In AI Agent's World - Emrecan Dogan on Building Glean's Next Gen Search and Productivity

So imagine uh a company being run on more than 100 different um IT apps uh software like SAS solutions each one having um its own approach to identity permissions whether it is person based or group based each user having different user identifiers in different systems. So resolving this to understand who is Dimitri and how is Dimitri presented in different uh pieces of software. What is the information? What is the data Dimitri should be allowed to see and what shouldn't be uh provided to Dimmitri understanding this in real time. This is really the core of what Gleam brings to the table. >> Hi my name is Dmitri Bonichi and I'm a content creator, agency owner and AI enthusiast. You're listening to the AI agents podcast brought to you by Jot Form and featuring our very own CEO and founder Idakin Tank. This is the show where

artificial intelligence meets innovation, productivity, and the tools shaping the future of work. Enjoy the show. Hello there. My name is Dmitri and welcome back to another episode of the AI Agents Podcast. In this episode, we're talking to Emer Dogen, the head of product at Clean. How you doing Ember John? >> I'm doing great. Thanks a lot for having me, Demetri. >> For sure. It's I'm doing I'm doing good. You know, I always tell everyone I'm living the dream and it's it's never a lie. Uh because, you know, when you're getting to interview cool products like yourself and those who are a part of cool products like yourself, doing stuff with AI and AI agents, we uh I am living the dream. So, just to kind of get things started, you have a pretty interesting background working at a lot of different companies. uh a couple

of I mean all of what I've seen at least in in the last like decadeish uh at pretty large uh and interesting companies most people would know. So for for first and foremost could you give us a little bit of background on on who you are and kind of how you got to working to be working at clean. >> Yeah would love to again thanks for having me here. Uh I'll kick it off. Um so this is uh I currently run uh the product or at Gleen. Uh so that is product management, design, data and operations. I've been here for over over two years now. Uh but first and foremost um I'm an entrepreneur actually the longest I have worked for like any company in my career is less than four years about four years and that was LinkedIn and that was the reason was

I sold my uh company uh to LinkedIn after running it for about six years. Um yeah. Um and actually as a matter of fact um I also worked at Stripe for a couple years and while at Stripe I actually stumbled upon an idea that I thought hey this is great enough that I will go back to uh being an entrepreneur again. Um I left I started building my company for a few months and it was going to be some sort of a Glean competitor without me knowing about Glean at all. So luckily um a venture capitalist I knew from my earlier founding years told me hey I will look into Galen um if you were building something around this area. Thankfully I discovered what Galen was doing at the time and I got really scared about the amazing products um the team was building. So

I kind of folded my startup and reached out to Gleen and said hey I can come in and uh build for you. Um and the rest again over the last two years and change I have been here building the product for them. >> That's really awesome. Um so how then I guess if we can go back a little bit more uh on the on Glean side of things. How long is just so everyone knows could you give a little bit of a background on Glean uh like its start and what it does? >> Yeah would love to. Um so Gleen's history uh goes back about six years and change and uh what I loved about uh Glean at the time was um the team started by solving um what data what I uh think as well by solving the hardest part of the enterprise AI

puzzle or problem and it is finding and understanding um company knowledge across every app across every person, every office, every project uh with the hybrid search engine and the very deep um and fully autonomous knowledge graph. And with this, Galen was able to personalize um the experience of finding the right answer while strictly enforcing permission. So um and what was amazing about Galen's uh inception was they were quite contrarian about uh the problems to be solved while these problems were not sexy top of mind um or big markets to capture um and u with that strong foundation um Glean was I think in stealth mode for over a year building a singular solution again for enterprise search and And uh when Glee launched his product, it was an overnight um I think product market fit with the first few uh customers and these customers being

very large enterprisegrade uh players not the startups or small businesses that we are used to hear uh about. >> Okay. So, you know, I I think in the context of of what we've covered on this podcast, we haven't probably had a product like yours on here, which is which is obviously great. Um, there's been a lot of really good growth and and obviously investment in in Glean over the last couple years. I saw, if I'm not wrong, in June, you guys secured another round of like series F funding if I'm not wrong. Right. So, the company the valuation keeps going up. It keeps doing better and better, I'm guessing. and obviously looked into it. The reason behind a lot of that is the innovation you are pushing on the front of AI agents and AI assistants. Um seems like you know you're able to interconnect

so many different aspects of the knowledge of a company uh with what you're mentioning. Uh and I'm just curious you know obviously as the head of product you have a lot of insight into how it's being built. what are you doing on the AI agent and assistant front to really like push the boundaries of your industry? >> Yep, love it. Um, so a few things maybe that distinguishes glean um from the from the rest and I will go through maybe a few ideas but uh please stop me if you want to dive into absolutely any one of these. Um so number one from the get-go from day one we are built for the largest enterprises out there right um sure so if you and I we start a business tomorrow uh with two people two founders we might not like we will not need glean

