Sim: The Visual, End-to-End Agent Builder

1. 0:00 – 0:36

   Building the Visual Agent Builder

   1. AEAaron Epstein0:00

      [upbeat music] I'm excited to welcome Amir from Sim.ai, who is fresh off of a Series A led by Standard Capital. Amir, thanks for joining.

   2. EKEmir Karabeg0:12

      Thanks so much for having me.

   3. AEAaron Epstein0:13

      Yeah.

   4. EKEmir Karabeg0:14

      It's great to be back.

   5. AEAaron Epstein0:14

      I appreciate it. Why don't you maybe kick it off and tell us what is Sim.ai?

   6. EKEmir Karabeg0:18

      Yeah. So Sim is a platform to build AI agents, so it's sort of a Figma-like canvas with cursor style side prompt, where you could build workflows, um, by dragging and dropping or typing in natural language to sort of build an agent to do anything, either inside of your product or to automate tasks for you in the background.

   7. AEAaron Epstein0:35

      Awesome,

2. 0:36 – 2:05

   What Sim Does and Why It Matters

   1. AEAaron Epstein0:36

      and give us the current state of the business.

   2. EKEmir Karabeg0:38

      Yeah. So we just crossed 60,000 developers on our cloud platform, and we're one of the fastest growing open source workflow builders out there right now. Uh, so we just crossed 17 and a half K GitHub stars, which is pretty exciting. So I guess those are, those are the main things we care about.

   3. AEAaron Epstein0:52

      Yeah. That's amazing, especially 'cause you guys were just in the spring batch not too long ago, right?

   4. EKEmir Karabeg0:56

      Yeah. It feels like yesterday. [laughs] It kind of was.

   5. AEAaron Epstein0:59

      Yeah, it kind of was just-

   6. EKEmir Karabeg1:00

      [laughs]

   7. AEAaron Epstein1:00

      ... like yesterday. That's awesome. Maybe take us back to the early days. How, how did you and your co-founder Waleed first meet?

   8. EKEmir Karabeg1:06

      Yeah, so we actually met in Berkeley. Uh, I was coming... It was my sophomore year, right be- right before, uh, my sophomore year started, and I was-- It was COVID my freshman year. So I came to Berkeley, and I knew nobody, and I was paired with my roommate, uh, at the time, who I didn't know either, and his best friend from high school was Waleed.

   9. AEAaron Epstein1:24

      Huh.

   10. EKEmir Karabeg1:24

       So we got to know each other, like, right before school started, and he was kind of my first friend at Berkeley, and then we've just been best friends ever since there. So we, like, spent all of our time together in undergrad, and then once we graduated last year, uh, I knew I wanted to start a company with him, and he was a year older than me, so he was working actually down in LA at Amazon. And I literally called him, and I was like, "Hey, dude, move to SF with me and, like, let's start a company." And, like, on the phone call, he just quit his job and moved to SF. [laughs]

   11. AEAaron Epstein1:50

       How long had you guys known each other at that point?

   12. EKEmir Karabeg1:52

       Uh, for about, like, three or four years.

   13. AEAaron Epstein1:53

       Okay.

   14. EKEmir Karabeg1:54

       But we had pretty much hung out every single day in that time span.

   15. AEAaron Epstein1:56

       And had you built stuff before together or-

   16. EKEmir Karabeg1:58

       No, not at all. Not even, like, as much as a project. We basically were just best friends, and he was, like, one of the smartest people I knew, like irrespective of that.

3. 2:05 – 4:20

   From a SF Apartment to 60,000 Developers

   1. EKEmir Karabeg2:05

      And so we were just like, I know I wanna do something with him, and so, like, just come to SF. And actually, I kind of tricked him a little bit. At first, I told him to just come to SF for a month, and I was like, "Just stay with me for a month, and, like, we'll figure stuff out." And then he ended up just never going back home, and so [laughs] things spun out from there.

   2. AEAaron Epstein2:21

      That's amazing.

   3. EKEmir Karabeg2:21

      Yeah.

   4. AEAaron Epstein2:21

      And so did you guys have an idea in mind when, when you first decided to start working together? Or you're just like, this is just a person that I want to work with, and we'll figure out the rest later?

