E59: The New Normal with Olivia Kory and Joe Wyer from Haus

Speaker A: I was just going to ask if we successfully got every CSO at house on the podcast. Chief strategy, chief science. I don't know if you guys have any other CSOs hanging out.

Speaker B: CISO Security.

Speaker C: We have a chief security officer.

Speaker B: Yeah, there you go.

Speaker A: Oh, you do?

Speaker B: Yeah, yeah, that's another one.

Speaker A: What are you securing? You're securing geodata. Like, what are you. What are you securing?

Speaker B: Uh, all of our customer sales data.

Speaker A: Yeah, got it, got it, got it.

Speaker B: Pretty, Pretty important to hold securely.

Speaker D: Why.

Speaker A: Why store it? Why store it at all? Why not just use it and get rid of it?

Speaker B: Why not use it and get rid of it? Uh, one, research, two. Like, it takes. You could pull snapshots at any given point in time, right? But then you're going to have, like, let's say the results change. You have to decompose. If you have really big changes in results and the customers are like, whoa, what happened? You have to go back and decompose. Like, hey, was it. Like, was it a change in data version? Was it a change in model version? Was it a change in code version? Uh, of all those things? Also, there's something called data settling, which is a, uh, whole lot of fun to deal with. Uh, and so that's why we pull it over. Uh, we could, you know, we've gotten to points with customers where, like, we have started to talk like, hey, are there different ways you want us to handle your data? And no one really takes us up on it.

Speaker A: Oh, my God. They would much rather you just secure it. Yeah, yeah, they want to keep that CISO employed. I love it.

Speaker D: Wait, wait, how do you abbreviate all of the CSOs then?

Speaker B: Well, I don't. I prefer not having an O. I think a science. Spock is a science officer.

Speaker C: Right?

Speaker B: I'm a chief scientist. I think that's different. So I go, I'm just chief scientist. Uh, Olivia, you got chief strategy officer and he got ciso. Like, we're good. We have enough acronym space. We can figure it out.

Speaker A: They just. They just add a letter. So it's like you start with Joe, two letters, then you add an O, and then you add an I, and then. And then we got it. And now we're winning. Yeah.

Speaker C: Yeah. The C titles are relatively new, though. This is like in the last probably six, six to 12 months. We didn't do C titles before.

Speaker B: We didn't have. We didn't have directors, we didn't have VPs. Um, you know, head up was everywhere.

Speaker A: It's like TikTok, uh, we're going to go backwards now with go to market and member of technical staff. So that's. I'm looking forward to that.

Speaker B: Nice.

Speaker C: I actually hate the term strategy. Like I cannot believe I'm our chief Strategy officer because I think it's meaningless. But I've been.

Speaker B: What do you hate about it?

Speaker C: I've been. Well, I've just been hearing a lot now that like, strategy is going to be the only thing left soon. So maybe I should lean into it, uh, a little bit more than I have in the past. It's just ambiguous. Like I'm. One of my core principles in everything we do is like, call it what it is like. And so I've like, I. Growth is a tough one. Like in terms of head of Growth title, it's just like say the thing that you do. And that's how we name our products too. Besides architect. Architect is the new one that's like kind of, ah, a fun one. But mostly I'd say we just try to like, say exactly what it is that we do, um, all over the company. And that's why I don't like strategy.

Speaker D: Like Fincott.

Speaker E: Yeah. Did you say like firma? Who said that?

Speaker D: Exactly what you think it does.

Speaker A: Jeremiah knows. Jeremiah knows. He's banking threes from the half court line. Yeah.

Speaker C: If you remember, we started talking about whether, uh, you're more productive with kids or less productive with kids. And I was telling you that I wish that the house was 10 years ago and you were like, we have to hit that on the pod.

Speaker A: That is a great place to start.

Speaker D: Yes, I agree.

Speaker A: That is a great place to start.

Speaker C: All right, I can, I can tee up the thought exercise maybe.

Speaker A: Yeah, go for it. Set us up.

Speaker C: Rashab and I were at dinner together in New York last week, which. Great to meet you in person. And I was telling him that I wish that house was either 10 years ago or 10 years from now because like, if I didn't have kids, I have two little kids. I have a four year old and a two year old. If I didn't have little kids, I would like pour myself in. Like I would go deep in flow every weekend. Um, I'd be like all in on the AI stuff, staying current, staying ahead. And I just can't do that. Like even just. Just thinking about this week. It's the last week of school here. My daughter's in preschool. So we had half days yesterday and today we had. My husband was out of town for work and it's just like constant Interruptions, Constant interruptions. And so, uh, I get frustrated and Joe, Joe's probably going to laugh because he's, he would probably say, I just need to delegate more. This, uh, has been his feedback for me always. But I, like, I want to be thinking about this company all day, every day. Like, it's that interesting to me. And I think that I have a unique perspective in terms of being able to, to like, I feel like I know what needs to be done in a lot of areas of company and I just don't have the time to do it.