you and I we will know everything about our startup it's not like we'll need that smart AI company or agents that much but if you think about scaling this company all of a sudden you have 20 people 200 people 2,000 people um information bifurcation or asymmetry um starts um exploding. So it grows um nonlinearly. What a single person knows about how the company operates, what projects are most important, uh what the company is investing in different areas, how to make the right decisions, etc. This these become exponentially more difficult. So um Glean is um built um on an enterprisegrade stack when it comes to handling the scale of information. That's one. But second, maybe as importantly, if not more, is this whole um aspect of governance and permissions. presented in different pieces of software and such that when Dimitri wants to accomplish something what is

the information what is the data Dimitri should be allowed to see and what shouldn't be uh provided to Dimitri understanding this in real time um reflecting all the changes happening in the underlying document permissions this is really the core of what Gleam brings to the table and from that core Then we start building. Okay, let's build um let's offer um enterprise search so that with one UI you can actually see everything the company has to offer that you are permitted to um access. But then on top of it then we build AI assistant where you are no longer finding documents or answers but you are actually um tapping into synthesis or you are generating new content um off of that search. And finally, the AI agents boom where um I think the pendulum swings from generating answers or getting answers to actually executing tasks and

getting AI to do things uh for you and to end. >> So um >> yeah, >> no, keep going. No, you're on a roll there. Keep going. >> What does that look like? What does that look like for executing things end to end? Yeah. So there are uh there are I think um two aspects that we um we should think about. One is every knowledge worker um they kind of have their own unique ways of doing things and being valuable to the company they work for. like you and I if we are um if we are tasked with the same function let's say in marketing or product management still the way we show up to work and then get things done they will have some slight differences or major differences right so um imagine the personal side of glean uh to be like the following

gle enables you uh to have an agent or actually set of agents just like you would have direct reports and And over time they actually understand you better and better. How you idiosyncratically do things. How you respond back to emails or select messages that you receive or when you are writing a strategy document or reacting to a Jira ticket how you would approach solving that problem versus um your colleague A or colleague B and reflecting on all that richness all like the full dimitry you bring to work. um glean agents at a personal level um start working in that way. So for the lack of a better framing is really um your team of agents that gets trained with the way you work so that um over time you actually rely less and less on your manual effort of doing things and then you can

put your own personal agents into work. So this is one major aspect of glean where we understand every employee um deeper and deeper how they do the work and the team of agents that employee uses um gets to do the work the way that employee does the work. Contrast that with what we call the departmental tasks or processes, right? So, and there are certain things in the modern enterprise where regardless of how many employees you have in a given function, let's say sales or marketing, you want a singular common way of doing things. Like imagine uh releasing um documentation or release notes about a new product launch. You don't want 100 different product managers uh to release notes in 100 different ways. You want a singular uniform high standard way of articulating what is being launching, what is being launched, what the launch date is

and how precise or concise the information is. So there is this whole aspect of departments um or business units implementing a singular agent that every employee in that function can use in a consistent way hundreds of times every day. So that's the other aspect of Glean where we power execs or department heads to build agents that can run for the entire company in a consistent way. >> Okay. So, uh, this is probably the part where I ask, I mean, you have some companies who are out there and, you know, there's different there's different levels of companies that are watching and listening to this >> podcast. What kind of level of companies are you working with like uh, size-wise? >> Yeah. Um, we work um, across the gamut of um, all businesses from SMB to enterprise. But maybe I'll take your question in the following sense.

um where is the sweet spot of uh Glee? Um the larger the company, the more value they unlock uh with Glee. So another way of saying this is um the AR the revenue we are making um it is coming from not tens of thousands of customers, not even thousands of customers but actually in the order of hundreds. Right? So these are some of the most iconic um businesses out there with hundreds of thousands of employees and in many of these Galen is deployed wall to wall. So all the employees within that company across geographies functions um are using Gleen on a daily basis. >> Okay. So, when somebody comes to Glean, um, are you, cuz this is what's interesting to me is a lot of the companies that will be on the podcast, it's a mixed bag, whether it's like made for you solutions or

customuilt types of things. This seems to be in the realm of kind of hard to make it uh made for you, I'm sure. But maybe there's a mixture there. Could you kind of walk me through because obviously you're pretty horizontal right from a product standpoint. So whether that's whatever industry that's what that means people um you can help them out. So is there any specific type of industry that you most find has found value in what you're doing at Glean? and how are you able to if you're able to give any examples uh kind of implement into their daily workflows? >> Yeah, would love to. Um so I think you are spot on uh by the way on this um let's say tradeoff or tug of for between the larger the customer you go to than the more custom or bispoke the solution uh might