   5. EKEmir Karabeg2:29

      Yeah. It's, it's actually kind of funny. So when he first came to SF, I pitched him on this idea, like, pretty much the first day he landed.

   6. AEAaron Epstein2:35

      This specific idea of Sim?

   7. EKEmir Karabeg2:37

      Uh, not actually this specific idea, a previous idea.

   8. AEAaron Epstein2:39

      Okay.

   9. EKEmir Karabeg2:39

      So the idea was it's kind of like a dynamic, uh, landing page where based on your, like, browsing history or if you have, like, data about someone in a CRM, you show up to a landing page, and right now, if you're on a landing page, like, it speaks to the general audience, right? And then typically, you try and optimize, like, the language or, like, the positioning of certain things on the page based on, you know, like, the overall metrics. But I was like, what if we could customize a website just for one person? Um, and then we ended up just trying to sell that to sales and marketing companies, but that was, like, kind of the worst idea just because we knew nothing about sales and marketing. We didn't understand their problems whatsoever. And so the first day he landed in San Francisco, I'm like, "I have this idea for a company. Like, it's gonna be huge." And he was like, "I'm in," and then he just quit his- wanna spend the next, like, twenty years of our life, you know, doing this. So we were just like, "All right, let's figure out another idea."

   10. AEAaron Epstein3:33

       And, and was that an idea that you guys had applied to YC in the past with?

   11. EKEmir Karabeg3:37

       Yeah. So that was actually the first time we applied to YC, and I remember that was, like, probably a reality check for us. I think it was sort of the end of November. I wanna say it was, like, November 30th.

   12. AEAaron Epstein3:45

       Mm-hmm.

   13. EKEmir Karabeg3:46

       And we got on an interview actually with Michael Seibel and, uh, Diana, and they asked us like, "Oh, so what do you guys do?" And we like explained to them this whole platform, and basically, we spent the whole ten minutes trying to explain what the, what the app did because it was so confusing.

   14. AEAaron Epstein4:03

       Yeah.

   15. EKEmir Karabeg4:04

       And I think eventually after that call, we just realized, like, we didn't know what we were doing.

   16. AEAaron Epstein4:07

       Yeah.

   17. EKEmir Karabeg4:08

       And so we spent all of December basically trying to figure out what we should build, and I kept coming back to this idea of building agents because that last company we built, um, or the startup that we were working on was called Scalemont, and

4. 4:20 – 7:05

   Finding Early Users and Product-Market Fit

   1. EKEmir Karabeg4:20

      when we were working on Scalemont, we basically had, like, hundreds of agents. Like imagine an agent that, like, did research on your background and then tried to create like a hero section, right? And so you have to create, like, all of the text for the hero section. Then you wanna generate images, right, 'cause, like, typically a website has images. So then you have to go have a small LLM that generates a prompt for an image generator. Then that goes, passes it to the image generator. Then that image generator gets you the image. You put it back in your website. You create, like, all these case studies and footer sections and everything. So it, it was just such a complicated, over-engineered project. But we realized, like, actually the biggest problem of that project or, like, the thing that we cared about the most was building agents. And just between Waleed and I, it was hard for us to even understand, like, "Hey, did you already build an agent that did that thing? Like, that researched that person and got their history or, like, maybe their background?" And so that was kind of the idea for Sim, where it's like, hey, why don't we have this platform where we can actually visually orchestrate these agentic workflows as opposed to just writing it in code? And so that was actually the birth of Sim.

   2. AEAaron Epstein5:20

      So it came out of your own pro- Was there, like, a light bulb moment that, like, this is what we should build, or, uh, like how did you realize this would be the thing?

   3. EKEmir Karabeg5:34

      Sort of. It wasn't necessarily, like, one moment, but I remember we spent all of December of last year basically trying to find what we were gonna build. And I remember we were looking into everything, basically like healthcare, logistics, all sorts of stuff. But I was like, I don't know anything about healthcare, nor do I know anything about logistics or any other industry. I basically only know things about LLMs and building agents. So I remember it was January 5th, and I was like-I'm a super big power user of Figma, and I love Jupyter Notebook, and I was like, "I would just love if those products were just, like, one and the same." I don't know what that would look like, but, like, I kinda wanna figure it out. And so it was January 5th, and I was like, "And I wanna build a platform, and I want it to be open source, and I want you to build agents in sort of this Figma environment." Waleed's like, "That sounds really stupid." And I was like, "Yeah, you're right, but, like, let's just try it for five days, and let's see what happens." He was like, "All right. I'll let you do that," while he does some research. So I spent five days working on it, and on January 10th, Waleed was like, "All right. You're dumb enough to do this for five days. I'll join you on January-" Like, "This is idiotic," and, like, "No one's gonna use this."