Speaker E: That strategy. No, I'm kidding. I'm messing around. Sounds like you need a wife. It sounds like you need a wife.

Speaker C: But Rishabh, Rishabh totally disagreed with me.

Speaker A: Yeah. Yeah. I don't, I don't think that's. I, I can. I see. We will never know the incrementality of your ability to understand what needs to happen in the company, because there was no holdout. That's good.

Speaker E: That's really good.

Speaker A: And I think that the, uh, here's, here's what I've observed.

Speaker E: At least.

Speaker A: Um, my, my ref. My, my, my reflection, at least on myself is a lot of the ideas that I have are because I'm doing something. Like my body is doing something else, my mind is doing something else. But in the background, you're still processing, right? Because, like, you can't really get yourself out of it when you're so deep in a company, either as a founder or C. And then whatever your favorite letters are after that, your job is to, like, build a company. That's the job, right? And if that's the job, then it's just like you're, it's always there. I don't, I find it hard to believe that actually, in fact, I would say on days when I'm fully back to back, I have no good ideas. Like, if I observe this morning, uh, it's like 11 o' clock for me now. I have had no good ideas all day. And it's because I've been back to back so far. But this afternoon, similar to you, it's the last week of school. So I'm gonna go to a TK graduation. And I'm gonna promise you during that TK graduation, something is going to bubble up and I'm gonna be like, thank God I'm at this TK graduation.

Speaker C: A work thought. You think you do good work. Thinking when you're with the kids, it's like anything else.

Speaker A: It's like processing. No matter what you do, it's either processing in the background or you can like when you're out doing things, you always have moments in between.

Speaker E: You always have moments in between.

Speaker A: And so I, I just, I just don't, I don't, I don't buy that you're fully disconnected. I think, I mean at least I have found that to be literally impossible to do.

Speaker D: Yeah, yeah, that, but that's a problem too. Cause I have the same problem. Like when I'm with my kids I'm thinking about my work and that's a problem. It's just like I, like I, I feel like they don't get the attention from me that they deserve either. Like, like being fully present with a 1, 3 and 5 year old is very challenging. Um, um, when it's just like this, this stuff that's really interesting to me intellectually is just constantly spinning in my head. So I don't know. I understand what you're saying. Like I uh, think I'm kind of the same way. I find myself if I have a lot of time to focus on something, I find myself like almost doing too much sometimes like going down rabbit holes that really weren't productive. I do agree on that side of things and like getting out of that can be helpful. Uh, and I do find myself working constantly, but I wish I wasn't. And I don't know how to turn that off.

Speaker B: I mean what Rishabh is talking about is I think is a really common pattern with like creatives and like from like my academic life. Researchers, engineers. There's like that unconscious back of the head work that's happening right when you take breaks. But you don't, you, it's, you don't have to be like still actively thinking about it. Like those things will just pop up. But like because I run, I run in the same thing. Like actually like me and my wife. There's a whole thing about how much I look at my phone when I'm with my 10 month old.

Speaker E: Uh, uh.

Speaker B: And so like putting the phone away, being really present with him, I still like might stand up uh, after playing with him and be like oh, like I, I just thought of something. But yeah, like the, and yes and the like when you do get deep focus time, it's so easy to go down rabbit holes. I find myself really often when I'm about to like go into some deep work time, I'm like, okay, I only have a couple hours of this a week. What can I do? And I like, I, I sit down and I say like okay, like what is what am I going to do for this block that I know will get something really effective out on the back end rather than just. I mean, I will say there is, there's totally value for play. I think that's where like experiencing AI right now. Right. If we're talking about understanding, thinking about problems and solutions with AI, like, play is a really important part of it because we're all just figuring it out. And I've, I've had the biggest breakthroughs and learnings just playing with AI, messing around on my own, like equipment and stuff.

Speaker E: There's gotta be some science to this, right? Like, why is Rishabh's divided attention sometimes more effective than his divided, uh, undivided attention? You have limited resources because you're occupied doing something else and you're not proactively doing. You can imagine, like in the process of play, there's all the second order effects of, oh, I just discovered this in the process of, uh, exploring and that's influencing, uh, decisions. Right? Uh, changes in direction. Um, so yeah, there's gotta be some science to this job.

Speaker A: There is.

Speaker E: You've signed it.

Speaker B: There absolutely is. I don't.

Speaker A: No, no, there is, there is. So there's a bunch of studies. There's a study about like walking while doing work. So when you.

Speaker E: That's what I was thinking when you were explaining it.

