Building the Future of Enterprise AI: From Harvard PhD to YC Founder w/ Gokhan Egri
Today’s guests — Gokhan Egri, Founder & CEO @ Brainbase (YC W24)
prev. PhD at Harvard University in Computer Science
Summary
In this episode of Masters of Automation, host Alp Uguray interviews Gokhan Egri, a successful entrepreneur who transitioned from academia to the startup world. Gokhan shares his journey from Turkey to Harvard, his experiences in Y Combinator, and the evolution of his companies, O'Leo and Brainbase. The conversation delves into the challenges of being a solo founder, the importance of customer feedback, and the future of AI and automation in the workplace. Gokhan emphasizes the need for transparency in automation processes and discusses the unique insights that drive his approach to enterprise AI. The episode concludes with reflections on the immigrant founder experience and the personal inspirations that fuel Gokhan's entrepreneurial spirit.
Key Takeaways
Gokhan transitioned from academia to entrepreneurship after realizing the potential of AI.
His first startup, O'Leo, was born out of a need for automation in education.
Brainbase evolved from O'Leo, focusing on enterprise AI solutions.
Customer feedback is crucial for shaping product development and messaging.
The concept of the 'digital worker' is central to Brainbase's mission.
ROI in automation can be measured through clear metrics and pilot programs.
The 'black box test' helps identify automation opportunities within organizations.
AI can significantly improve the efficiency of enterprise operations.
Transparency in automation processes is essential for enterprise adoption.
Immigrant founders often possess unique grit and adaptability that contribute to their success.
“I realized that while academia was slow and bureaucratic, startups moved at lightning speed, and that’s where I could really make an impact with AI.”
On Brainbase’s Approach to Enterprise Automation:
“We don’t aim to replace entire teams right away. Instead, we automate the small, repetitive tasks—the ‘paper cuts’ that bog down big organizations—allowing them to focus on what really drives value.”
On Y Combinator:
“YC taught me that being able to build is one thing, but learning how to sell and execute at scale is what separates the good from the great.”
Transcript
Alp Uguray (00:01.248)
Hi everyone, welcome to Masters of Automation. In today's episode, I have the pleasure of hosting Gokhan. Gokhan, welcome.
Gokhan Egri (00:09.121)
Hey, how's it going?
Alp Uguray (00:11.224)
Thanks for joining. Gokhan has a stellar career. He came to Harvard, he was doing his PhD, then he left his PhD, went to Y Combinator and started a company, Leo and then started another company called Brainbase, which main mission is enterprise AI, which aligns very well with the future of work driven by AI and automation.
I have a lot of questions for you today. To kick things off, let's start with the early days, So let's start with the PhD days. So initially, what led you to get started from Turkey, come to Harvard, and then chase a vision on the research and in academia?
Gokhan Egri (00:41.557)
Yeah, for sure, for sure.
Gokhan Egri (00:51.832)
Yeah.
Gokhan Egri (01:00.571)
Yeah. Yeah, a hundred percent. So, basically the, just to give some background, did my undergrad in Turkey, where I did it in actually in electrical engineering, not in computer science. And then by the end of it, I did a bunch of machine learning and I really liked it and I was very interested in it. I did a bunch of research, got to work with transformers very early on. and then I basically just looked at the, the options of industry and academia. So.
When you're in Turkey, when you're in a company that's, when you're in a country that's usually not United States, doing a startup from the get -go from undergrad is not like the, the most viable option. So I was just looking at kind of academia versus industry. And that was a pretty easy choice that I wanted to do academia. Cause I essentially love building and I, I had a pretty good grasp on what I wanted to build. wanted to build in machine learning and computer science. so from then on, it was just basically a question of.
What kind of research excited me? What kind of groups excited me? And I looked around and I found my advisor at Harvard who's working on computer vision. Amazing guy, Professor Zickler. Shout out to him. And yeah, I just loved the work he was doing. He was a great advisor. And so I just went in and it was an awesome experience overall, just getting used to that.
Alp Uguray (02:20.866)
And when you join Harvard, you stayed there for a bit, right? Like you've done your research and what sparked your interest in going from research to then start doing these projects because before Brainbase, you had another product that you rolled out as well.
Gokhan Egri (02:39.531)
Yeah. Well, a hundred percent. yeah, when I was at Harvard, during my time at Harvard, I got a chance to work on a lot of really cool technologies. got to work with transformers and machine learning and models, you know, not only language models, but computer vision models, you know, from zero to one. So I got to work with the nitty gritty, just like the tensors of it all just got to code them from scratch all the way to getting, you know, early access LLM. So was actually one of the first people
to have access to GPT 3 and beta. And the way I got access to it was I just spammed the CTO at the time was Ilja, Ilja Sitzgever, just ideas for what I would build with it. Like every day I would just send like two, three emails to him. Finally he caved in and just like gave me access. So I got the experiment with LLMs from, know, I was one of the first people to experiment with it. So that just gave me an idea of kind of how fast the industry was moving.
and how comparatively slowly the academia was moving. And I, as I was doing research, I kind of got bored of all the bureaucracy in academia. of, know, I saw that I was spending, you know, maybe a month building a new model, building a new architecture, and then six, seven months writing a paper that is, that does not really add anything to the model, but kind of is just like a, like an internal process in academia where you try to get it accepted into these.
these literature reviews and kind of buy these reviewers. So there's definitely a publisher parish, you know, idea in academia. So I just saw that that wasn't really my calling. And I, now that I was in the United States, it was a much more viable option to actually start a startup. The way Leo came about was that, I was a TA at Harvard. So you have to be when, if you're doing a doctorate, you have to be a TA, at some point.
And I was spending, you know, nine, 10 hours every week, just grading papers, you know, reports, and they all had very clear rubrics. So I wasn't really doing much. and this is, this was after chat, if you came out. So I just wrote a script one day. That was like a hundred lines, that took these submissions and then. Formatted them in a way and then fed them to chat GPT and had it grade them. And then that took about 10 minutes total. So my nine hours went down 10 minutes. I was like, this is amazing.
Alp Uguray (04:54.734)
you
Gokhan Egri (04:58.147)
So I went to my other TA friends, was like, do you guys want to use this? And they did. And they all loved it. So I enlisted another one of my friends from grad school. I was like, you know, we can, we can probably sell this to other people. And then we actually saw that the real market for it was not universities, which are usually very well funded, but rather K1 through 12, where there's so many more submissions. So, you know, so much less time and, you know, resources. So that, that company is still going on and you know, it was.