   4. AEAaron Epstein6:41

      Yeah.

   5. EKEmir Karabeg6:42

      I was like, "You're right, but it's something I care about." Honestly, if nothing else, I was gonna build agents on that platform just because we saw the pain from our previous project.

   6. AEAaron Epstein6:49

      Yeah.

   7. EKEmir Karabeg6:50

      And what's funny was when we started, he was like, "What's your goal with this?" And I was like, "If we get this to 100 GitHub stars, I'll be, like, super happy." He's like, "100 is way too many." I was like, "Actually, you're right. So if it gets to 10, I'll be satisfied." And that was back in January, so it just...

5. 7:05 – 10:30

   Designing an End-to-End Platform for Agents

   1. EKEmir Karabeg7:05

      And, and it's-- We wouldn't even have believed if you had told us that we were at, like, 500 stars now, um, in January.

   2. AEAaron Epstein7:11

      What was the moment after you decided to pivot to this where you were like, "I think this is actually the thing that's gonna stick"?

   3. EKEmir Karabeg7:17

      I think once Waleed joined, uh [laughs]

   4. AEAaron Epstein7:18

      Okay.

   5. EKEmir Karabeg7:19

      I think once-

   6. AEAaron Epstein7:19

      So he wasn't a believer, but you were convinced at that point.

   7. EKEmir Karabeg7:22

      Yeah [laughs] . I think I'm more optimistic, and I think Waleed's more of a skeptic. And so I think once Waleed gives me the go-ahead, um, that we can, like, work on something, then I think that's, like, a sign that it's actually gonna succeed. Um, and so I think once he joined in January, I think from that point onwards, there wasn't a single day we looked back. We were just building from then until now.

   8. AEAaron Epstein7:43

      Any advice that you would give to people going through that same, you know, kind of pivot journey and, and hopefully help them find their way out sooner?

   9. EKEmir Karabeg7:50

      Yeah, definitely. I think the biggest thing for us while we were pivoting was we were trying to find... perhaps reach too far away from what we understood.

   10. AEAaron Epstein7:57

       Mm-hmm.

   11. EKEmir Karabeg7:57

       You know, I, I think as an engineer, we often underplay what we understand, um, just, like, about coding or just about the world, and if you have to-- If you're really, really critical about what you understand perhaps better than others, um, then you should lean into that more. And so I think oftentimes, at, at least when we were pivoting, um, you know, we were looking for all these far-fetched ideas. But what we should have done is just looked at what we had already built together, um, and just try things. You know, just ship ideas and, you know, you can iterate, like, 15 times before landing on an idea. And I think that's one of the things about YC that I really loved is just, like, that openness to, to pivoting.

   12. AEAaron Epstein8:35

       Yep. Had you tried a bunch of the other, like, workflow agent builder tools out there and discovered that they weren't right for you? Or talk about your experience there and how you look to differentiate the product that, that you're building with Sim.

   13. EKEmir Karabeg8:47

       Yeah, it's a good question. I, I think when we started, um, I had tried a couple of other workflow tools, but it sort of felt to me like the traditional workflow automation that had existed for basically decades, actually. Workflow software is not new. It's been around for, you know, decades. Um, and so when I tried them, I felt that they were not AI native in the way that I wanted them to be. For instance, if I want to orchestrate one agent to do the same thing 100 times over and then, you know, make, like, some synthesized response for me or if I want to even build a workflow with natural language, um, some of these things that I felt like were already accessible in programming were not accessible on workflow platforms. And so I figured that workflows are kind of the next paradigm for programming, where we're kind of moving up the abstraction stack. You know, you can see this with Cursor, where, you know, most of code today is just generated by an LLM, and I think in the future that actually most, um, software will be orchestrated inside of workflows rather than just in pure code.