Speaker A: Yeah, like when you're walking, you're more, you get more creative ideas. Um, there's like plenty of studies about, uh, intersections of like, different fields. So like, when you're thinking about one thing and then you intersect it with something else, you get more creative ideas. There's now, I mean, like the specific thing of like, has someone conducted a study about, like, spending time with kids having impact? I don't know. But all of the ingredients are supportive in actual science. This is like not. Ah, a. Yeah, this is like not, not, not unstudied. Like, the ingredients are not unstudied.

Speaker E: You know, what's a better time than in a podcast for conjecture? Could it be that salience bubbles up? Because, like, if you're working and you're deep work and you're going into the nitty gritty of like, something that needs to be solved now, or you're pulled in all the directions. You know, Rashab, you're in an office, there's like 50 people that want your attention at any given time. That might be like, you know, a low number compared to the amount of people that actually want your time. Whereas when you're with your kids and you know, it's taking up very little of your mental bandwidth. Maybe what's salient and what's important is allowed to bubble up to the surface. Let's call that science. Yeah, that was very scientific.

Speaker D: But Olivia, is your experience like this? Because part of the problem I have with having little kids is like, let's say it's Friday and there's something I really feel like I need to get done this week. I'm gonna be with my kids all weekend. Uh, I won this thing like live out doing something all of next week and then you've got like this time to do it and you get interrupted, you don't get it done and now it's like, feels like three, four days later you're actually getting this thing accomplished that you wanted to do today, which you could have done if you didn't have children distracting you.

Speaker E: I don't know.

Speaker D: Like I, I experience that kind of stuff a lot. I don't know if that's kind of. If you experience that as well, but totally.

Speaker C: I mean, you and Rashad probably thinks this too is there are people who say like, I'm more productive with kids. Um, I think like I don't subscribe to the people, um, pushing back on the argument that, that uh, that kids are a distraction for founders. Um, but I actually, I just like objectively think I was doing better work five years ago than I am today. Like it was better. Like I know it was better. I was putting more thought into it. I just don't feel like I have time to like get into a deep flow state and like put together a framework for how we might need to approach a problem. Um, so I don't know, I think I might be, I um, might be the outlier here. But Jeremiah, you said something else which is like mom guilt. Like when I'm with my kids, even when I put my phone away, I'm still thinking about work. And like I kind of, uh, I think that, I think that the, the there's probably a more positive way to, to um, to like work it into my process versus feeling guilty about the thoughts that come, you know? Mhm.

Speaker E: Do you guys think that there's been a bit of a collapse in leverage, right, with just how much what individual people can do? Because like before the answer was clear. You would get someone else to, to do these things and you could enjoy your life, right? Enjoying um, spending time with your kids. That is probably if you weren't so much more powerful now with AI, exactly how you'd prefer to spend your Time, always with your kids, kind of within reason. And there wouldn't be a, uh, discontentment about splitting your time between your kids and all of the amazing things that you could be doing to further the business. So, yeah, um, is this a bit of a collapse in leverage for operators? And this kind of reminds me. Oh, I've got a loop back into house here. Um, you know, what you guys do is so technical and really, um, to think that it could be done by just humans without software might, uh, be very ambitious. Like a nice way to think about it, but really it's like the combination of incredible humans and incredible software. Right. If I'm understanding it correctly. Uh, and then what do you do with that information? Who do you get it to? You get it to the wrong person at the organization and it's not going to be used as adaptively as it could be. You get it to the right person or you get it to the machine, the software. Um, and maybe you've channeled that information to where it can be used most adaptively for the business. What I mean by this is that I used to say to people, I really actually don't care about looking at the tracking. I want Meta to know what the tracking, uh, the best information because who cares what I think? Meta is the one that's delivering all of those impressions on behalf of, uh, our businesses. So we need to get that information to Meta. I don't need to see it. Um, so, yeah, what do you guys think about this? There was a few things. Collapse of leverage, um, flow of information to the people that it matters most and then actually just to the places where it matters most. That could be software, machine learning, um, and AI.

Speaker A: Can I say something slightly spicy? Joe made this comment of like Olivia or Olivia said, like Joe says I should delegate more. And I think that the critical problem with AI especially is sometimes you'd rather build an agent than tell someone on the team. And that's just true. Now, like, agents are pretty good. And like, um, I don't know if you listen to like Jason Lemkin at all. I mean he's the most extreme about this, but it's like agents are better than a lot of people, like a lot at particular tasks, especially when well directed. Now the problem is that they need consistent direction from you. They're not a, they're not actually autonomous. They're, they're like, you need to like constantly nudge, update the memory files, update because like, because you're giving it constant context. And I think that that's actually the core problem is that we don't know how to operate. We're all figuring it out in real time. And I'm curious if you would agree with this, but, man, the number of times I think to myself the output of this task is worse than an agent that I could build, and then I just build the agent and prove to myself whether or not that's true is very high now. Very high. And I think that that's actually the re. I mean, the reason to not delegate is because, Jack, the way. The way I think about your leverage question is leverage is no longer human to other humans. It's like there's this piece of technology now that can compete with the human pretty well for a huge amount of tasks. A huge amount, not a small amount.