need to look like but Gleen happens to be in this I will say magical sweet spot where we get to build a very standard core product that many many customers find value from day one and it's very visible value they turn on glean and then they start getting the value. Um I think one secret source there is this notion of uh what we call knowledge graph. So what Glean gets to learn about all the subjective ways or idiosyncrasies of your organization that is not just like bits and byes um in your data store but the derived understanding of what project names mean, what acronyms mean, how one team relates to another and a given employee. what are the formal or informal belongings the employee has to different offices, site locations, um teams etc. So the magical buildup of that knowledge graph takes care of a

lot of the idiosyncrasy you would expect from a large enterprise and it is so well ingrained in the core product that the product starts working for even the largest enterprises despite um all their custom bespoke needs. That's like one aspect. Um and the second aspect and we'll get into that I think uh deeper is the whole notion of agents. It is actually an a very versatile way for these large enterprises to take glean take the core product get a lot of daily value from it but then deepen its impact by customizing what glean does for each employee each function um or the whole business units. So I think earlier you asked about some examples right if we take a modern enterprise let's say >> 10 20,000 employees >> they start using Gle search and assistant uh seamlessly from day one uh thanks to knowledge graph

Gle gets to answer hundreds of different varieties of questions use cases across let's say sales engineering IT customer service security HR um and as the employees get on boarded to glean and they develop the habits of using lean, they start noticing, hey, I can go beyond these everyday horizontal use cases. I can actually get my team uh to operate with AI more. But in order to do that, I have to give more custom instructions to Glean. I have to get Glean to follow a set of instructions, a set of steps. Sometimes these steps need to be highly deterministic. So I'll give you some example. Finance teams they love using Glean for everyday questions like um hey what is our churn risk um on our newest cohort of customers but then they start noticing that they can customize Galen deeper for more uh finance use cases

like hey I want to improve financial planning with Glee. No single company does financial planning in a um uh in a like horizontally in a generic way, right? Every company has their own flare of doing financial planning. So then they start um on boarding to what we call glean agents where they can say hey I am almost building a different glean assistant but I want to supply um all the instructions to it so that it behaves um just like my coworker. It is as if I am hiring a financial um analyst or a planner and I am educating or training this hire to act in a certain way. Um or accelerate financial reporting so that the next time the company is doing the board meeting um they are actually cutting down the time required to put together all the financial artifacts and the uh closing

of the quarter a few days earlier than what they are used to. So um we kind of enable the companies to operate at two levels. The core product search and assistant where every employee gets to see the value and then the agents platform where um again departments like finance or legal or operations or engineering can take the core glean assistant and customize it in a wide variety of rich ways to conform to their specific business processes. Did this make sense? >> Yeah. No, that does make sense. And and how does your um maybe on a practical side, I'm just imagining as a as a company myself, like how do you get the information from the backlog of current things to kind of get piped into Glean? I'm guessing it's rag oriented or something to that effect. >> Yeah. Yeah. Love it. >> Okay. Um so

I will say Gleen was uh one of the few companies that put the R in in rag maybe for our audience like REG standing for retrieval augmented generation right uh without retrieval generation had a lot of let's say deviations or hallucinations >> and I won't say glean invented rag per se but it it is one of the most beautiful and impactful implementations of how um the raw horsepower of LLMs get to work with the company data um and um all the company knowledge and idiosyncrasies that that get to run the business for that company. Um so um when we started doing uh like creating this rag architecture um as you might also recall this was a few years ago now the context windows were uh quite small um and at the time the discussion was >> hey um do we need rag why don't we

um train an LLM based on company information or fine-tune an LLM um I think that's where uh the impetus of this whole thing started um and we quickly ran into uh this very clear let's say boundary or obstacle. The moment you take um an LLM, let's say you have the resources to train or fine-tune a large large language model with company data and you supply in that training let's say some corpus of company data now you are losing something incredibly important which is all the permissions and governance on that data because the moment you fine-tune an LLM with company data the LLM will no longer have any idea whether Dimitri the employee should accept has um certain data about the company strategy or salary or HR practices versus enrage on the other employee. Now you have the same LLM using the same data and then

using that data to answer anybody's um requests not knowing who should have access to what data. So the market clearly understood hey regardless of how cheap or effective fine-tuning or LLM or finetuning or training an LLM is um the right way to bring LLMs to the enterprise cannot be that because no single company in this world can afford to lose the per permissions enforcement and governance of the company data. It is not like using the open web internet where anybody can see anything in the enterprise. there is a permission profile for every bite of data, right? Um so there was maybe one uh hugely important uh problem that we needed a graceful answer for and then this is how um rake came to life. >> But then over time uh as you have observed models ended up having larger and larger context windows. So then

the next challenge was hey is rag relevant anymore when you can supply all the documents you want um through the context window interface into an LLM and we are still landing on uh a a very precarious problem and it is the fact that you can supply more information doesn't mean that information is the highest quality information right so >> every additional line of a statement you supply uh to the LLM LLM will actually try to use it. So in the end the more information you supply you might actually regress more towards the mean you might get more consensus response more generic response and the way we create value in the enterprises is not by generating consensus responses we actually generate very specific idiosyncratic um answers as knowledge workers. So larger context seldom means a better answer, a higher quality answer or a more precise answer.

