might just be my weakness, but at least, that’s how I think, we don’t have this, but it becomes apparent so now, for example, one of the strengths of Glean, of course, is that we have all of this enterprise data that we brought together in one place. And, it’s a hard thing for somebody else to go and build and take the time.

Rajan: I think what you said, Arvind, which is, “Look, if somebody starts from scratch today, it’s going to take them five years.” That is, you’ve been working at it for so long I think maybe that’s the, you know, this idea of building a product [is that] it takes about two years to build a really deep product. That itself because somebody else starting from scratch is going to have to spend their two years [on it] and then you’re always ahead because you’re…

Arvind: Exactly.

Rajan: Allright, let’s open it up.

Audience Q&A excerpts [27:31] 

Audience: So, how did you go about finding the ICP (ideal customer profile) because it is a very horizontal product that everyone in the world can use. How do you go about defining, okay, these are the right set of customers for me and these are not?

Arvind: Yeah, that’s a great question. In fact, one of the most challenging things for us is that our product is useful to every single person in every single company in the world. So who do you focus on? And so sometimes, this huge market, this huge opportunity actually becomes a liability because it can easily make you defocused. So for us, what we chose was, first let’s go and target the segment that we understand more. I’ve worked in technology companies my whole life. And I know this is a problem there. So we started with that as our first, vertical. And within that, you can then think about, do you want to be a large enterprise or mid-market or SMBs? We picked companies with around a thousand people as our segment because we felt that first, they can, we will get a good number of users from them. Second, they have a real problem. And third, they actually have money to pay us, and fourth, they will be agile and fast. So that was the reason why we chose that as a segment. And when you grow up once you reach some success there, then you start to grow from there. 

Rajan: Got it. Mic here.

Audience: So I have a question about how you classify yourself as a RAG [based company]. So RAG is more, from what I see, more on unstructured data. That’s what [it is] generally, and I guess primarily what you’re targeting as well. So I’m curious to know that many enterprises have information locking as structured data in databases, etc., and when you build on top of that, you have to have more complex concepts like knowledge graphs and semantic layers, et cetera. I would like to know more about if you see these actually happening there are companies and the trends in this direction. 

Arvind: Yeah. So you’re right. Enterprises have lots of unstructured data in the form of conversations, and documents. Then there’s a semi-structure. There’s a lot of data that is structured, but [there are] some unstructured parts, for example, tickets and there is data in salesforce. And then there’s like the real data and databases. Glean will actually connect with all of those pieces of information, whether it’s structured or semi-structured or unstructured. And then we’ll help you leverage AI to provide value to the data. But there are a lot of information-seeking tasks in enterprises that require you to do complicated analysis of structured data, and that’s not a business that we are in. There are companies like Snowflake, Databricks and ThoughtSpot and actually many startups we’re seeing that are actually trying to build these new AI-based structured data analysis products right now. So there’s a lot of things there. 

Rajan: Go ahead.

Audience: So I think enterprise search would have come earlier also because search engines were there earlier. Why do you think this problem had to be solved now instead of 20 years earlier. 

Arvind: First, I think, it’s not solved yet. We have a long way to go to actually build a great product ourselves. A bunch of companies—there were quite a few companies—in the early 2000s in enterprise search, but they didn’t succeed. It was actually partly because I think in the enterprise, it was very hard to get hold of the data, they used to be in these custom systems in the pre-SaaS world with package software with 25 different versions. So any search company that was built found it very hard to actually build a turnkey product in the past. Even Google had a product, in fact, that product predates even the advertising revenue that we got at Google, but it never became big because it was just so hard to actually get value in it. Like, only the largest companies in the world built an internal search, experience, and that was more like engineering exercise, not buying a product. So there were like those technology trends which made it hard in the past. I would say cloud, AI, open source, and search technologies, are some of the things that actually made it possible now for a startup to actually solve this. 

Binny Mathews: Arvind, thank you. I’m Binny. I’m the co-founder of ProjectPro. We help analytics teams ship faster by giving them a repository of reusable assets. So I have two questions. The first question is, what are people searching for inside enterprises? When you look at the queries, what are the most common things? And, the second question is, in the early days when you’re trying to sell, would specific teams come to you asking for customizations to map your search results closer to what they are used to in a Confluence or a SharePoint? And how do you address those customization requests and things like that? 

Arvind: Yeah, I think what people are looking for, lots of different types of things…you know, there are some common needs in enterprises like people looking for corporate information, the benefits, IT-related information, but then different functions are different needs. Salespeople will typically try to look for information about customers like product features engineers use Glean a lot to debug problems that they run into. So yeah, it depends on the persona and a lot of different needs. And actually, it varies a lot, company to company. 

In terms of customization, that’s a great question. Especially once you start to sell to enterprises, every enterprise will have different requirements. And, a lot of companies struggle to meet the demands of their customers. They’re bringing revenue but then taking the product in one direction, and other companies taking it in another direction. So you have to have that discipline and confidence to tell them, “No.” I think I also struggle. My product team actually constantly tells me that I keep saying, yes to all my customers for things that they want. So, yeah, I think you just have to maintain a good balance where enterprises feel that yes they have an influence on your roadmap. But at the same time, you build something which is like, I think if you make this rule that you’re not going to actually have any lines of code in your system, which is meant for one customer. There should only be one release, one binary, and one system you will take requests from them but then combine them into a common set of features that will help multiple customers and then go back to them. So that’s [the thing] just striking the balances is important there. 

Rajan:  Okay. Question from Nikunj Bajaj. 

Nikunj: Hi Arvind. I’m Nikunj, co-founder of TrueFoundry. We’re building an ML (machine learning) platform for developers. The question is that Glean is building, let’s say, “A managed search experience for enterprises, right?” You plug all your data in and you get one final search box out. We’re also seeing pockets of developers trying to build out more specific search experiences for some specialized data sets where they have some Jason, where they have to build out some parts, some PDFs with charts and tables, etc. Do you see a lot of that happening? And is Glean trying to support that or that falls outside of the scope? Like, how do you think about that? 

Arvind: Yeah, so certainly like there are a lot of needs. Businesses are trying to build AI applications for specific workflows that they have. For example, in a business, an HR team, won’t actually bring AI as the first responder in their, let’s say, HR help channel in Slack. So they have this need, they don’t want a search or an assistant that actually answers questions using all of the company knowledge. They want this first responder to work only on that specific, curated HR information that they wrote and use that to answer questions. So, there’s a lot of demand for the search in specific, smaller, constrained environments within enterprises. And so we do allow that, using our AI platform. So you can take Glean and create these different applications for each one of those specific functional needs.

This transcript has been edited for clarity.