Speaker B: Well, this is.

Speaker D: This is part of where, like, I. I definitely agree with Olivia on this because I have, let's say, 40 to 50 hours a week of time that's dedicated to my family, which is time where I can't be reliably delegating things. And in a world where anything can be happening at any, like, around the clock, that is a bottleneck. I mean, I'm already a bottleneck half the time, uh, because of the fact that, like, I have so many things that I'm responsible for. But now I. I have, like, these blocks of time that I can't take action within those blocks of time, at least not reliably. Sometimes I can sneak away for 15 minutes, go do something, but it's not, like, a reliable thing. Whereas if I didn't have kids, that block of time is probably like, you know, five to ten hours a week of dedicated, like, hey, wife and I are going to go on this date or do this or do that. Um, and the rest of the time is kind of like, okay, I can just operate fully within that amount of time. So I think that's, like, for me, that's the biggest challenge with having kids. And I think the. I think I'm way better as a operator post kids because I've learned how to use my time more effectively. But I probably will be better in 10 years than I am now, uh, when I don't have as much of that. Just, like, constant demand for my attention, um, um, which I like, by the way. I want. I want that. I want my children to want to be around me, uh, just to be very clear about that, if they ever listen to this 10 years from now. But. But it will be different, and it'll be easier in that context, uh, to just have to not be a bottleneck, dude.

Speaker A: They're smart enough to not listen. They're going to use Gemini to summarize what you said.

Speaker D: Yeah, exactly. They're gonna go searching when they're old enough and be like, what are all the awful things my parents have said about me publicly?

Speaker B: Yeah, I think, I think this, this, this discussion or like the argument over, hey, like, how much to delegate. Now you have these tools that like can do better than maybe a person doing it manually or versus, um, you know, versus those people you could delegate to now have those tools too. Right. And they could be more effective. The thing that I think about when I'm trying to balance this of like, do I just ask Claude or do I ask like one of my team members to go do this is like, what, what is the final output? Do I need, Do I just like, need a report to read? Okay, I can probably get that from Claude, or do I just need to like go and learn some methodology or some. Do some research? Sure, Claude, but if it's, if it's something where I need to go and change the system or I need to go and get adoption, like, hey, like the rest of the team needs to be doing something this way or something like that, like there is more, more swing through that you need and that you, I think like you as an operator or manager, like Rashad, you were just talking about being in back to backs all day. Like there is an extra amount of just people work needed to I think, fully land some things. And judging ahead of time, is this going to take people to fully land or is uh, in like the risk that I see is, oh, hey, I got the agent to like write something, say, this is our new process for this. And then I ship it out and people will look at it, be able to tell like that was very heavily AI written. How real is this? And so if you're trying to drive adoption on your team, you really need to think about, do I need to be having meetings? Do I need to have someone to like finally run this down and make sure I get adoption or not?

Speaker E: This is what I mean by the flow of information and the collapse of leverage though. And I'll bring up the analogy of House again, because if you get that information that you get from House, that's useful for uh, the organization, it's adaptive to informed decision making and you give it to the wrong person. It's not good. Right. It might actually be useless. It might just be expensive.

Speaker D: Right.

Speaker E: Um, whereas if you give it to the right person, it might be Adaptive for the business. It might be actually extremely important. And this is like delegating that work to someone else or doing it yourself. You might be thinking, oh, well, I could get someone else to build the agent for me. But there's a difference with founders. And maybe this is why this tension between personal life and work, you know, what you're working on in your business comes up. It's because I don't know about you guys, but nearly all of the founders that I meet have such an aggressive bias to action and they're so in it. Right? Is their life, right? They are. And, um, this might not just be founders. These are founding members as well. People who have that high, um, agency, and they're just so consumed by what it is that they're building and they really believe in it. But, yeah, I think it's hard sometimes to delegate, especially if you can do so much now by delegating to an agent. Right. But that flow of information piece is interesting to me because I think there's very few people who can actually operate in ambiguous territory where they actually, you know, where we're playing. Right. We don't actually know, um, what, uh, step one, step two, step three are. We're just figuring out as we go along.

Speaker D: Right?