   14. AEAaron Epstein9:43

       Mm-hmm. Yeah. That's a, a common trend that we've seen where everything just keeps becoming more and more abstracted, and you get further away from the pixels or the actual source code, and, um, it seems like you guys are, are really, uh, leading that trend for a lot of people.

   15. EKEmir Karabeg9:57

       Totally. I mean, if you were going back to, like, the '50s or the '60s and you told people they wouldn't be programming with punch cards, they'd be like, "What does programming even look like?" Right? [laughs] And so obviously, like, the modality in which we've programmed has changed a lot over the course of history, and I feel like this paradigm is the next step. And I feel like we're always going towards the path of least resistance, and if we can truly be the easiest way to ship some software or some, some agent, then why wouldn't we be the next paradigm?

   16. AEAaron Epstein10:25

       Yeah. Um, you would've been really happy getting to 100 GitHub stars. You're at, like,

6. 10:30 – 13:25

   Powering the Open Source Agent Ecosystem

   1. AEAaron Epstein10:30

      17,000.

   2. EKEmir Karabeg10:31

      [laughs]

   3. AEAaron Epstein10:31

      More than that right now.

   4. EKEmir Karabeg10:32

      Yeah [laughs] .

   5. AEAaron Epstein10:33

      Uh, not that much later. You know, so you've really been able to figure that out, and, like, your open source traction has, has really been blowing up. H-how'd you pull that off, especially for any other founders out there building open source products and that maybe are in your, your shoes early on, where, like, they'd love to just get to 100? What advice would you have for them right now?

   6. EKEmir Karabeg10:50

      Yeah. I mean, I would say two things. Um, the first thing is that we found a space where, first of all, the other platforms are not fully open source, uh, and we really actually wanted to... We believed in the ethos of being open source, um, which is that, you know, we, we think that in the future, basically LLMs are gonna just be able to one-shot some code base. Like, imagine if you had AGI, and you gave it some task, it could probably go and replicate an entire application within minutes. So code is no longer the proprietary part of a company anymore as it used to be. We also wanted to kind of, like, give this software away to the world because we figured that the platform is much more valuable once it's interconnected and kind of having this network of people on it, as opposed to the workflows running themselves. But we should not sort of constrain the workflows away from the open source community in the sense that we want people to build on top of Sim. We kind of want it to be, like, the foundational workflow platform, and then everyone can customize it to their own needs. Like, for instance, if you're bringing it into healthcare or into logistics, then you can just, like, take out pieces of Sim and put your own layers on top of it, uh, and we're kind of this, like, underlying engine.Um, so that was kind of like we wanted to just stick to the open source philosophy, and I, I feel like that actually bodes well with the community because I, I feel that developers are honest and they, they can kind of see through BS a little bit. And so if you're just truly open source and you actually are just Apache 2.0 or MIT licensed, and you give your code away for free, I think people will be responsive to that. And then the second thing is, like, actually just finding developers where they are. Um, so we posted a lot on Hacker News. We went to, like, Twitter and LinkedIn, I mean, LinkedIn less important, but, you know, Twitter and Hacker News, and we kind of just put our project out into the world. Uh, and the response was obviously great, as you can tell, um, so we're kind of continuing in that direction.

   7. AEAaron Epstein12:42

      And how are you guys ... You hit number one on GitHub Trending and, and that seemed to really boost a lot of your, your stars and adoption. Were there any strategies that you used to, to make that happen, or was that just kind of luck that a lot of people got into it at the same time?

   8. EKEmir Karabeg12:55

      Yeah, it's really interesting. This kind of trending concept, uh, is something that we've been really focused on, um, in the course of Sim. And what we've realized is the majority of our stars come while you're on GitHub Trending, and to be completely frank, we weren't even really sure what was GitHub Trending before we had started our project.

   9. AEAaron Epstein13:12

      Mm-hmm.

   10. EKEmir Karabeg13:12

       And then I realized later that GitHub Trending is super important, uh, because when you're featured on GitHub Trending, typically you'll get in- anywhere between like, you know, 250 to, like, thousands of stars in a single day. Our kind of drive and our, uh,

7. 13:25 – 16:20

   Raising a $7M Series A

   1. EKEmir Karabeg13:25

      launch schedules are always, like, around how we can optimize for GitHub Trending.