That taught me quite a lot and that gave me the automation bug in a way.
Alp Uguray (05:34.434)
In a way, you've actually found the mundane part of your job and then went ahead and automated it and made it faster. It's interesting because automation education is really powerful. It also enables everyone to have access to intelligence, especially in other schools. Not every school is as lucky as Harvard. have great teachers.
Gokhan Egri (05:59.299)
100%, yeah, for sure, for sure.
Alp Uguray (06:03.086)
Well, this just came to my mind, but when you were pitching different use cases to Ilya, which use case was the most dear to you?
Gokhan Egri (06:10.093)
Yeah
Gokhan Egri (06:14.551)
I think it's just, just like looking back, it's a good question. So looking back, I think the one that was the most dear, was the one that he actually caved in and was like, so the, the, an email from Ilya Sutskova at the time saying, yeah, that's actually an awesome idea. You should definitely build it. Here's access. was great. And the idea there was, this thing that I was trying to build called Genie, which funnily enough was essentially a semantic smart search that's
being done right now by a lot of companies. I basically said, I want to build something for e -commerce and kind of knowledge source.
websites where instead of having it be a deterministic search, I want them to be able to put something in the chatbot that I want to have the LLM transform into a query that can run through that database. So today that's just, that's very ubiquitous. Just it's everywhere. But at the time it was just like, it was a crazy idea. I don't think it even worked that well with GPT three. but yeah, was, it was awesome getting a yes from him. I definitely had a bunch of.
not so good ideas that I pitched. So I'm glad I came up with at least one half piece in one so that I got access.
Alp Uguray (07:26.092)
Yeah, it became the core of the applications as well, like essentially doing the drag piece, right? Like getting and the informatics search and then pulling the information.
Gokhan Egri (07:31.329)
Yeah, yeah.
Yeah, 100%. Yeah, yeah, yeah. It's very cool to see how kind of the technology went around and, you know, things that were barely possible or even impossible with GPD 3 suddenly became, you know, very possible with 3 .5 and then, you know, leaps and bounds exceeded with GPD 4 and 4 .0.
Alp Uguray (07:53.358)
So for that leap of intelligence, everyone has access to that intelligence and you describe it perfectly. You found a case within your own job and then automated it and then said, wait, a lot of people can't get use of it. And then went on to selling it. What was the next evolution from that that led you to, wait, this could.
Gokhan Egri (08:12.033)
Yeah.
Alp Uguray (08:19.702)
maybe automate not only or improve the lives, only of education, but maybe of all industries. So that maybe ended up the brain base.
Gokhan Egri (08:25.473)
Yeah. Yeah. Yeah, a hundred percent. yeah, Brainbase was, so Brainbase came sort of as a very natural conclusion of Leo as well. just like ideologically, Leo was sort of the education verticalized version of Brainbase in a way, at least the first version of Brainbase. Going from Leo, basically saw that, you
key or, you know, these LM models are amazing at analysis when you give it some prompts, when you give it like a rubric. So I was like, you know, it's, this is kind of a problem of just going from unstructured to structured data in, like the most basic form. I give it some submissions. There's some formatting that I do, and then goes into a structured format, which is the scores or like some, some other output I get from it. We also did like feedback giving an annotation. but you know, I was just like, read something unstructured, analyze it to the LLM.
just like it would be a human, it's, as if there's a human on API and then you just get a structured output. It wasn't the biggest leap to just go from that to, know, I think other people would be able to use it. So I kind of just went ahead and expanded the product and expanded the scope of it, to the first version of Brainbase, which was, going from unstructured data to structured data. So you would be able to put in a bunch of, file types, know, PDFs, docs, PowerPoints.
you know, images, audio, video, whatever you have, and then it run them through a pipeline, which based on the user's prompt, based on the user's instructions, we'll take them into structured format. this at the time, the first version did not have any the automation capabilities, cetera. that's kind of how it came about, but there's another portion of how Brainbase came about that is actually kind of rooted in the research I've done at Harvard, where I, in some ways to, you know,
side projects slash some miscellaneous research I've done there that I've developed the core part of Brainbase, which is the base language, which I can talk more about.
Alp Uguray (10:29.838)
Yeah, that sounds very interesting. so it's like, I the side projects always is the indication of where the curiosity takes us. then like that sparks into becoming actually like the like, like you mentioned earlier, like, like life's work. And so let's talk about that a little bit, right? Like, so you've been doing your research and then you got curious on the side project and then, then
Gokhan Egri (10:37.655)
Yeah. Yeah. For sure.
Gokhan Egri (10:45.165)
Yeah.
Gokhan Egri (10:49.124)
Yeah.
Gokhan Egri (10:57.915)
Yeah. Yeah. Yeah.
Alp Uguray (10:58.85)
you took a bet on it and they said, okay, I'm going to go all in and then there's the portion off where you apply to Y Combinator as well. How did that all evolve? How did it shape?
Gokhan Egri (11:11.363)
Yeah, for sure. Yes. I, during my time at Harvard, I did a bunch of side project. Most of them not worth mentioning. So I, brain base was kind of the one that I was like, yeah, this is the one. And I just went ahead with it, which looking back was the correct choice, obviously. it kind of came about from this kind of miscellaneous side research, not in my group research that I was doing, which was, I kind of thought to myself, you know, after GPT 3 .5, I sort of.
we were using LLMs for a lot of these novelty use cases and some productivity use cases. But for the most part, people were very not convinced with the fact that it would be able to do the job of an average employee. So I went in and looked at it because I saw that, know, GPT .5 is not that much dumber than an average employee at a company. Overall, if you look at it in a way of just intelligence.
In a lot of ways it's, it's, it's smarter. It's able to retain information better. It's able to reason in a deterministic way better. but it obviously was not able to do the job of an average employee. So I kind of went down this rabbit hole of understanding why it was not able to do that. And that brought me to understand that LLMs are essentially a way to compress information. So LLMs just take the entirety of human knowledge as a corpus and they smash it down.
to these comparatively low number of parameters, which is still four, you know, billions and trillions of parameters. But essentially what it does doing that is that it makes intelligence, makes information retaining much better. But what the average employee has that the LLM doesn't have is the ability to use tools and use tools in a native way. So that led me down to kind of how we use tools, which is we have these programming languages called Python, JavaScript,
that have been designed for us to write and every function calling that has been designed on LLMs have been very ad hoc that has been added on top of an LLM, which is just an information mechanism. So that prompted me to develop this new language called BASE, which is an AI native language. So the blocks of the language are tokens that you can put on an LLM. And what that does is it allows the LLM to actually provision workflows. What that means is that, you know, just like how there's a token for the word the.