Speaker B: Right. When we were preparing for this, um, Olivia reminded me of something that I said like, uh, a few years ago in our, in our, one of our leadership slack channels, right? This information flow, is it going to get to the right person? Can they even act on it? Like a founder can always do, take whatever action they want, Right. But is VP or head of whatever, whatever the right person to get this information and take the action? Something that, that, like, we thought about and we're really conscious of is that in the world, the state of the world where we're successful, we will have had to, like, create a change in the industry of how people are taking it and making decisions on this information. Because you can give out that information of like, hey, here's recommendations and actions that you can take. And it's the kinds of stuff that people are used to not touching at any sort of high frequency. We even still have conversations where people say, oh, I can't move that line item, or what have you. Uh, and so that that value just dies, right? That opportunity just goes away. Uh, and so really like, and I think this is true for any sort of AI system or any recommendation system we're going to be putting in. We're going to have to think about, like, how organizations are going to have to restructure around it.

Speaker A: Okay. At our last cab, the most intense conversation was around who inside the organization, like of our customers are actually going to make the shift with what is demanded of them. Now that we have AI versus who is not. And that I think is like the critical issue. It is not to say, by the way, I don't have any belief that like um, hey, we're going to have like mass unemployment over time or anything like that. But what I am saying is that there's a set of organizations where with AI like ours, like your guys', where AI adoption is very high, especially in a tech company, it has to be. And so we're going to see it first, which we have with square and so on and so forth, where there's a set of people who understand the new job and there's a set of people who don't understand the new job. And it turns out when the cost of taking an action is very like or doing a task is low because you can build an agent to do it, the expectation on judgment goes up a lot and the expectation of understanding business impact goes up a lot. And not everyone is trained to do that. Right. And so now what's happening is inside of every organization you're over. You have to over lever on those two skills and under lever on the like action oriented skills. Because the action oriented skills in uh,

Speaker E: a, in a uh, in an information

Speaker A: business are done by agents. It's, it's like kind of not that hard. And so how are you guys thinking about not only how you will like continue to build house the organization in that context, I mean I'm happy to share for us. Uh, like I have no problem being transparent. I say the same thing to my team. But like how are you guys thinking about it for house? And then how do you then communicate that to your customer? Because like yeah, it's not the, it's not as simple. Like let's just say you give your customer some sort of action layer, whatever that means, the total amount. I mean by the way, your customers, most of them, they're like hey, I've got to go from like ChatGPT. You're asking them to, you're asking them to leap two steps from I, uh, use ChatGPT to I am using an agentic orchestration system. They're not going to leap two steps, guys. That's just not going to happen. They have to walk those next two steps of like now I know what cowork means. I understand how interacting with agents works. I understand that I need to interact with it every day. I understand that I need to give it context. Then they're going to go to agents and they're going to be like, okay, I understand that there's a thing called memory. I understand like I, I, I am m understanding how these systems work before I get all the way to agentic orchestration. But you're going to show up with this thing that is like three steps ahead of them. Right. So, so you have to tell the executive at your customer, we gotta do some training or we gotta hire some new people. What are you gonna tell them?

Speaker B: Yeah, I would uh, I would say that there was two things that Rashad brought up there. One, like how do we think about internally what we need to change in terms of like what we're hiring for, coaching, looking for and then also how we're working with the customers. I'm going to just, I'm going to follow Olivia on the like how we're thinking about it with interacting with customers. Uh, but we can come back to the internal story if you all want to uh, another pitch. So talking about how we're interfacing working with customers. Ah, I really like how Zach teed this up. He had a recent article on Architect which is right like Rashad, you're talking about jumping straight to AI Orchestrator. Like that's what essentially we'd be doing if we just like handed Architect over to somebody. Where in finance these kinds of systems have existed for a while, like high frequency trading, all of that. People are not too scared of that. Something like, why is that in finance you know how much money you made or lost at the end of uh, any given cycle. In marketing you can do all the work but at the end of the day you are not a hundred percent sure what drove what and what the marginal returns on any given thing are. And so it is ambiguous even not just what bets to place are ambiguous, but then what bets even paid off are still going to be ambiguous. Our job is to try to make it as clear and reliable as possible. But how do we do that? Everything we do needs to be explainable. We need to be able to trace where the recommendations come from. And that's why we're been slowly building up a more platform first approach. There have been people that have been trying to build very machine learning driven like decision multi arm bandit type tools that just didn't really get traction because they were just, they were just starting all the way to at the end problem of like yeah, let's just go. But the signals that they were building on. I mean everybody intuitively knew that they're just, they're noisy, they're biased. It's not capturing everything that we need to. And so that's why House is like, hey, let's start with experimentation. Let's use that to calibrate all of this messy information that we are getting from the platforms from marketing mix models and then build the orchestration on top of that so that we can show our work all the way up that chain.

Speaker A: Right.

