Founders in Arms
01:11:53 · Immad Akhund, Jean-Denis Greze
[01:11:53] S0: Good question. You wanna take it first? I I [01:11:55] S2: I think it's a good Parker Conrad question. Mhmm. You know? Because I think he showed that you could do many things at once, and then maybe even... I would say the product quality for any one thing is not that high, but the bundle is worth a lot. So that's, like, one. Part of the answer for you is like there are different kinds of companies even in the old world that you could build. I think with AI, yeah, what's different? Like, it is that the product strategy [01:12:17] S0: that would have made
01:12:17 · Immad Akhund, Jean-Denis Greze
[01:12:17] S0: that would have made [01:12:18] S2: a company work twenty years ago is no longer the right product strategy to have. I think you have to be ready for that. That being said, you gotta be crisp about where you're gonna win in one place. Like, gotta be crisp about what really matters. Because before, maybe you could only do two things. You have to pick the right two things. Now you can do a 100 things. You shouldn't pick a 100. You should probably only just pick five. Five is more than two. Before, you couldn't do five. You probably shouldn't pick a 100. So, you know, we have... There's a document in my company that has three things that are, like, three things that either build network effects or strong differentiation that we think are worth it. And then on the list, there's three other things that are core building blocks that we need to get right. Right? And it's just six things on the list. There's plenty of things that are not on the list that other people are doing. Maybe we get the things wrong, but I think we can only do those six things at the right craft and level of quality that I think we need to in the world.
01:13:13 · Immad Akhund, Rajat Suri, Jean-Denis Greze
[01:13:13] S0: I have three versions of this answer. Number one, I think the best things to build are things our customers are asking for. I think it's very easy to imagine a 100,000,000 features. But if you go and talk to a lot of customers, like, you know, they tend to have really good feature requests. If you have the same thing several times, and then you're like, oh my god. I can't believe we have to do this. That's, like, the easiest, simplest things to build. Number two, I find, like, certain things unlock, like, aligned with the vision and unlock a bunch of things in the future. And then there's other things that are, like, very custom and niche and specific and don't really, like, lead to, like, lots of future time or something like that. So I think, like, things that unlock other things are, like, a useful... A risk to to use. And my third thing was gonna be good, but it's giving me now. Alright. I can't remember. I'll I'll let you know if I think of it. Oh, god. It's funny to [01:14:18] S1: me now. [01:14:18] S2: Anyway, Rajat, I I agree.
01:14:18 · Jean-Denis Greze, Rajat Suri, Immad Akhund
[01:14:18] S2: Anyway, Rajat, I I agree. [01:14:19] S1: Just interrupt me if you remember it. Yeah. I I think, you know, it depends where you are in your journey as well. Like, if you're a start up, you're just making your name in the world. You're you're trying your, you know, heat seeking missile for product market fit and getting some growth. So you're gonna optimize for what's most likely to get you that, and it doesn't really matter what it is. It's like, you know, everything you do is just an experiment to get you to that crazy growth. And so that's what you gotta optimize for. And then even if you're building a different type of company and you're not just looking for growth, you're also just looking for money to just cover your expenses first, and and then you can go for, you know, a larger vision. So, you know, now it becomes more interesting actually once you're profitable. And and I've been in this situation like a few times where it's like, oh, this is a cash machine and generating tons of money. Now what do we do? You can actually take moonshots at that time and and do things that, like, you couldn't even think about before and and go for a much larger vision, take a much more risk. But I I think... So it really... It's really stage dependent in my view. [01:15:24] S0: Yeah. Agree. [01:15:27] S1: Maybe someone whose hands are gone.
01:15:27 · Rajat Suri, Jean-Denis Greze
[01:15:27] S1: Maybe someone whose hands are gone. [01:15:29] S2: Yeah. I have a question for, actually. How do you evaluate how good the answer that you generate are and and use that to, like, maybe self improve your own system? Yeah. So the the answer for you is we don't because we think for business customers from a privacy perspective, looking at what they're talking to their agent about is the wrong stance. So that's like an example where we have a pretty different position than maybe others do. So what we do is we go talk to our customers and we ask them what are you actually using Town to do. And then they tell us, and then we go and try to do those things and learn from it and see how we can improve the trajectories. I think in today's day and age, the reality is the labs are investing in the core that's gonna make our models better. Right? And so the things we control are contacts and contacts objection, like synthesis that that we do so that trajectories are shorter. We can measure how long the trajectory is. And the tools that are being used. So we look at those three things, but we look at it more in, like, an eval part of the universe. And then the last little answer is when things go wrong, some users tell us, and then we ask them, hey. Can we look at this session to try [01:16:47] S1: to understand? Well, we don't do... [01:16:48] S2: We only do it with our permission and not an aggregate way. So we don't have aggregate, let's say, like, evals based on on user data today. [01:17:00] S1: One last question, and then we can go to chatting.
01:17:00 · Rajat Suri, Jean-Denis Greze, S10, Immad Akhund
[01:17:00] S1: One last question, and then we can go to chatting. [01:17:04] S2: Go ahead. [01:17:06] S10: I have a question for all three of you. You all have successful games. So can you name one thing that... Maybe one core thinking or skill set that leads you to where you are right now? And then second is what you're trying to develop a skill set or maybe core thinking that you're to learn or adopt in the next few years of setup. [01:17:24] S0: So the question is, number one, what's the core skill set of thinking that has everything has led us to success? And number two, what do we wanna develop in the future to, I guess, continue? I'll go first. Yeah. I I really like learning new things, and I think I'm, like, pretty good at learning new things. And I think as a CEO, you're often, like, yeah, what your job is for this six months won't necessarily be the same job for the next six months. So I think, like, getting really good at learning, you know, new spaces. Like, every startup I've done has been in [01:18:03] S1: a different space and, you [01:18:04] S0: and learning about, like, you know, what it takes to succeed there. So I think that's, like, a really valuable sort of skill set. Like, not just being good at learning, but enjoying learning. I think for what I wanna develop, that one's, like, kind of a tricky one, so I'm gonna I'm gonna pass. [01:18:20] S1: You're perfect already? [01:18:21] S2: Well, I