Gokhan Egri (13:35.485)
Now in the, in the way the base uses, there's a token where there's a token for provisioning a workflow that is able to send an email. So that made it much more efficient, much more kind of a straightforward to implement. And it just overall made it in, in a lot of ways, just, you know, able to work with, able to do the job that you would need from an LLM to put into work, more robustly.
Alp Uguray (14:04.654)
That's very powerful, especially that I love how you explained that the ability to predict the token, the next token, has the ability to predict the next maybe best action, next action to be able to then execute an automation is very powerful. And from my research that the
Gokhan Egri (14:05.571)
how it came about.
Gokhan Egri (14:22.773)
Yeah, exactly.
Alp Uguray (14:33.538)
the cost of predicting the next token, even it's like a simple word, is same as the cost of predicting the next action, right? Like in terms of the energy consumption side of it. So in a way then you're enabling with the same cost and a new layer of automation on top of intelligence of processing knowledge.
Gokhan Egri (14:40.454)
Yeah.
Alp Uguray (15:03.394)
So how did you take this idea and then the part with your research and then the product and go to the next step? Like go out there and then maybe transition to YC as well, like just going there and then hearing from their experiences.
Gokhan Egri (15:13.386)
Yeah.
Gokhan Egri (15:17.998)
Yeah.
Yeah, for sure. yeah, once I had this idea, like I said, the first version of Brainbase did not have this automation part. was like, you know, let's just see if people are interested in the unstructured, the structured part even, and then I'll add it forward. I was thinking kind of conservatively, I think I just thought like I would add it in like a year or two years. And I ended up adding it in like the next two months, the automation capability as well. but going into YC was, YC is a very well, you know,
It's a very well marketed machine, first of all, you know, kind of separated from their capabilities and kind of the things they add, which I'll explain gladly. so, you know, Paul Graham's essays, I, anyone who's looking at startups, who's looking at building stuff that is going to go out to consumer or business, you know, have heard of has her of YC. So I definitely entered a YC. I thought it was like a very good status thing. It's to give something that I, I thought I lacked.
very much, which was network, which was guidance in how to sell these things. Cause I knew how to build things, but I always had this kind of chip on my shoulder that I didn't know how to sell. and why Cecil point is that they get people who are able to build and then they're able to very easily guide them into, know, being able to sell it and become the best sellers essentially. So with that in mind, it was just not a, it was like a no brainer to kind of, apply to YC.
Having said that, I procrastinated quite a bit and I've actually applied at the very, very end day, which a lot of my friends also did at YC. I think that's like a very common thing to do with YC. But yes, I had my interview. I interviewed with Harj and Emmett, Emmett Shear, who was actually right after we interviewed, who became the interim CEO of OpenAI. And that was very interesting. But yeah, so they liked the idea. So they chose to...
Gokhan Egri (17:17.997)
fund me even as a solo founder, is not super common. But yeah, after that, I just moved to San Francisco. I was already in the process of dropping out to pursue this, but went to SF, met a bunch of cool people, got a part of the community. yeah, with a lot of companies that went through at least our batch, and I think a lot of other batches, YC is definitely a make or break thing where, you know,
the kind of the confidence that they give and the network ability and the guidance and all that. It is just very valuable.
Alp Uguray (17:52.718)
And it nurtures you to be a founder, like being a tour
Gokhan Egri (17:59.395)
It's very lonely. Otherwise it's, it's, it's very, very lonely actually. So as a solo founder, it's, it's especially lonely. I, one of the main reasons I was able to survive is that I actually, shared a house with some other YC companies. So with, with an otherwise company in that, you know, going through the whole thing alone as a founder is very hard. I think moving to a new place and have knowing no one, because you're working, you know, you're a 20, 22 hour days is.
just insane. So YC kind of helps with that. Even if it's just like mostly working, knowing that you're working alongside other people who trying to take these moon shots as well is just very empowering.
Alp Uguray (18:39.948)
And as a solo founder, what were some challenges that you had during the YC days? Because everything at the same time is the most important thing to do.
Gokhan Egri (18:49.268)
Yeah.
Yeah. Yeah. I did not, I did not sleep that much. I can tell you that, it was definitely rough. think it's, I mean, it is obviously rough. It's obvious. It's rough in the ways that you would be able to kind of predict and, know, I, I had to code switch every day, like every hour. So I would be coding something for a feature for a customer. And then I would have to switch into sales mode to talk to someone. And then I would switch into fundraising mode. So one of the best advice YC gives is that.
You know, you don't, you don't want both co -founders to be in fundraising mode. And since I was the only founder, I did not have a choice there. you know, there's a very big portion of it where you're able to do a lot of, you're able to kind of do a little of everything, but not, you know, a reasonable amount than any of them. That is the risk of the solo founder. So navigating that was the most challenging part through YC And then now we're hiring, now we're actually expanding.
You know, it's like the mundane stuff that I would love to have some help on, which, know, our team is amazing and we're getting people who helping me so much with this. you know, mundane things like payroll or, know, just like registering for legal things, you know, that kind of stuff is not stuff that excites me. I don't think it's stuff that excites anyone, but there are people like my co -founder from Leo who were just inherently good at those things who were just like, you know,
very productive in doing that kind of just like bookkeeping work. And I'm not the best person to do those and I know that. So that is like a challenging part that still goes on.
Alp Uguray (20:30.228)
I think it's also like one of the things that will get automated over time as well, especially now with LLAMs like...
Gokhan Egri (20:36.023)
Yeah. Yeah. I mean, if you have any say on it, yeah, for sure.
Alp Uguray (20:40.258)
Yeah. So then you got out of YC and you're a solo founder now you're backed and you have a product and you are doing POCs and like getting feedback from customers. How was the initial reaction with the customers that if you compare your days at Harvard, like you told about the idea and then
Gokhan Egri (20:54.349)
Yeah.
Gokhan Egri (21:03.989)
Yeah. Yeah.
Alp Uguray (21:05.887)
And then you found out and unlocked an insight from the customer that you didn't know before.
Gokhan Egri (21:09.411)
Yeah. 100%. So yeah, mean, definitely. So the product has definitely changed quite a bit. I think the product has changed most in the messaging that we provided it with.
way I started messaging our messaging through YC was AI power spreadsheets, which resonated with smaller medium businesses did not resonate with enterprises because enterprises are like, you know, we have Salesforce to do that. So what, what, what is an AI power spreadsheet? So that did not resonate. We switched it over to AI workers, which started resonating, but people were very, people were not able to figure out how to price it. So it was just like, you know, if I have a 50 employee team and I want to replace it, do I need to buy 50 workers?