So that's why um rag today is as important if not more important than what it was maybe about two years two and a half years ago when it was actually invented by by the industry angle. >> Interesting. Okay. So when you're uh taking these uh obviously rag like you said is kind of your bread and butter. What I am curious about though is just from a practical standpoint the moving of all this data. >> Yeah. >> From a workspace you know a lot of people are using could be just general PDFs and stuff >> uh like that for like SOPs and whatnot. Is there integrations with tools uh out there? Like for example, do you if a company's kind of on G Suite, is there an easy way for them to move everything over from uh their G Drive to to Glean? And same goes

for Microsoft or kind of how does that what does that process look like? Because I can imagine for wellestablished companies, it could be kind of a hassle, right, to to get everything in there. >> Exactly. Um look that's a um that's a very opaque part of um glean but maybe it is one of the most important pieces um in glean stack in how we engage with a customer. >> Um so some of the customers we are working with they have more than a billion artifacts in their main document repositories and they might have exabytes of data. So when they engage with glean nothing gets transferred over to glean servers at all. So when we think about um building gleans index there is no um aspect of let's say copying um the um the customer's data uh replicating it duplicating it etc. um Glee has a

very lean minimalist way of understanding um a company's um document corpus or knowledge corpus while the data sits where they are right so you gave the example of let's say G drive we can talk about Microsoft 365 or all the select conversations and channels um Glee never creates a copy of it right what Gleen does is it has what we call continuous integration um into these data sources through what we call connectors and um we create an index of every conversation, every document along with the permission profile um of these conversations, channels, documents etc. such that um Glean has a um has an understanding of what this document is about but also it starts building the relationship between one piece of document in G drive to a particular select conversation that might be about a related project or the same project and then Glean focuses

on maintaining this relationship understanding of different documents um and in different data stores. And the crucial part of it is a company's knowledge corpus is a living and breathing organism. So if we were to make a copy today um in a day or in a week and the relevance would already go down because we go back to our documents conversations we add stuff we remove stuff we change the permission so that a document that you have access to today um you might lose access to it um the moment the documents author decide say I'm going to make it private for example so Glee manager comes from this a very lean minimalist approach to always be in touch with the um company's documents corpus without creating any duplication or replication of it. This also um is extremely crucial for the customers we serve because glean doesn't

become yet another security attack vector or risk. So whatever documents or proprietary information you have that stays within your cyber security guard rails within your own cloud. Um Glean doesn't ingest something off of your cloud. Um every Glean instance is a single tenant instance. I don't know if this is going into too much detail but managing the security risk profile of our customers is is of paramount importance. This is why Gleen invented this highly opaque but magical way of creating and updating an index continuously without duplicating or carrying over any of your actual data outside the boundaries of your cloud. >> That's very smart. No, that makes sense to me. So you're saying that basically it can it uses the connector to have a continuous um sort of like stream of uh it's essentially using the current things that exist. >> Mhm. >> Via making

it their own sort of uh real time rag repositories. >> Yes. Through the connection. Okay. Yeah, that makes sense. Um >> very smart. Yeah. No, because I I could imagine where that would be a concern for everybody. It's like oh well I'm connecting it and then I'm like essentially re-uploading everything to another company and yeah I could see where they would get uh people get >> exactly maybe if I may there is also the other side of the we I focused a lot on the defensive posture of it like why it is important to >> no but yeah talk about like how it helps uh to perform better and makes the makes it give great answers and whatnot that because obviously that's the point right >> yes and so let's think about the following as well in in this conversation so far for example I

haven't shared any of my screen etc but as importantly if not more is the relevance and freshness um angle of things so as you as you can easily imagine in a company we create so much content um and some of this content is duplicate and they become outdated stale very quickly so when you ask a simple question um there might be hundreds of different documents where on the paper each one of these documents could be used to help answer that question. Right? So um the challenge um is as much about um deciding which document or documents are the authoritative sources um to answer which document is the freshest, which document should be used and which document should be ignored. So um that what Glean um takes advantage of is all these signals about engagement, collaboration and freshness. So imagine you wrote a document or you

created a slide deck today. Glean will say hey um Dimmitri is a expert in product road map and he created a product roadmap deck. So I am already thinking that hey this is a high um let's say highly reputable document. But if you don't make any updates to that document uh to that slide deck, if you don't present it in any of the Google meet or zoom sessions, Glean will also consider the lack of those signals. So it will understand which decks are getting more comments from different collaborators. Who are these collaborators? Are they product managers or unrelated people? Um is this deck getting presented in different meetings? And all these activity signals, Gleen is also using them to understand freshness, ranking, relevance and authoritiveness. So that the next time you say, "Hey, what is our product road map?" Gle can decide across 57 different