Speaker B: And talk about it that way. And I think that's how you drive the adoption for something that's like as ambiguous because like it feels like a black box. And it's a big leap really. It's. There's a human element of bringing someone along the way, like having a transformation agenda whenever you come in with a customer and handholding, um, them through like, hey, this is where this is coming from. Here's where all the information is. And also here's where we're going to show. We talk a lot. I mean, Olivia and I just got out of a conversation just uh, about an hour ago of uh, how are we showing and demonstrating the value of. Once you take the recommendations, how much did you get out of it and

Speaker A: do you think who, who at the customer is most likely to change their way of working first? So it sounds like you have to go to the founder, cfo, exec level. How does it percolate from there?

Speaker C: It. I think that there we've like, like Cody is just such a good example. Like Cody is. Cody from Jones Road Beauty is a design partner, um, on the system, on architect. And it's like the, I think the marketing team, when we, when I've talked to marketers about why not use an agentic system, they start to name all of the things in terms of business context and all the things in their head that they uniquely kind of understand about the business, uh, around like reasons why they need a human in the loop. And I remember being at a, at a summit, um, I was at Daski. Jeremiah was there. I saw him like kind of pushing on this of like, why can't a system know your promotion schedule? Why can't a system know your new product strategy or your new product launches? Why can't a system understand your seasonality? And so I think that what I'm, what I'm sensing is it, it might perhaps need to be more top down because a marketer is feeling like they like, what is the value that I bring if I'm handing over the keys to an agentix system when in reality we know it could just give them more leverage. But I'm feeling like it's not something that the marketing team is bringing back into the organization. It's a push from the organization to get more efficient is like what I'm seeing. And, and it's, it's not even a push. It's like permission um, to do this because it feels so risky for the marketing team. Like there's, there's so much that could go wrong, there's the extra zero. Um, like I don't think they feel the safety to go take risks in this world. And so that's where, it's where I'm seeing like the founder who really wants to move into this is just moving so, so fast. Um, and they're not getting caught up on what can go wrong and what does this mean for me personally.

Speaker E: And I guess this kind of comes back to my last question about where is that information most adaptively used in the organizations that you serve? Right. If you think about a human, what are they doing? Create, read, update, delete, uh, in an ad account maybe or in the marketing operations of the business. But AI can do so much more. We were just talking about it in terms of leverage, how much execution it can have, but just in, in terms of the amount of processing that can happen that is outside of what's possible for a human. I would imagine that the information that you get out of house could be so much more adaptively used by meta.

Speaker C: This is often a lack of clarity at the executive level on what is the goal of the organization. If you think of a head of growth and their title has growth in the title, they might be gold on growing top line revenue. You know, like the, the heads of growth often sign most of our contracts. But it's, it is uh, it is a muscle that requires the founder to understand and be on board with the approach. Uh, in a, in a, in a very big way. It's such a big shift that a marketer is just not going to be able to get it done alone.

Speaker D: Um, so well and if you listen to Taylor Holiday, head of growth as a CEO anyways. So yeah, sure, yeah, that's just heading

Speaker B: at something to pile on. Like our company goes nowhere without marketing leaders who really do care about this and like, and, and adopt and people are differentiating themselves in the labor market by taking this stance on like this is what they really care about. Uh, and so there's absolutely leaders out there that we work with um, that really care and are like focused on getting it right. And they push us too. They, you know, they. They call bs. Um, not that we're giving any bs, but, like, they really push, like, hey, this doesn't make sense for this reason.

Speaker A: Look closer before we lose the topic on the internal operations, like, there's a bunch of the. I think the most common audience for this is, like, a bunch of other software operators. And so I, I am curious, Joe, about the internal operations, like, how you guys are deciding to run the company internally and what are the expectations that you're setting with people? And. Yeah, like, are you. I mean, a different way of saying, like, Olivia has trouble delegating is like, every executive is an IC and the best executives are ICs. And we. We're experiencing that in a very big way. And so I'm curious how you guys are thinking about what it means for how you operate a software company right now.