Where the answer is obviously, no, you just get one worker who replaces the entire 50 worker, 50 employee team and more. So we finally got into AI automations, usage based pricing, et cetera. there was a lot of changes on that from customer feedback, I that customer feedback is something that sort of. It's like an incremental improvement. I believe companies have.
a big vector of direction. they have a big vector of direction, which is founded by the founder and like the intuition you have and the grasp you have and the unique insight that you have that other people don't. And I can talk about the unique insight of Brainbase in a bit, but you kind of have that and that puts you in a position. And then I think what, you know, when you talk to customers, all they give is sort of incremental movements on that trajectory. So it definitely changes the position, but
I think unless you're doing a full on pivot, unless your unique insight has changed, everything you got from customers essentially just allows you to pinpoint the exact point that you're going to get the product market fit. So that's kind of what happened with us. The main feedback we got was around messaging and there's definitely been changes to product, but the underlying, the core of the product, the core automation, the core services, document and voice that we provide has mostly remained the same.
Alp Uguray (23:19.544)
So then in that case, you found out what resonates with the customers, like how they perceive the product, how they understand the product. And from the unique insight that you had, what was it? What was the unique insight that drove the idea?
Gokhan Egri (23:23.958)
Yeah, yeah, for sure.
100%. Yeah.
Gokhan Egri (23:39.487)
Yeah. Yeah. Well, the unique insight that we have, we had and still have a brain base is that, LLMs are on a good trajectory to replace the average employee. and you know, that's not a very unique insight. That's an insight that a lot of people got. The unique insight we had was that it does not make sense at this, at least at this point to go ahead and try to replace, you know, sales, dev, HR, know, these entire
any adjacent roles. So I had this idea that you cannot really go to a CEO, CIO, and tell them that you're going to replace their entire self organization with AISDRs and that they should commit to this. And then they're going to see the ROI in about a year and a half or two years. That's not something that resonates with enterprises. That was my unique insight. We've since experimented with it and talked to people and saw that that's exactly the case. Our insight was that.
We want to keep those roles intact. But when you're an enterprise of a certain size, you essentially start having so many of these small teams that we call paper cuts that actually start to bog you down quite a bit. And they're not even revenue adjacent. So example teams in voice parsing, you know, compliance, the teams that do reference checking during a, you know, a hiring process. So teams like these are, you know, usually small, but
They essentially cause a lot of managerial overhead. They have very clear inputs and outputs that can be automated. we were like, basically, since the market sentiment is not what sells HR dev, we're not going to replace those. We're not going to touch those roles, but we'll go to a company and we'll tell them, we're going to clear out everything else. We're going to make everything else incredibly efficient so that your revenue impactful roles, you can put more focus on, you can put more money in and everything else you can kind of start.
know, automating phase by phase. That kind of resonated. That is something that has resonated with investors and customers and, you know, fellow founders alike and has given us a very unique market positioning that we're very happy with.
Alp Uguray (25:53.006)
It's sorry, from customer's side, right, especially enterprise, I think like when they look at functions like, like you said, finance, HR, and those functions have, especially in a large company, a hundred thousand employees. like, it goes, it's the roles became so fragmented, like each individual employee just does a very little thing. And then there's that handshake of.
Gokhan Egri (26:10.37)
Yeah.
Gokhan Egri (26:18.882)
Yeah.
Alp Uguray (26:20.972)
Yeah, I review the background, this candidate has no criminal history, let's go ahead and hire, pass it. And then like it keeps going that way. And like from that perspective, how does how does brain base gets augmented within that workflow of like different employees and their handshakes?
Gokhan Egri (26:24.123)
Exactly. Yeah. Yeah. Yeah, for sure.
Gokhan Egri (26:43.594)
100%. So essentially what we do in a company is we've developed this test called the black box test that allows us to essentially go to companies and you know, when we see that companies are excited about automation, but they don't know where to start, which is, which is the most common case right now. They have the funding. They want to go with automation. They want to go with AI. They saw the writing on the wall. They don't know where to start. This is kind of what we employ.
And it has four, you know, bullet points to check. So does this team have standard inputs? Does this team have standard outputs? Is this team revenue adjacent? Which can be proxied by the, do you hire people for this team that are as qualified as the best people you get for other teams? You know, do you spend as much time hiring for this team or not? Cause if you're not, it's probably a team that can be replaced. And the essentially the last one that's the most important one is.
Does this team's head count grow proportionally with operations? Take for example, a moist parsing. You're doing a thousand moist a month this quarter and you have X amount of people. Next year, you're doing a million of moist that you need to parse on the same quarter. The number of X is gonna go linearly with that number.
That is a very bad news because it means that the more successful you get, the more bogged down you're going to get by this managerial overhead. So that's a very good place to be automated. So we take this to the companies and we tell them, okay, so we'll just go ahead and identify these sub teams. know, these can be in the larger sales organization as well. So in the larger sales organization, if you have a team of, you know, seven people who's just doing, let's say, you know, enrichment.
that just you have your leads, you get them from HubSpot, they just Google them and they enrich them and they find the CEOs and they pass it on to someone else. We can take that part out. We're not going to touch your SDRs to actually directly go out to customers and talk to them. We're not going to take over your AEs, which are the ones that are actually keeping you from churning. But this enrichment part, you know, it's very well connected. It's essentially a black box, which is where the name of the test is coming from. And we can just take that out and make it 10x more efficient.
Gokhan Egri (29:04.961)
without any loss and possibly a lot of benefits.
Alp Uguray (29:10.99)
When people ask about it, which happens a lot in the enterprise automation use cases, the ROI, how does ROI is driven and how do we measure, like we will get our money back after we the investment. And of course, it's contingent on the process, it's contingent also on whose budget it's coming out of. How do you view that?
Gokhan Egri (29:25.121)
Yeah. Yeah.
Gokhan Egri (29:34.275)
Yeah, 100%.
Alp Uguray (29:37.856)
our why story for customers who are adopting BrainBase.