   2. AEAaron Epstein13:28

      Mm-hmm.

   3. EKEmir Karabeg13:30

      Um, and so how we basically got on there is just to organize, like, some orchestrated launch, like, to let a l- big community of, of, like, developers know about your product all at the same time. So if we go out on Hacker News and on Product Hunt and on Twitter kind of all at the same time, and you coordinate these launches, then you can start to develop some traction around your GitHub repo.

   4. AEAaron Epstein13:48

      How have you been able to convert some of this open source interest into, like, actual paying customers?

   5. EKEmir Karabeg13:53

      Well, we have a cloud platform. Uh, so Sim.ai is our cloud platform, and it's linked in our GitHub repo. And so if you actually map, um, charts of our, um, o- like, our GitHub stars to the new users on the platform, it kind of follows the same pattern. Uh, there are obviously some external factors that'll make the cloud platform grow much faster. Like, for instance, if some workflow or AI influencer finds out about, you know, Sim and they make a video about it, that's kind of a channel that typically goes to the cloud platform as opposed to the open source repo. But originally, pretty much all of our new users trended in the same exact way with our GitHub stars, and so we could basically map, like, a one-to-one plot with, um, our GitHub stars and our, and our cloud users. And then once they come to our cloud platform, then it's sort of just easier to get set up on Sim to build an agent as opposed to running it locally. Of course, you still have that option, but for people that want a little more convenience, they'll just come onto the cloud platform and then pay for inference on there.

   6. AEAaron Epstein14:50

      Cool. And any kind of general advice for open source founders?

   7. EKEmir Karabeg14:54

      The first thing i- I think is to be, like, truly open source. Um, I feel like people are worried that their code is proprietary or that, you know, someone's gonna steal their idea. Um, but no one can steal your vision. And so you can put all your code out there, and we know that there are competitors that are looking at our platform and definitely wanna take ideas from there. But we're not scared whatsoever just because we have our own vision for what we want the product to look like, and it's gonna change the platform drastically in three months from now or six months from now than it is today.

   8. AEAaron Epstein15:25

      Mm-hmm.

   9. EKEmir Karabeg15:25

      Um, so I'm not necessarily concerned if someone takes the product right now because ultimately, we believe in the vision of the company, and no one can take that away from, from your repo. So, you know, the thing that I would, I would advise founders that are building open source products the most is to be, one, truly open source, and to, two, engage with your community. Um, you know, we do a lot on, like, Discord and on Hacker News and on Twitter. Uh, and so kind of, like, constantly talking to your users and, like, the dev community and understanding what they like and what they don't like about your project can also help you kind of, like, learn how to build a better product, but also build this community of developers around your repo.

   10. AEAaron Epstein16:02

       You've grown i- incredibly quickly and, you know, had a, a lot of success since the batch, which was, I don't know, seven months ago or something like that probably.

   11. EKEmir Karabeg16:10

       Yeah.

   12. AEAaron Epstein16:10

       What's been the hardest part of the journey for you that you feel like you've had to overcome so far?

   13. EKEmir Karabeg16:14

       I, I think probably the hardest part, um, is just forcing ourselves to

8. 16:20 – 20:10

   Building Agents That Actually Work

   1. EKEmir Karabeg16:20

      be better every single day, to never be satisfied. Um, and so for instance, like, our whole team will have a discussion about, like, the border radius of a button for, like, three hours un- until someone just, like, [laughs] gives up on the, on the argument. Um, but I, I think the, actually the hardest thing is just, like, challenging or, like, being content with certain things and just, like, moving forward as quickly as possible. Um, I think that when you're in YC, you're kind of in this mode of, like, launching all the time. And when we got out of YC, we were extremely deliberate about consistently launching, you know, outside of the batch, and we actually grew much faster outside of YC than we did inside of YC. And I think it's because we kind of treated the batch just like, as a small stepping stone on where, you know, our company future will be. Um, and so we kind of took those philosophies from YC of, like, launching, like, very often, being very public, like, building momentum, and we kind of took that after the batch. Uh, and so I think that's wh- how we were able to grow so fast. I mean, I think when we ended the batch, we were, like, around 5,000 developers on the platform.

   2. AEAaron Epstein17:23

      Mm-hmm.

