Founders in Arms
00:19:44 · Immad Akhund, Rajat Suri, Jean-Denis Greze
[00:19:44] S0: Oh, on the assistant side. Yeah. This is what? [00:19:47] S1: Does anyone now use AI for email calendar today? One person does not use AI. Two people. Because that... That's, like, six months ago, everyone... And that's, like, a this year thing. Right? You... AI assistant for email calendar. For sure. [00:20:00] S0: Yeah. I have a complete four things. Go ahead. So, obviously, you were a engineer and a CTO for a really long time. Yeah. What's been the most surprising thing about kinda starting a startup? Like, how are you building things and how how does that mean, like, kinda... Like, do you do you read the code that you deploy nowadays? [00:20:26] S2: No. Actually, I read the code that I deploy because I should probably not be writing code that often, and I do. And I don't write code all the time, and so I no longer know what the code base is doing. [00:20:38] S1: So I
00:20:39 · Jean-Denis Greze
[00:20:39] S2: think I have to be a little bit more careful. What is most surprising? Well, I think one of the things you don't... When you're an exec somewhere, there's someone called the CEO, and, like, the CEO kind of always has your back. Even if you're gonna make a decision that you feel is just yours, at the end of the day, if it's a really dumb decision, the CEO will probably tap you on the shoulder and be like, bro, or bro et, let's talk. And and you have those conversations, and, you know, then there's a hard call to be made. And even though the CEO says it's your call, you know he only says that or she only says that to you when it's a call that could be yours because it's not company ending if it's in the other direction. So the weirdest thing about being a founder and a CEO is there's nobody behind me. And it it just really means, like, anything that happens, anything at any level that's not, I think, the way it should be, I do wake up and stress out about whether it should have been the other way. And there's no excuses. And and, obviously, you have to then decide, do I bother people and... By imposing my will or by convincing them, which takes time and effort. Not always worth it. But, like, I didn't internalize how much pressure just accumulates in a CEO role at all times. And you asked me about being stressed. I'm relaxed, but I do wake up at about 4AM, I would say, like, six out of seven nights just thinking about How do you sleep? Like, eleven. So, you know It's that stress. Yeah. No. No. Exactly. And I wake up, and I I look at the ceiling, and I'm like, I think about the things. Yeah. You know? And I think before that, I slept a little bit better because of of Zach, who was, you know, Pod's great CEO.
00:22:23 · Rajat Suri, Immad Akhund, Jean-Denis Greze
[00:22:23] S1: It takes many years, I think, right, to get used to it. It's not like a overnight thing, and it is such a different role even though, you know, you know, you... People have been in other c suite roles. Sometimes they think it's like, oh, it's just one step, but it's actually dramatically different than what I found. I don't know you found. I might just [00:22:43] S0: I've been a CEO forever. So Only been for forever. For twenty years. So I had... I only had one job. [00:22:51] S1: And the first five years the first five years, like, ten... You know, how have you adapted to the to the role? Like, did you feel that stress earlier on more? [00:23:02] S0: So, yeah, if I don't sleep, I get, like... I'm very low productivity. So, again, I have to come up with mechanisms for sleeping well. I think the key for me is, like, stop work at least three hours before I go to sleep. And then, yeah, go read or do do something else. But if I'm if I'm sleeping, if I'm working and then trying to sleep straight away, it doesn't work for me. Yeah. Yeah. Did you get stressed out? [00:23:26] S1: I will see you stressed out. [00:23:27] S2: Not stressed out. And I I get stressed, [00:23:30] S0: but not not that much. Yeah. I mean, I I feel like I'm in bonus levels of life now. [00:23:35] S2: Yeah. He's winning. So much winning. [00:23:37] S0: But Right. I I think I was relatively low stress founder, which I think you can have to find a way to [00:23:45] S2: be low trust in order to, like, [00:23:46] S0: do it for several decades.
00:23:46 · Immad Akhund, Rajat Suri, Jean-Denis Greze
[00:23:46] S0: do it for several decades. [00:23:48] S1: Yeah. I think you get comfortable with those level of decision making and the stakes of every decision. Yeah. [00:23:54] S0: Have you kinda thought about the types of people you hire now versus where you were hiring a client? Like, is it a super different profile, you know, smaller team, like, dynamics? [00:24:05] S2: We have a lot of clients at town. All of nice Plaids. You know, look, when you're when you're smaller and you need to be close to your customers, I personally have always valued engineers, especially who like talking to customers. Engineers who do not care about the technical complexity of work, but simply about whether the work is important for the business. About people who are great communicators with others at the company and with others externally, and people who do not need to be managed, meaning who are never complaining. I like those people at all scales of companies. I think having only people like that early on is extremely beneficial, and those dimensions, I think, are the same. What is different for me is AI nativeness, but not an engineering role. I mean, engineering roles, it matters a ton. But we... We've realized that if anyone's in product or marketing or community or whatever it is, if they can be AI native, it is a incredible benefit. Like... How you how [00:25:25] S0: do you test for AI nativeness? [00:25:27] S2: We ask them what they build with AI, either on top of town or something else before, you know, and we just ask them to show us what it is.
00:25:27 · Jean-Denis Greze, Immad Akhund, Rajat Suri
[00:25:27] S2: We ask them what they build with AI, either on top of town or something else before, you know, and we just ask them to show us what it is. [00:25:36] S0: What... Like, what's the best answer you've got? No. No. [00:25:38] S2: They show us... They, like, show us an agent that they have that, like, help them find jobs or... You know what I mean? They're just like, I wanna see the thing that you've tinkered on. I would say the thing that I wanna find is the tinker gene, where once something's not working, you're... You have the engineering reflex of, like, can I build a thing for that? I almost don't care how good the thing is, to be honest. It's the tinkering gene. The desire to do that is very important. [00:26:02] S1: How have you adapted your engineering interview process at [00:26:05] S2: the age [00:26:05] S1: of AI? You can't do the coding, deep coding anymore. Right?
00:26:05 · Rajat Suri, Jean-Denis Greze, S3, Immad Akhund
[00:26:05] S1: of AI? You can't do the coding, deep coding anymore. Right? [00:26:09] S2: Yeah. We... I mean, we do an architecture interview and a previous work interview. And then on the coding, we have, like, a two and a half to three hour assignment where, like, in person, where we actually put you in our code base with Claude or whatever agent you wanna use, and we ask you to add something to the product. And what we're testing actually is on how you use AI along the way, and how are you making the right trade offs between not looking at the code and looking at the code. And so usually the feature will have some places where it doesn't matter if you look at the code, but have a couple of places where you actually do need to look at the code to understand how to prompt to to get the right outcome. Yeah. We're just trying to find people who have that, you know, like the right judgment about one, [00:26:51] S3: and other. [00:26:51] S0: This is like on your full code base or some sort of simplified version [00:26:55] S2: of [00:26:55] S0: the [00:26:56] S2: code base? It's like a stuck in time a while back version of that code base. Yeah. We can't do it on a movie code base, so I don't wanna expose my entire code base to candidates. Yeah. So it's it's, like, six months old or something [00:27:06] S1: like the same feature for all candidates? [00:27:09] S2: There's a couple of variants. Couple of variants. [00:27:11] S1: Yeah. Yeah. Pretty similar. Yeah. Right? And and then you'll see how we how we do it.