Gokhan Egri (29:41.539)
Yeah, 100%. I think that's one of the most kind of underrated, but important parts of selling to enterprises, especially in AI automation or any automation, because you know, you can sell a very big, a very large grand doors thing that is just like, you know, we're going to replace our entire sales organization with AI and it's going to be 20x more efficient. And then the question is in what metric is it going to make more sales? Is it going to make the sales?
time go down and when you try to do something with that large of a surface area, there's so many things that go in and go out. There's no standard input, standard output that, you change. You don't know where it's actually coming from. So ROI tracking ROI becomes incredibly difficult from just a metric point of view, but also it becomes such a long time horizon thing. When you try to go with this large surface area,
It becomes a thing of, you'll see the benefits in about a year and a half and two years when the adoption hits, when, when your organization gets used to it. Whereas with us, since we're replacing this six people team, we know what they're doing. The six people team, we identify that it's a black box. Essentially the metric is what is the rate that they're able to go from input to output that sometimes that looks at the volume of data coming in, coming out at a time.
Sometimes it looks at, you know, the amount of time it takes for them to go through one request that comes into that box, but it is very easily measurable. So we tell them when we start the POC, when we start a pilot, we tell them, okay, let's measure what is there right now. And then our promise is that our expectation or at least is that it's going to fall down to, you know, one third of that one fourth of that in the next month, in the next two months. So that's the criteria we're going to give it to you as a pilot. So let's go forward.
That makes so much sense. That makes me people so much more comfortable and actually share something. So one of the main metrics that we track at Brainbase for our success is not the amount of people, not how much someone uses a worker or not, you know, how much we're able to charge them for a worker. So those are all kind of proxy metrics. The really kind of the North star metric we have is what is the amount of expansion we have in a company that means
Gokhan Egri (32:06.357)
Every month on average, how many more workers, how many more use cases does a company add using Brainbase? And right now, every month, a company usually adds at least one and a half more workers on average with us based on current customers. And that's, you know, we're trying to increase that, but that kind of shows us that the places that we place the workers, we place the initial worker, it works well enough that people start brainstorming internally of
Alp Uguray (32:09.768)
you
Gokhan Egri (32:36.119)
yeah, actually we could use BrainBase here too, and then here too, and here too. You know, one of the most recent customers we're onboarding, we started with one use case and then in two months they've gone up to roughly eight use cases. And that's incredibly kind of rewarding for us. So that's when we know that what we're selling actually resonates and actually works.
Alp Uguray (32:55.63)
And has the network effects within the customer and seen that the term worker and digital worker shaped a lot over the years. For you guys, how would you define digital worker?
Gokhan Egri (33:13.155)
Yeah. So I think the definition to that would come, somewhere into the black box test that I described. It comes down to how we pick, these teams. So, we kind of visualize it as, know, we go to an organization, we imagine that we can have a layout of all the thousands of employees that sort of organization has. And then we just look at these little segments of connected dots and then we can cover, you know, 10, 20 of those, and we can just put a single brain base worker on that.
So that single brain base worker is doing the job that is being done by all of those, you know, that, that 15, 20 people team, that it covers. that's what a worker looks for us going into a little bit more detail. So technically what a worker looks for us is, our take on how software, at least enterprise software is going to turn out is that the Salesforce and Twio's of the world are paying, you know, billions of dollars.
their engineers for them to be able to hard code what happens when you click on a button. But essentially enterprise software is just crud on a database. It's create, read, update, delete operations on a database. It's just that there's a UI on top of that where every operation has been hard coded. AI where we're going though is we can take that part out completely that you will be able to interface with an AI in one way or the other, you know, through Brainbase, through chat GPT, wherever you have.
And then it'll have a data source that it reads and writes to and it'll just run the correct action on the fly. So that's why Brainbase the workers at Brainbase are structured to have underlying database that they get to access. So if they're an email worker, they can store the emails there. They can retrieve the emails there. If they're an InMoise parsing worker, once the InMoise come in, they can save them there, extract them there. So there's a database portion. And then on top of that, we put our AI, we put our automation.
So that's what a worker looks like for us on the inside. And the idea behind that is that the, the regular humble database is not going anywhere. AI will be able to replace everything else in software, but not database simply because AI is a means of doing amazing approximate retrieval. can do amazing approximate retrieval, but when you need to pull a user's data, AI is absolutely useless in storing that data. It's incredibly inefficient.
Gokhan Egri (35:37.407)
And it's, I would go as far as to it's useless. So you will always have this component of an actual exact retrieval database underneath and the AIs will just need to interface with that and everything else most likely in my opinion will just fade away.
Alp Uguray (35:53.09)
And that's really interesting because as AI, the AI workers, let's call them digital workers, are executing tasks, doing those automations, like retrieving, pulling data or deleting data. How does the transparency side work? For example, when an enterprise, especially adopting it and finding out, wait, we deleted this much data or we added this much.
Gokhan Egri (36:06.605)
Yeah.
Gokhan Egri (36:19.75)
Yeah.
Alp Uguray (36:22.37)
How does that portion work?
Gokhan Egri (36:24.215)
Yeah, 100 % that's one of the most important questions and I think going back to your question about kind of what changed Going forward. I think our messaging around that change But I think also how we visualize it to the end user and how we kind of convinced them if it has also changed so one of the most powerful parts of our brain base is that There's a lot of automation companies out there, know, the Zapier's of the world like the incumbents, but also, you know the newcomer You know, I can name a bunch of them that are like just like
you know, let's go with auto GPT, for example, which works completely in the ether. You don't know what's happening. And that is a huge no -no in terms of enterprise workflow. You need to know what's going to happen for the most part before you sign off on an enterprise workflow. So, you know, the idea of running automation in the ether and just seeing the input and then seeing the hopeful output is not, it does not go with enterprise. And that's what we kind of built Brainbase around.
Brain Maze allows you to see every step and is able to show you every step before the fact, before you run it.
all there, so I won't, I'll say details on like the exact kind of how it works, but the base language, the way we do workflow provisioning is very much unlike auto GPT. It's a place where you use it. It uses only sufficient AI. There's a lot of parts of it where you don't like sending an email is not something that AI should be a part of every part. The AI should probably just write the text and then the sending email, that part is just deterministic. So the way our system works is we call it.
single shot agentic behaviors, SSAB. So what it does is the AI is able to, we have an AI that is able to provision a workflow and then pinpoint locations in that workflow where it will need to come in and actually be in the loop and look at it. Other than that, it's able to provision a mostly deterministic workflow on its own that, you know, it'll be able to remedy if it fails or something like that. So, and the user before they start, before enterprise start this, they're able to see that laid out in
Gokhan Egri (38:28.353)
You know, we, we visualize it using a decision tree, like, you know, it's three structure that we, we use for visualization, but essentially the idea is that you have a good sense of what's going to happen. So you have a good sense of the vulnerabilities and you know, where this might fail and how the AI might remedy it. And that's very important. And that actually is reflected in our UI as well in that. We.