versions of product road maps, which one is the right one to show you. Does that make sense? >> That does make sense. Yeah. I think, you know, it almost feels like a big issue that you could be solving in general is uh general lack of organization inside of companies or lack of source of truth within companies because everyone has so many different uh pieces of collateral made. Um, so to that I would just say I'm practically understanding how one could ask questions of >> the uh the assistant and whatnot and get answers while a lot of people are well a lot of people are happy with that. That's awesome. What are you doing necessarily on the agentic side of things whether it being proactive or performing tasks uh with that information? because I could imagine, you know, you mentioned sales and marketing earlier, you know,

people maybe do the same process over and over again. What is what does that look like from an agentic standpoint? >> Yeah. So, I want to uh maybe I want to take a multi-step approach uh here because the um the AI adoption or maturity journey of different customers, different enterprises, they are in different parts of this. We are all learning it on a daily basis. Some of our customers are so advanced that they are surfacing us the problems that I barely know about and I love learning through our customers what we should solve next. But as you can imagine some customers are just saying hey what are these agents for? What are the some good agent ideas we should adopt now? Right? So my prerogative as at Galen is how do I push the boundaries push the frontiers of my most advanced customers but also

make sure that innovative thinking trickles down to other customers that are early in the maturity curve. So um here is here is a like a huge insight that uh I f I feel very fortunate uh to to deal with over the last six months. um some of our customers they end up building their first agents. So for example in the world of sales um many of our customers build agents around similar business objectives and how do I increase the win rate um how do I um increase pipeline generation of my salespeople? How do I increase net new meetings I have with prospects? Um accelerating time to deal close. accelerating time to address um prospect objections. The first few agents we see our customers creating in the realm of sales in B2B sales are around these ideas, right? And when a customer goes from, hey, I

have no agents to hey, I now have the first one, two, three agents, that is already a moment of truth. is all of a sudden they realize, hey, I have this agent that can stay consistent um across different sales people. They can use these agents for a wide variety of prospects and their objections, but also they le these agents leverage the upside of the LLM, the stoasticity. So these agents are very versatile. So I will say uh the big moment of truth with glean agents and I think this holds true for the broader agents movement in the industry is agents bring a the right combination of determinism like the traditional software paradigm is determinism and software does the same thing again and again with the huge upside of stoasticity that LLMs bring to the table. So when we think about an agent, it is really

the right combination of how much determinism you want to have with the stoasticity like dynamic decision making that you don't need to qualify. But we are talking about one agent let's say for sales. Um once the enterprise realizes the value of an agent, guess what happens next? They build tens of agents, hundreds of agents. So all of a sudden um let's say you and I are employees in a 10,000 um employee organization. We go from hey I had glean assistant I was using to generate answers for everything and now there are hundreds of agents that are specialized in different things. So how do I remember which agent to go to to get my job done right? Maybe there are actually multiple agent versions that might get the same thing done like digesting um like creating a digest of what happens in a select channel when

you were away. So um many of these customers that realize the value of agents and start scaling their agents investments they come back to the table with this amazingly valuable insight that again I feel very fortunate to have worked on with the team and that is how do you maintain a singular user interface where the end user the Dimitry the employee um doesn't need to remember or recall whether there was an agent for it like if you If you recall the the amazing um Apple marketing slogan for the app store, hey, there's an app for that. >> Yeah. Yeah. Yeah. >> Right. With the world of agents, um saying that there's an agent for it doesn't cut it because people don't want to see through pages and pages of different agents to remember, hey, this is the agent that can do the work done. So

increasingly with our more advanced customers that get to value with agents, they have hundreds of agents across sales, engineering, marketing, product, um operations where the onus becomes the glean um assistant can match user intent, what the user is trying to do right now to the best agent available where that agent mind you might have been built on Glean, but it might be a third party agent that is um There's another aspects of um Glean that maybe we'll get to discuss but um our customers don't need to build all their agents on Glean. They can actually build agents elsewhere but Glean assistant becomes that uh brain that orchestrator that singular interface to say Dimitri is trying to achieve this and there is the perfect agent to help Dimitri with that. I'm going to invoke that agent. Dimitri doesn't need to know whether that agent exists, whether

that agent is built on Glean or on Snowflake or on Salesforce or on Microsoft Copilot platform. Glean can route the right intent to the right agent in runtime. >> Interesting. Could you explain a little walk into that a little bit more? >> Yeah, sure. So, um let's think about the following. Um as it happens um at Gleen we recently closed our uh performance review cycle and if you think about uh companies every company without exception have some cycle to review performance of the employees um to promote people to give feedback to people to reorganize etc right let's say this happens every six months every 3 months uh whatever um so the moment um a manager wants to do performance review, the manager needs to lean in to skills that she hasn't practiced over the last 3 six months. So performance review is a very infrequent

thing to do, right? So without uh without Glean, for example, performance review has been a very manual process. uh with Glean many forwardthinking AI force managers started coming up with different ways of prompting Glean to generate a very rigorous evaluation of what their direct reports have delivered have worked on have shined what the improvement areas for their direct reports could be. So this is how for example in a very bottoms up grassroots way AI first managers started building agents in many of our customers to help with the performance reviews. Then the next cycle um HR department realizes hey many of people managers are actually using their own ways of doing this but guess what as HR um the department would like to ensure fairness rigor a common way of doing things across all managers so then um HR department might lean in and say hey