   3. EKEmir Karabeg17:24

      And then now just, like, four months later, we're at, like, 50 thou- or 60,000. Um-

   4. AEAaron Epstein17:29

      Yeah

   5. EKEmir Karabeg17:29

      ... so I think it's just about, like, constantly launching and, like, constantly forcing yourself to just, like, be better and improve the platform every day.

   6. AEAaron Epstein17:35

      Yeah. Are there any tactical things that you do to, like, maintain that sense of urgency? Hours and two weeks, we gotta hit our goal. What are some of the things that you do post-batch to be able to keep that sense of urgency, and it sounds like even ramp it up?

   7. EKEmir Karabeg17:54

      Yeah, definitely. Well, we have a launch calendar, and so I think that thing is something that we stay very true to, where every single two ... Like, every two weeks, we launch something.

   8. AEAaron Epstein18:01

      Mm-hmm.

   9. EKEmir Karabeg18:01

      Uh, and there's a really good quote that I love and Waleed loves. It was from one of the creators of Saturday Night Live, and they saidUm, we don't go live because it's ready, we go live because it's Saturday night. You know, it's 10:30 PM on Saturday night. And so that's kind of our philosophy as well. It's like we al- you know, engineers I think typically wait to release product until they think it's perfect.

   10. AEAaron Epstein18:20

       Mm-hmm.

   11. EKEmir Karabeg18:20

       Uh, and that's just kind of never really works, and I, I'm sure you've had experience with this at your startup as well. Um, a- and for us, like we're, we're just consistently launching. So every two weeks we put something out there. It doesn't matter if it's almost nothing, it's just that whatever we have ready. So it kind of forces everyone... It's a kind of, it's a forcing function to get everyone around like some product release.

   12. AEAaron Epstein18:41

       Yeah. That, that's great. Um, it's an important muscle to build and, and it's important that it's in the culture, 'cause, you know, culture drives a lot of these decisions. Um, a- along those lines, I know you guys are hiring, um, what is the ideal person that comes to work at Sim actually look like?

   13. EKEmir Karabeg18:56

       The ideal person that comes to work at Sim is someone who's just extremely excited about the future of AI. Um, something that I fundamentally believe is that, you know, a lot of people talk about AGI is coming, AGI is coming next year, maybe it's here right now. Uh, it actually doesn't matter if AGI comes or not. And my opinion on that is we actually have solved all the scientific problems that we've needed to completely revolutionize the world. I think what's actually left is an engineering challenge, not a scientific one. And so I think actually being able to democratize the way that developers build agents, or actually can, anyone can build agents, not just developers, is extremely powerful. And so I think if you gave a platform like Sim into the hands of, you know, people who have aspired to build agents for their own personal use cases or inside of their company, um, then they could completely transform their own business and perhaps like their lives, um, with agents. And so I think for someone who is interested in, in the future of AI and kind of what AI holds from all, like, the latest model developments to, like, new tools and new stacks, someone who's just curious about the future of AI and would like to actually implement that in some kind of product where we can actually change the

9. 20:10 – 23:40

   The Infrastructure for the Agent Era

   1. EKEmir Karabeg20:10

      way that people build on top of AI, I think this is a perfect candidate for Sim. And so I think someone generally that's just intelligent. They don't even have to be like a TypeScript developer or... I remember like one of our founding engineers, when he first came, you know, we were talking to him about like all these features we're gonna build and, um, he was like, "Yeah," openly like, "I literally don't know TypeScript, but I know Go." And we're like, "Sure, just write it in Go." And so [laughs] it's just like it doesn't necessarily matter about your capabilities, it just matters about your drive and, and how curious you are about the future.

   2. AEAaron Epstein20:38

      Cool. Yeah, I heard curiosity, people building on the cutting edge, interest in AI, and obviously ship, right? Like, that's, I'm sure that's a requirement for you if every two weeks you're shipping, you know, whether it's ready or not.

   3. EKEmir Karabeg20:51

      Yeah, and I think also that we're launching, and it's because we care so much. Um, and I think the one thing that you can't replace in someone is just their passion and their love for building something. And so I think if you don't have that, you can't have anything. Um, so we're just really looking for passionate people who really just wanna drive the, the future of AI and this vision that we have.