Speaker B: I think there's, um, one. The, like, at the beginning of the year, what we did was we. We brought on. I. I hear a lot of companies didn't do this. And we did. We. We gave everybody access to everything we said. Here's Claude, here's Cursor, here's Chat, here's Gemini. Go play. Do what you want to do. Um, try out new things. And I think something that was really important in driving adoption at the start of the year, too, was leaders going in and trying things and then showing them to the team. Hey, this is how I'm operating. This is what I do. Really, really showing it. I started taking transcripts for all the meetings that I was having with folks and putting that into my. What used to be called Zettocast. And then it was second brain. Now, I think they're called carpathy graphs or something like that. Now, whatever the buzzword is for building your second brain for your agent to use, showing people that I was doing that, creating artifacts off of that, um, talking to people about skills. And so really tons and tons of experimentation out in the open and showing your team that you're doing that and giving them free rein, don't really give them handcuffs within data safety and security. And something else that we did too, is that, like, at least for my team, specifically acknowledged fairly early on, like, hey, like, the. The expectations or like, the needs for really, really great, like, execution of writing science code and data analysis. Like, Claude can do that for you. What I need is much more judgment. You all are moving up the value chain. And there's two, like, there's two different ways that people, I think, sort of react to, like, the execution becomes really cheap. They just explode out with all the different things that they could do, and they just go do them. And so me and, uh, one of my, uh, directs, we wrote a doc that has a picture of Jeff Goldblum in Jurassic park where he says all you could think about was what you could do, but you didn't stop to think about whether or not you should. And that's. So it's like really going back to the judgment is really, really valuable now. Right. And making that very clear to the team that, hey, the expectations of your role has now shifted. Judgment and how you use these things and what you decide to do with them is going to be how you're going to be evaluated. And I really, I think product, um, management, product strategy, sort of practice and expertise is really helpful for teaching folks how to do that. And so referencing a lot of the canonical product management books. Good strategy, bad strategy, um, Teresa Torres's continuous discovery habits, like pushing those stuff out to folks of like, really teaching them. Like, hey, here's how you think about judging what you should do as opposed to just like picking the thing that you see you could do.

Speaker A: Oh, wow. So you're investing the time to teach people and you're like, telling them where the books are.

Speaker B: Yeah, absolutely. In a former life.

Speaker A: And you have the time to do

Speaker B: that in a former life, I wanted to be a liberal arts college professor. So maybe this is just my bias, but I think that's like, you. So, like, my reaction to, uh, good executives or ICs. Yes, but also we need to be teaching and setting the tone for the whole team of like, hey, like, this is how we have to operate. This is what you need to learn. Like, if the. Because at a certain point you can have the kind of people you want. You need to create the people that you want. At a certain point, a certain point of view, if you're being that transformational.

Speaker A: I think there's. There's. This is like. So this funny phenomenon of like, you see all these graphs of like, the most number of job openings at the same time as the. All of these large layoffs, which is easily reconciled by the following statement, which is, I don't have the time to teach these people, so I'm going to go find the people who are predisposed to doing it this way. M. That's how you easily reconcile those two graphs. Right? M. All I'm asking, uh, and it's genuinely a question. There's two different things you can teach people. One is you can teach people how to learn, which is good strategy, bad strategy. Teach them how to be a product mat. You're teaching them how to learn and you are teaching them by virtue of doing the work yourself and showing them how to use the tools, the actual actions that you are taking. So you're teaching them both the skill and the knowledge and you're choosing to take that, you're choosing to teach both. Which I think is, I just think it's interesting, right? Like you could, you could make a different decision and say, I'm happy to teach the knowledge. I'm not going to teach the skill. I'm going to hire people with the skill.

Speaker B: Mhm.

Speaker C: Joe's a good, Joe's a good people leader.

Speaker B: Thank. Thank you.

Speaker C: Probably be better than me.

Speaker B: The way I see it is that it's, it's much, it's much faster, especially in technical, in technical teams the hiring lag can be very high. Also you want to get people who are mission driven, right? This is a big principle in hiring missionaries, not mercenaries. And so if you have people who are mission driven and it's, it could be much faster to get them plugged into like, hey, this is how to operate than it could be to kick them to the curb and find someone new who's already working that way. Especially when I'll uh, go back to this. If we're in a transformational moment, I, I, I just have a hard time imagining finding people who already operate that way. That also hit the Venn diagram of like technical knowledge that I need. There's a certain extent that we could say like, well, Claude has the technical knowledge, but still the whole pruning of the tree of all the things they should do, that judgment does come in with that technical knowledge and experience.

Speaker A: You're really challenging me, man. You're making me question a lot of the ways I think about things.

Speaker B: Thank you.

Speaker C: Joe. You should, I don't know if this is, you should share what we just launched with DocMaster.