I'm going to build a very rigorous let's say performance review agent so that every um IC um can be evaluated against the same or similar rubric as opposed to managers creating different flavors of performance reviews. So this is the next step. >> But then some companies might say hey you know what we already have built a performance review agent because most of our HR information lives on let's say workday for the for the sake of example. So then they would tell us hey Gleen we want to um run performance review with the help of Gleen but we don't want to build our performance review agent from scratch on Gleen. Can we bring um the agents we built elsewhere? So in this case let's say it is built on workday and if so how can we bring it? So then um the solution there is um

glean assistant today um can integrate with just like the connectors we discussed earlier where glean is indexing information from different data sources. Here it is the opposite way of um moving or interacting with third party data stores where Glean is the interface but Gleen um recognizes that there is a performance review agent in workday and when a manager asks um to create a performance review write up of a direct report Glean matches that intent to the um let's say metadata to the um descript description of what that agent does and say okay I'm going to invoke this um workday agent which is a third party agent and supply all the necessary information to that uh workday agent so that it can get to work and it invokes the workday agent uh the agent might work on whatever it is supposed to do and bring the

response back and then glean uh displays that response um to the end user. So this is how Gleen maintains a a singular commitment to a IT stack or the agent stack where again the core solve here is as an employee I shouldn't need to remember which agent or platform I need to go to to solve my problem. I can tell Galen what I'm trying to do and Gleen will connect the dots uh to the right agent to the right platform and then invoke that agent and get that uh job done. Wow. That's Yeah, that's very impressive. Um because I feel like a lot of the effort that people would put in here is figuring out, okay, well, I know I have an agent for this. Um but like you said, unlike the app store, you know, no one wants to kind of like sort through

necessarily all of the um different agents that exist. So that that's pretty impressive. Um how many uh companies do you guys currently have as uh customers? >> About about I think 500 to 600 somewhere in that range. >> Very cool. Very cool. >> Um, so yeah, that's And is there any specific industry that you feel like is like the highest uh you know like that you work with like um I'm just curious trying to trying to think like who's adopting this, right? Because this is a question that I often ask is like who's um who are you finding is like most adopting >> these >> uh these products and like what kind of impact has it had practically on their um you mentioned earlier about their employment. Yeah. >> Yeah. So I think the um I will answer in two ways. one um who is

adopting and like the companies that have the companies that have um more bifurcated IT stacks right um the companies that are using different software solutions as best of breed solutions as opposed to let's say companies saying hey I will I will build all my IT stack in just one mega vendor so those companies tend to see the visceral glean value much faster because it's not very surprising um the more um let's say different types of software you use then your data your um information is bifurcated um in different silos more so they tend to experience the pain points more often so that is kind of like maybe after the fact it sounds obvious but then this also means glean really matches what we have been seeing with the SAS um industry over the last 10 15 years where really like The modern enterprise went from

hey I am using everything uh from Microsoft as my IT stack to hey I will use Microsoft word but I will pair it with zoom for my conferencing and I will use Salesforce as my CRM and I might use a startup like notion for like cross project collaboration. So there is a clear movement in the in the way enterprises are putting um their IT stack together over the last 10 years or so. the great unbundling of the enterprise software and Glean um has been unsurprisingly um delivering a like great solution to that um to that unbundling. So that part um the more again uh the more software you use as part of your IT stack then glean brings the magic faster and more viscerally to you. maybe that's the obvious side but I will also share the nonobvious side. Many of our maybe listeners here

might expect that um Gle's core customer base is Silicon Valley tech enterprises and that is actually not the case. So for every um tech or engineeringdriven organization that we have as our customer, we have a non- tech traditional retail manufacturing driven organization. So that is maybe something that I really enjoy and many of um outsiders don't get to realize with glean that we are not we are not dominated by tech first um software oriented enterprises but many traditional businesses manufacturingdriven businesses get tremendous value out of clean >> you know it's interesting I think the um there's probably not a lot of people who are considering That's sure knowledge workers and tech and etc. would probably benefit from these a lot and might be the first to adopt. But even in manufacturing um there is a quote knowledge work uh that uh ends up occurring and