Strive for a hundred percent automation, obviously, but we have three phases of automation with all of our customers that we try to go sequentially. So phase one is 20 % automation, automate the most annoying part of a workflow. And then the rest of it, can still, you can still look. So we want them to test it on phase one for a month or two months, maybe if they want, if this is a very important workflow. Phase two is 80 % automation. Most of it automated. And there's just this last mile part that you still want your humans to do until you have.
you know, faith in the system completely. And then phase three is 99 .9 % automation where, except for the odd kind of thing that happens, you know, once in a hundred thousand requests, you know, it's fully automated. And that's kind of what we strive for. But our UI, our entire interface is designed just like enterprise software that allows you to kind of do that last mile option or, you know, be in charge and be able to visualize everything.
Alp Uguray (39:49.626)
From the perspective of like subject matter experts of like the people who are executing and doing the process, in most cases they're not. They share what they do is typically limited or like typically they're blind spots, right? Like they forget to mention that they do an additional step or they forget to mention or reveal certain things related to that.
So as you're approaching to automation of that task and then like in phase one and in the phase two when it's automated and now it's incumbent on them to grow it or like scale it up. How does that scaling up work? Is it that an employee takes
Gokhan Egri (40:34.985)
Yep. Yep.
Alp Uguray (40:43.394)
the brain base software and then like starts the scaling up like a citizen developer or maybe a low code way or or or the brain base come in and then help them and teach them to to grow it.
Gokhan Egri (40:49.251)
Yeah.
Gokhan Egri (40:57.488)
Yeah, for sure. So that's a great question. So we do it in both ways. So we have multiple ways of doing sales, multiple ways we can go to market. So we do top down as well as bottom up. So top down is at least for us the easiest way of doing it because we talked to a CEO, a CIO, a CTO, and then we tell them what we're providing and we tell them, look, we have a pretty good sense of what we can do with for your company.
For companies in your industry, we've done this before. We think they're going to be applicable, you need to commit to this and then put us in touch with your VPs, your directors, and our team will be able to help them figure out.
Gokhan Egri (41:42.197)
responsibilities in this process is the ability to prescribe what is needed for these companies. And that's not like a consultancy situation. It's not like it's going to take us six months to do this. It's that we need to talk to these people and you know, they know that they want to use AI. They know that they can benefit from it at this point. They've all used chatGVD. They've all seen mid journey on their Facebooks and they need to kind of talk to us to understand, you know, what can it do for me exactly?
In which case we are the experts of the solution. They're the experts of the problem, but we're the experts of the solution. So we come in and based on our knowledge, based on what we've learned, we prescribe to them and we say, okay, you know what? I think for this sales organization, this company, you should start with this solution that we provide. We have provided this before. It's going to take a day for you to implement it. And it's going to take only two, three weeks for you to see results. So let's just get started. So top down, that's how it works.
Bottom up is something that we've more recently started experimenting, but two very good results in that we try to give it as a pilot, an upfront cost -waved type of pilot.
a bunch of leading companies who want to automate it for their teams. And then once that works, what can happen is that once that team works, they usually expand it to other teams just by word of mouth. And then if we get a request from enough teams from that company, we can go to the CEO, CIO, and then we can do the top down motion. But now we don't even have to go into an exploratory mode with them because we know what they want from from Brainbase. So that's kind of how works. then
for the implementation part, we do give a bunch of training. We do give a bunch of documentation, but we also, for large enterprise clients, especially non -technical organizations, which are our most common customers, we essentially kind of help them get onboard and help them get set up. And one of the things I like the most about the process with which we do it is, especially in our fundraising,
Gokhan Egri (43:53.217)
We got a lot of these questions. were like, yeah, we love the product. We love you guys. We love everything, but very horizontal. And how are you guys going to stop from being a consultancy for these companies? And I think that's a thing that, you know, a lot of people have in mind.
key insight there that we have and that we're very happy with is that AI improves our product. Yes, that's obvious. AI improves almost everyone's product. But one thing that gets left behind, one thing that gets overlooked is that AI actually improves the way you build and deploy that product as well. So assuming that you're going to have to put in the same amount of work next year and the year after that in building the product and deploying the product is, I think, very kind of misleading and it's
does not go with kind of how AI is progressing. To that end, we developed an internal AI called Kafka at the Brainbase. there's, have an internal tool called Brainbase Kafka that essentially whenever we get a request from a customer on a worker they want to do, the first place that it goes to is Kafka, which creates the entire worker, at least 80, 90 % of the worker for us. And then we just come in and do the remaining 10 % if needed. So that takes down the
Alp Uguray (45:01.132)
you
Gokhan Egri (45:08.127)
amount of time it takes for us to deploy this product, you know, substantially. We have another one that allows us to feed in documentation for a new integration and it was building integration for Brainbase. you know, the reason we keep from being consultancy and being an actual product is that we actually use AI and we actually count on AI in not only in the product itself, but in the process leading up to the product, leading up to the deployment. And I think that's very important.
Alp Uguray (45:34.83)
That is super interesting. That's very interesting. in a way, AI, as soon as it says Kafka, is reviewing the product feedback requests and then coming into the feature requests and then ranking them, analyzing them or forming the, like you said, like some parts of automation already and then sending it back to the customer, right? Like, and then the customer iterates and that's...
Gokhan Egri (45:42.551)
Yeah.
Gokhan Egri (45:46.369)
Yep.
Gokhan Egri (45:51.747)
Yeah.
Gokhan Egri (46:00.483)
Yeah. Yeah, exactly. So tomorrow we get a request for a new invoice parsing bot, let's say, that says, okay, I want you to sit on our. And then every time there's a HubSpot update, we want you to look at the last invoice from that employee and then, you know, parse that invoice and then check if it's above 50 ,000, you know, send them an email. it's less than 50 ,000, send a Slack message to this employee in our team. So some, some kind of,
complicated process. So when that comes in, we don't look at that. looks at that and gives us a very good prediction. Usually it's able to get to a point where we can deploy that almost immediately what comes out of Kafka. We just kind of tweak a little bit before we bring it to the customer. So we actually have for any new integration, we have a 24 to 48 hour promise to our customers. If you want new integration from us,
By the time you tell us the integration, starting at the time you tell us the integration and give us the documentation for the integration, for the API of that integration, we will get you that integration in 24 to 48 hours because it's fully automated on our backend to create that integration. And that's what makes it scalable and powerful.