   4. AEAaron Epstein21:15

      Yeah. Cool. How have you changed as a founder since starting this?

   5. EKEmir Karabeg21:18

      Yeah, I mean, [laughs] I think, uh, I've changed in a lot of ways. I'm certain, certain that Waleed has changed in a lot of ways. And I think the main way that I've probably changed, um, i- is realizing that it takes 10 years, um, to do, to, to, to build something great. And so, you know, when you're first getting started, you might think that things are gonna take, um, a shorter amount of time than, than they are. And I think Brian Chesky always brings up that quote, which is like, "People, um, overestimate what they can do in a week, and, like, underestimate what they can do in 10 years," or I for- it's like a quarter or 10 years or something like that.

   6. AEAaron Epstein21:49

      Yep.

   7. EKEmir Karabeg21:49

      Um, but I, I, like, fundamentally believe in that. And so I think that what we, like, really aspire to do is just show up every single day. Um, so we come into the office at 9:00 AM, we typically leave around, like, 10:00 PM, and we do that seven days a week, um, or six days a week. Having that... I, I think if you do that every single day, it just compounds your, your growth, um, over a long period of time. So it's, it's also about having balance. I think a lot of people, especially in YC, when you're in the batch, you're, like, constantly ship- It's really important to have balance. Like, find time to go to the gym, find time to eat healthy, cook the food you're gonna, you know, typically eat, and to keep, like, a healthy lifestyle. You have to make sure that you're taking care of yourself as much as you are the company. Uh, and so I think that, like, my mentality had to mature a little bit to be, you know, like, just a, a good founder as opposed to when it was just Waleed and I in my apartment and just, like, sleeping on the couch and eating whatever and never going to the gym. [laughs] So I think if we wanna operate for the next 10 or 15 years, we have to keep this consistent and healthy lifestyle.

   8. AEAaron Epstein22:58

      Any advice for, for founders that are, are watching this and, and hoping to follow in your footsteps?

   9. EKEmir Karabeg23:03

      Again, a- as I said, just staying true to yourself. Um, so, you know, Waleed and I, um, we always were just, like, being honest with ourselves, being honest with each other, um, being healthy, and just, like, listening to our own, like, minds and bodies and, and pushing things, of course, in the way that we, you know, we both wanted. Um, but kind of the important thing for, for both of us is just to, like, be very honest about what we understand, especially what we don't understand, uh, and just to keep learning. And I think, um, one actually big thing that I would say is that oftentimes in the past I struggled with optimizing for, like, short-term growth or short-term learnings, whereas we

10. 23:40 – 25:05

    Advice for Builders in AI

    1. EKEmir Karabeg23:40

       should kind of... You're, you're kind of in a balance of optimizing between, like, long-term learnings and, like, long-term growth. And so I think if you think of things in the timeline of, like, 10 or 20 years, you would much rather optimize for being, like, constantly learning for the next nine and a half years as opposed to spending the next half year doing something that perhaps, like, you know, you're not interested or you wouldn't be doing. So for instance, like, the sales and marketing company that we tried to start, you know, we were kind of too shortsighted in the sense of, like, could we see ourselves doing this in, like, the worst moment of our lives in, like, six years' time? And if the answer is no, then you'll probably fail at that six-year mark.

    2. AEAaron Epstein24:17

       Mm-hmm.

    3. EKEmir Karabeg24:17

       And the thing that Waleed and I realized when we started this company is, like, we could be doing this for five or 10 or 15 years with completely no success, and we would just be having so much fun doing it that we would keep doing it. And I think it's even important not to lose sight of that. For instance, I spent a lot of time selling and not programming in the past couple of months, and I realized that the thing I love most is programming, and that's also the thing that I'm best at. So I got back to programming, and I think we're always trying to optimize for putting ourselves in a position where we're, like, l- learning for the long term and, like, trying to succeed for ourselves in the long term.

    4. AEAaron Epstein24:53

       Well, Amir, thank you so much for joining.

    5. EKEmir Karabeg24:55

       Thanks so much for having me.

    6. AEAaron Epstein24:56

       Yeah, it's great to hear your story.

    7. EKEmir Karabeg24:56

       Really appreciate it. Thanks. [outro music]

Episode duration: 25:05