they they probably no offense to knowledge working companies uh in general probably understand efficiency you know uh and how stringent one could and should be even more so than maybe knowledge work because the manufacturing companies are have over the last couple hundred years essentially done this to you know a ridiculous level with the the physical components that they're working with. So that's actually yeah that's a good that's an interest yeah I wouldn't have cons I wouldn't have thought of that but yeah it's it's almost like >> when I said earlier horizontal funny enough I still thought of like is it tech is it finance is it um none of the the list in my head was manufacturing or physical so awesome is there and and is there anything that you know you know we're we're kind of coming close to uh the end of uh

the episode I I We always like to ask this question as well. How have you how do you feel like your product is going to help uh businesses in the future in regards to kind of the hiring whether it be process or lack of need of hiring and how do you feel like it's going to impact like the job landscape uh moving forward? Because I feel like agents for a lot of people more and more as it's come out as a phrase and understanding people get that like these are things that can do tasks and people generally do tasks. So they they're worried about that in the in the job landscape. So what do you feel like your product and other products kind of mean for uh the job landscape in the next 5 to 10 years? >> So I'll answer I think in in

in two parts. Um number one um if you think about any employee and their um their first meaningful interaction or adoption experience of AI in the workplace. Glean plays such an incredible foundational role with it because like I know hundreds of millions of people wearing our consumer hats. we are interacting with a bunch of AI solutions like CHPT, Gemini, Perplexity etc. But at work um connecting the dots between how can AI help me um deliver more value become a better employee there is a big chasm there and now we talked about the reasons of that chasm it is security it is access to data etc. So if you imagine the employee of a Glean customer, Gle plays a hugely foundational role at bringing AI for the first time to how an employee uses AI to get better work or to get their work done faster.

So Glean has a mission here. Many of our buyers, our champions, the CIOS that sign the Glean contracts, they think of Glean as either one of their major bets or the major bet itself to uh to drive AI adoption in the enterprise with their employees. So if you think about the future hires like the future of the job marketplace um I will venture to say that um the candidates the talent that knows how to bring AI into their own craft whether they are doing manufacturing work, sales, finance, engineering, legal work but bringing AI to your craft so that you are more creative, you are more productive then those employees um will be highly sought after. So how those employees get to learn how to be that employee that AI enabled AI native employee. Gleen is playing a hugely important role today >> because you get

to use Gleen in a safe um and everyday environment and you get to build agents agents for yourself. You get to build agents that your teammates adopt that give you feedback. you can iterate on these agents in a highly zero coding required um let's say safe and productive way. So I will say that like glean um contributes to that mission of making employees a lot more AI forward AI native or AI enabled. This maybe the um the the crucial aspect glean uh plays a role in in today's talent transformation. Well, that that I mean that was uh that was awesome. I think this it's funny this episode kind of flew by. I can't even believe that we're actually at 50 50 minutes and I'm a little bit disappointed um to be honest with how fast the the the time went because I I would love

to keep talking. So, um if you do uh do we have a few more minutes? We can actually keep talking if you have a few more minutes. Um if not, okay, sure. Yeah, we can probably run it till a couple more minutes. Um so with uh your kind of growth over the last few years, what is the main goal and vision of what this company can do for people? Right? Because obviously we've seen such growth and such improvement in so many different ways in the agentic world. What does like I don't want to say the final goal but like the current vision look like for the next couple years of what somebody uh what a company is going to be able to accomplish with clean >> yeah um so maybe I'll um I'll try to answer in uh two ways um I'll go back to

a framing um that I love using and it's about the craft we we bring to our work again regardless of the function seniority etc we do um like the work is actually like what we put in every day it ends up being our craft. So I I am call me an AI maximalist but I sincerely and deeply believe regardless of our function not just the engineers or product managers but any function with AI we will get to do better craft. So our creativity will increase right and so AI is not just about time saving or getting tasks done faster. It is actually about craft. If you get to do more insightful podcast episodes because with AI you can prepare for your next episode like better deeper ask more insightful questions etc. This is really the core of the craft and I do believe AI will

help each one of us to do our craft better and with craft the productivity of the enterprises will increase. They will ship innovation faster. They will ship innovation that they wouldn't have thought about before thanks to the increased craft of every employee and then this is going to lift all the bots deciding like where the where the lion share of impact uh will be. >> No, I think that's a that's no that's a very good actually that's something that not a lot of companies are talking about. Um, the reason I say that is because I feel like there is focus always from what I've heard on getting more done and or getting stuff done so that you don't have to do as much as a employee or company, right? And I mean the employee would have to do less and the robot does more. So

therefore there's like a time increase on the person to get back more time for leisure, right? But I actually haven't heard anybody bring forward the point the work could just be better. >> Right. And not better in the sense of uh some sort of I guess pragmatic concept when you know you hear business talk about better. You're saying because this is a this is a term that is not used often b in business but I do think can be used for you you're saying that it will lead to more work being what is essentially craft right when and this this concept I believe there's a couple books I remember reading on uh this back when I was in my self-improvement uh productivity book phase um but there was a book that specifically talked about work being craft rather than being work and it also I