Alp Uguray (47:18.806)
Yeah, that is very powerful. It enables actually within an organization as there are more requests. And I've seen that a lot with the automation COEs, right? Like they are typically bombarded with different automation ideas for their pipeline generation. like you would get if you're like there are two types to federated COE and a centralized COE. Like federated COE would sit at
within the department itself, like finance is automating finance and HR is automating HR. So as like I see a direct use case actually when the COEs are getting contacted by the employees and departments themselves, they're like, okay, you get this off my plate. It could be a Kafka as a service even for.
Gokhan Egri (47:53.463)
Yeah.
Gokhan Egri (48:13.377)
Yeah, a hundred percent. mean, yeah, we that's definitely in the books. It's, we only like to sell products that we are very comfortable in that it's going to be used reliably by people who are less informed about it than we are. we were very much. We kind of have the sense that, you know, of any product that we built, since we're the ones building it, we're the best users of it. So we know the tweaks, we know what not to do even internally, like inherently we know it, but.
You know, when you give it to an external user and the supply side with like, you know, things other than brain base for any company, when you give it to an external user, they will be much more cool and substantially more coolest than you are. So you want to minimize the inherent risks in anything you give. And, you know, with some like Kafka, you really want to kind of dot your eyes and, know, cross your T's before you put into production. And that's kind of what we're looking for.
You know, we hope to get a version of it in production for our earlier customers to use by, you know, mid 2025 or, you know, quarter to three of 2025. But, you know, for the time being, it's an internal tool that's just very helpful for us.
Alp Uguray (49:24.824)
So for the time when it scales, let's say it hit thousands of digital workers and then everyone is using it, what is the enterprise AI that, like let's say an enterprise that uses AI that you imagine at the end game, right? the end goal of it. And how does that company look like?
Gokhan Egri (49:30.402)
Yeah.
Yeah.
Gokhan Egri (49:48.951)
Yeah. Yeah, for sure. I think it's going to look like, so I think it depends on kind of the timeframe that you're looking at. think in a long enough timeframe in the next five years, definitely they're going to be companies that are doing above a hundred million to a billion, maybe more that have only one or two employees and the rest is just automated. I think, you know,
SaaS is very, very clear for that. Services companies, I think, are going to be very well automated. And there's going to be a lot of services company that essentially wrap around something like Brainbase. And then they become the human interface for these services for the end customer. So the end customer comes in, talks to this human on this wrapper company. The human just feeds it into Brainbase and the Brainbase does the rest of it. And the human just is able to put a very high markup on it.
because they're the human interface. think that's going to happen very, very fast. And I think brain base and companies that are like brain base are going to be at the forefront of kind of powering that. like I said, our immediate goal, at least for the next two years is that I don't see that there will be very big enterprise. Like I don't see a century 21 anytime soon, getting rid of all of their real estate agents. Cause just from a point of view of like, it's not that they wouldn't want it.
It's that it's just very risky and that I think they would follow in the steps after something is established a little bit more. think it's definitely going to happen at the end. And I think they're going to start slowly experimenting with it, but they're going to keep those in two years. What I think is that these large companies are going to have something like Brainbase doing taking over these four 15 people teams one by one. And then you're going to be left with the really revenue adjacent parts, you know,
your SDRs, your devs, your, you know, the, the HR people who kind of run, keep running it. And I think that's going to be much very important. Another way to look at it, I think is that anything that you have a funnel, you know, there's a funnel and hiring, there's a funnel in sales. Funnel and means that if you have a hundred employees to place on that funnel, you have to place roughly 85 of them at the very top of the funnel so that they can take a lot of those requests. They can take a lot of those things that are very simple.
Alp Uguray (51:56.259)
Mm
Gokhan Egri (52:12.931)
and repetitive. what something like Brain Maze allows immediately is to move people from the top of the funnel to places where they actually close deals and, you know, are actually able to talk to, you know, new applicants. So it's not removing those hundred people. It's taking those 85 people and putting them somewhere where they're to have so much more, you know, value and impact. And I think that's going happen very fast. So you're not going to see from the outside a
Alp Uguray (52:28.908)
I see.
Gokhan Egri (52:41.699)
substantial decrease in the amount of people in that organization, but you're to see much increased efficiency and much increased kind of top line from that because they'll just be able to work much more efficiently. And, you know, I think there's a lot of ways of going to the sci -fi in terms of kind of what we expect companies to look like in the next 50 years. But, you know, I think this is kind of my, you know, my bets that I would put actual money on.
for the next two, five, and ten years.
Alp Uguray (53:13.464)
Just that from the workforce management side of it, that the employees are going to be rewarded for their soft skills, human relationship, as well as dealing with maybe customers, maybe being more on the active side, as well as maybe be very technical. So that they'll be able to understand how to build AI agents.
deploy them, monitor them. So it's like the threshold being knowing a little bit of both, I think won't be enough to be able to excel in that. So for the next stage, I want to do a speed dial if you're okay with it, like to ask you like quick questions and then you can answer whatever comes to your mind first.
Gokhan Egri (53:51.404)
Yeah.
Gokhan Egri (54:01.164)
Yeah, for sure.
Sounds cool. Yeah, sounds cool.
Alp Uguray (54:08.718)
So what was your favorite book that inspired you the most that you tell in even at parties?
Gokhan Egri (54:18.979)
let me think about it. little Prince comes to mind. I don't know how it inspired me, but I have a interesting thing with little Prince in that whenever I travel somewhere, I collect the local copy of little Prince from any country that I go to. So that's what comes to mind, but, I don't think that has impacted me in any kind of meaningful way. it's, I can't really delineate it, but that's what comes to mind. So I'll just go with that. I'll go with little Prince.
Alp Uguray (54:47.874)
Do you get the different languages? Like little prints in different languages?
Gokhan Egri (54:51.137)
Yeah, yeah, yeah. I get, yeah, I have a bunch of languages. So wherever, whenever I go to a country, I get their copy of Little Prince because the reason I pick Little Prince is because I like the book a lot. And also that it just, you know, every country has that book. That's a very common book that every country will have a published version that you can find in the first book store you go to. So I have it in Italian, French, German, you know, I got one from Singapore. I got one from United States. I got local dialects of it in Turkey.
It's just like a little fun thing that I usually try to do.
Alp Uguray (55:25.666)
Yeah, that's cool. mean, it's I wonder how different the book is in between different languages because it's such like the language is very how to say it, right? It's interpretable, right? Like in each language it's written.
Gokhan Egri (55:31.272)
yeah.