think ends up leading to almost like a compounding form of productivity in a sense, right? You take somebody like me who the last thing I want to do is spend a bunch of different time clicking around trying to find ways to ask better questions, right? It's sure, some might like doing the nitty-gritty of things, but most don't like doing the nitty-gritty of things. They'd rather do the the craftful type of things, right? like the person who does the sales calls would rather do the sales calls than find get the people on the sales calls. You know what I'm saying? Um or figure out how to do better sales calls. And and that's uh that's something that I Yeah, it's kind of lost on a lot of us in this race to get more done with AI agents. We can make the work better and better

and more craftful. That's that's a very that's a very good point. >> Exactly. Mhm. Yeah. No, >> I'm loving that. It resonates with you like like you said, I think the the market looks at the more obvious things um where hey with AI we get to do things faster, save costs, reduce the time etc. It's of course like less obvious to think hey the craft level will increase like the creativity will increase. Um but I think we are going to get there. I'm already seeing like many different examples from different industries or job functions along those lines. So I am quite hopeful and I have proof uh that convinces me. >> Yeah. And there's there's no reason that we can't move more towards that type of environment. I think uh because to be quite frank I think there's excuse me a lot of focus on

people's time savings and whatnot in order and that it's funny though that leisure component I was talking about with like time saved on work is at some point you talk to anybody who's retired like outside of like actually like maybe 10% of people most people are like yeah I mean they kind of like sputter out like the the joke goes for middle-aged men in the Midwest like ah I can only golf so many days a week I do kind of miss work right um and the reason for that is it the the the craft that we would wish to to live in for ourselves is something that everyone does yearn for right I wake up I run my own company so I kind of have I have a high stress component but also I do have the thing that others don't maybe in a lot

of jobs where people >> I think there's actually not as much of a correlation between uh work volume and uh negative feelings about work as people would might think. It's uh more so uh in line with whether you actually find value in what you're doing. Um, >> and I think when we get more towards that that craft deep layer, that would be the case because somebody who's, you know, >> I used to work in paid ads, uh, and about a decade ago, maybe more, there was not like bidding algorithms to optimize for pay-per-click. You you literally had to manually bid on keywords and you could purchase keywords and these sorts of things. These were not edifying things, right? This is not edifying whatsoever. Data entry is not edifying, right? So, when we move from that to something that we can be proud of and it

feels more like something that we're building that that means something, I think if that's something that you guys are striving for, that's a that's a beautiful thing. >> Love it. >> Yeah. I'm going to use some of your some of your narratives in in my future conversations with customers and prospects as well. >> Well, you should. No, I'm just kidding. Beautiful. >> No, I I I really No, I I think it's true. Like uh I love what I do. Uh I I'm stressed because I I work for myself and it's not like the longest like people who start their own company and they've reached a certain level of success, I'd imagine, have less stress. And you know, there's there's like different reasons why you're at different stress levels and and running a business. Um, same goes for when you're at a at a job, right?

Funny enough, everybody the first week enjoys work for some reason, like at a new job, and it's because it's this new exciting thing. You're thinking of all the cool things that you could do, and then some people get pigeon holed into like a specific monotonous task, and then the reality of said uh job isn't quite what they uh imagined. But um if you can help move more people in the direction of you know uh it being what they'd imagine that's probably the >> that's probably the way to go you know. So all right well um >> you say >> no I was going to chime in uh like in a very personal and recent way just this podcast by the way I I enjoyed this uh a lot. I don't know, we should do this weekly. Um, but like ask me how I prepared for

this podcast and I actually used Gleen a lot like your previous interviews with Leor and Ellen. >> Um, I actually used Galen's AI >> to better prepare and I do think that improved my craft. Like I don't know if you like the podcast but whatever I am here without Galen I would have done worse. >> But the thing is it's not a time saving for me. like how can I be more insightful or concise or sharper with my uh communication? How do I adapt to your communication style? How you want to run your podcast? Um Glean helped me prepare better or AI helped me prepare better and that I consider part of my craft. So I think the opportunities are endless to improve craft by the use of AI rather than again hey help me cut the time to prepare for this meeting or for

this podcast or the interview. So I think I'm quite excited um and optimistic about about the future of AI in the workplace and for our employees. >> Absolutely. No, I totally feel that. So with you know with that um being said, my last question would just be what necessarily uh would be the final thing you'd like to plug? I'm guessing it's uh glean.com if if I had to guess anything that you're you're interested in plugging to to close this one out. That's uh yeah let's let's go with that and um we are hiring across product design and data I mean across all functions but yeah um AI first product folks or designer data folks uh come find me on LinkedIn could be another flag. >> Awesome. Well uh we really appreciate you being on the show. Uh everyone please go uh check out Glean on

their website, LinkedIn, all the socials. We we'll have links down below. Thank you so much for being on this episode. Thanks everyone for watching and we'll see you in the next one. Bye.