Gokhan Egri (55:41.691)
Yeah, for sure. I know this is speed dial, but I actually have the correct answer to this. So I don't know why I think of that. So we have a, have a book at, Brainbase that I give to every employee that comes in and I've actually read it and we have it on our copy table. So I have like loads of copies of it that I give out. It's like our Bible at Brainbase. I can't believe I didn't think of this. It's the Oracle book. So it's the Oracle book called Softwar.
that talks about Oracle and Larry Ellison and Larry Ellison actually has commentary on the footnotes. And this was sent to me by one of our investors, the Javid Kareem, who is the, who's one of the founders of YouTube, who once he invested said that we were like the AI Oracle and that I had to read this book. And then I read the book immediately and then I loved it. And then I give it, I basically give copies out to anyone I meet now in terms of understanding how to run a horizontal company and
how to build a generational company. So that one is a better answer. As much as I love the little prince, that's the actual answer. Yeah.
Alp Uguray (56:41.432)
They're both great answers. mean, yeah, like there are different reads to it, right? Like there's one who you read Harry Potter and then there's the ones that we read. that's amazing. I'll definitely check it out. I haven't read that book. I'll learn more about it. Yeah, you have to. I'll buy the coffee. For the next one, which person inspires you to keep
Gokhan Egri (56:50.433)
Yeah, exactly. Yeah.
Gokhan Egri (56:56.715)
Yeah. Yeah. If I'm ever in Boston, I'll get you on. Yeah. Yeah. For sure. For sure.
Alp Uguray (57:11.064)
pushing forward when you adopt yourself.
Gokhan Egri (57:14.444)
Yeah, so, I mean, my parents are the obvious answer, but to make it something that, you know, the listeners can actually relate to more. I would say Bob Dylan. I love Bob Dylan for a lot of reasons, but I think one thing that is very respectful and some, you know, amazing about him is that he's kind of been like very contrarian in the sense that, you know, he was never very comfortable in one spot that he was in, you know.
acoustic to electric, all that. went into this, know, of Christian, you know, born again phase. you know, a lot of people had a lot of thoughts as he was doing any of these things. But I think he has a sort of restlessness to him that I relate quite a bit. think our culture brain base is very restless. And I think I personally, it's coming from me that I'm a very restless person that
I like to kind of move places and I like to move places in the sense that I like to try new things. And he has a quote that I have as a poster in the BrainBeats offices, is from one of his songs, that he not busy being born is busy dying. And I think that's kind of a very good mantra to live by of, know, just like, if you're not, if you think that you're like where you're supposed to be, you're probably not. And that I just find that to be very helpful personally in business, everything.
Alp Uguray (58:40.928)
And it's like keep pushing yourself because knowing that in where you are at, you'll always be comfortable. It's like that restlessness of keep exploring.
Gokhan Egri (58:51.151)
Exactly. Yeah. Yeah. Yeah. You just got to keep moving and you just got to keep kind of banging to your own drum in a way that, you know, is thoughtful and intentional and is kind of respecting other people's opinions. But it's just kind of, I don't, I think you just keep trying to find yourself and I don't think you ever find yourself. I think you just, the finding yourself is just like the keeping trying to find yourself, which
I think he's done an amazing job with both in his writing, in his songs, and as well as in the way that he's lived. So yeah, he's just, I love him. Yeah.
Alp Uguray (59:28.642)
Yeah, and the exploration keeps going too. In the podcast, I had Gaeko Wazaki come and then we talked about Ikigai, so like the Japanese principle of forming what you love. Yeah, so it's like your Ikigai keeps shaping and then changing over time, aligning with what's in front of you in life.
Gokhan Egri (59:39.937)
Yeah, yeah, Yeah. there's a book on it too. I think I read a book on it. Yeah, yeah, yeah. I remember it.
Gokhan Egri (59:52.194)
Yeah?
Gokhan Egri (59:57.505)
Yeah.
Alp Uguray (59:59.118)
So I think the the the Ori asked for my next question actually.
Gokhan Egri (01:00:04.323)
I'm not doing the speed part of it that well, but I hope I'm doing the answer part of it well.
Alp Uguray (01:00:08.78)
No, this is great. So the other part is as part of being from Turkey and then living in the United States, what was the number one thing that you appreciated as being an immigrant founder in the US and what may be challenged at first?
Gokhan Egri (01:00:35.689)
Yeah. Yeah. Well, a very real challenge is that you, when you're in the United States, you have to keep trying to be in the United States, even though you're, you're creating jobs and you're actually bringing money into the country. The legal immigration path in the United States is difficult to say the least, but in a good way, think immigration, like being a founder from,
a country that is not known for creating a lot of these startups. mean, it's changing. Turkey is creating more and more startups now, which is amazing, which I enjoy very much. I think being an immigrant founder just gives you more of a grit of, you to be there, you've essentially moved up and left your country, your family, everyone, and have come into this new city. Like I went to Boston.
And you have to figure out how to live there. have to figure out how the culture works there. You have to be adaptable. So I think, I think those are all very good soft skills that feed into being a good founder. So I don't, I think it's kind of a chicken in the problem in that it's, it might be survivorship bias, but I think immigrant founders work well because immigrants who have come to that stage are people who have been, who've been filtered through this process to be these adaptable, resourceful, know, relentless people.
And that is something that directly translates to success in almost any field, but definitely in a field like, you know, building a company or entrepreneurship. So I think it's kind of like, I think there's a, I think there is an underlying reason that a lot of immigrant founders are, know, some of the most successful founders ever are immigrants. And I think there's a very big reason for that.
Alp Uguray (01:02:27.192)
think so too, the grit and then the adaptability, like you said, and at an early age to have to be resourceful to be able to do everything at once is one of the biggest drivers for sure. That said, this was a fantastic conversation. Thanks for joining the call. I really enjoyed it. again, anytime, if you're in Boston, I'd love to grab a coffee together.
Gokhan Egri (01:02:35.437)
Yeah.
Gokhan Egri (01:02:42.243)
Yeah.
Gokhan Egri (01:02:46.625)
Amazing. For sure. Yeah, me too. Likewise.
Gokhan Egri (01:02:54.851)
100%. Yeah, if you're an SF, I'll get you your book and you can you should come visit us in our offices. But yeah, I really enjoy this. This is awesome. And I've been really enjoying the podcast anyway. But yeah, just keep doing the good work. And, know, hopefully, hopefully the listeners enjoy it as much as I have during conversation.
Alp Uguray (01:03:13.71)
Yeah, thank you very much,