Speaker B: Oh yeah, yeah, absolutely. I mean along with talking about internal usage of AI, um, we just launched, this was a really big effort with like got, got a really big staff, uh, IC engineer, um, our CISO and a bunch of other people involved into essentially setting up the loop of like the software development life cycle. Hey, like cut a ticket with these acceptance criteria and an internal tool called DocMaster is what called DocMaster is watching that ticket queue and picking them up and taping it through the whole process of hey, is this clear enough? Does this meet all the criteria Blah, blah, blah, blah. Taking all the way to. Now there's a nice PR ready to go. Uh, and like, I think it's like an important principle that like I'm coaching the team on and I think everybody needs to build towards is letting it, letting the AI take the first shot at it. Never ever as a human. Write the first draft. That's the hardest part for a human in anything is writing the first draft, taking the first crack at it. Editing is much easier and faster. Let the AI take the first crack at it. The AI is really good at first drafts, especially crappy rough drafts. Uh, AI is very good at that. And so DocMaster is like, is a really nice flow. Just like anything that we think is fairly simple to write out the criteria, especially a lot like front end tweaking, things like that, just write it out. Let AI take the first crack. Uh, and then thinking about, I think the next step is thinking about like one, you know, a lot of basic stuff drive adoption. Like how do we have a feedback loop for that system? So put the feedback loop on top of the loop. Uh, and then how do we create systems like that for non engineering stuff too? There's a pattern that you can think about with a bunch of individual contributors, have their own Claude skills and have all these different things that they've built in their cloud coworks or what have you. Uh, to me I think it's really important to be abstracting away from the foundational models and having an internal system. Um, I know that um, HQ by Indigo I think is really big on this and I really agree with that direction. Creating a system where you can really easily share and have those skills, those processes are like, those are assets for the company. It shouldn't just be living on people's laptops or like individual cloud instances. It uh, should be manageable by team leaders and editable and curated by team leaders. And so how do we create systems or ways that they can access that and curate it that they don't have to go through engineers, let's say. Like that's a big failure mode that I think that it'd be like asking an engineer to like change a cell in an Excel sheet for you. Like we need to get away from that as fast as possible.

Speaker A: It's absurd.

Speaker D: I'm um, the coaching thing is really interesting to me because this is, I think of it similarly in that I want people to like for their benefit and for the benefit of the company. I want people to learn how to adopt and use these tools. I also think that the speed doesn't matter as much as we think it does in a. Like, there's a, uh. There's an aspect of speed matters, for sure, but it's like, it never needs to be as fast as we think it does. Like, the, The. The way that I think things need to move, in my mind, reality is always go slower than that. And I think that's true of markets. I think that's true of, like, customers adopting your products. What. Like, whatever those things are. Like, that is. That is just how, like, it doesn't move as fast as you think it's going to. Um, so that's kind of my. My, uh. The way I think about this. And I'm curious, Joe, like, when you're talking about the coaching, what I want to know is, like, how successful it was. Like, what. What do you. Do you feel like you've gotten people where you want them to go? Do you think there's still a gap? Do you feel like it's like it worked for some people, didn't work for others? Like, I want to know, like. Because I also understand the benefit of, like, you go find somebody who. We've done this, we've hired a few people who really, uh, who are already doing this. And there is a massive difference in the way that they operate and move compared to the person who's learning how to do this. And so. And it's like, it takes a while, I feel like, to get to that point. So anyway, I just. I love the theory. I think it's the right theory from my perspective. But I also have seen the success of bringing in outside people, so I'm curious how it's worked for you all.

Speaker B: Uh, Uh, I, I think it's. It's like, not. Not gonna lie, there are mixed results. There's some people that just, like, jump in, they get it, it clicks. There's some people that are a little bit slower, and then there's. And then there's some people where, you know, you have to repeat yourself. It's just like coaching in any situation. Um, and that becomes like, frankly, it becomes. It starts to become a performance conversation then, right? Um, when you're at, uh. If you're. If you're not responding to feedback and say, you know, the, the. We're talking about the company, the universe, like, things are just in a different place. The role changes. You know, we're. We're not in a situation where you can post a job posting, and then that's gonna be the same thing for 20 years. Like, that hasn't been true for a very long time. And so then it becomes a performance conversation. Something I will say is that I was after, like, demonstrating the behavior a lot with, like, showing how I was using the tools. I came and checked in and started asking folks directly, like, hey, like, what are you doing? Um, with it? And I was pleasantly surprised. And it was with a lot of stuff that, you know, you could think, like, oh, like, they just, like, put it together just in time for the bosses. No, no. These are things that definitely had to have been built well before I asked them. Show me what you've been up to. Uh, so I was pleasantly surprised with, like, how much they were diving in from the skills because they were starting to see the value of themselves. Every good I see, like, they know about their BS work that. That, like, is pretty routine and showing them how to. How to, like, automate a lot of that stuff. I think they jump on it pretty quickly.

Speaker A: I think, Joe, I'm curious if you're experiencing this and Jeremiah, if you're experiencing this also. I just worry those people who do get it and for whom it does click, I don't think that they want to sit next to the person who is not getting it. I don't actually even. This is my. My fundamental problem is it's actually nothing to do with my pa. Look, um, I'm impatient as hell, so, like. But ignoring that, let's pretend I was perfectly patient. I was like, sage like Joe. If I was sage like Joe. The problem is, it doesn't matter what I think. What matters is that there are people on my team yelling at me, saying, like, come on, seriously, like, uh, that, I think, is actually the bigger problem.

Speaker E: This is one of the luxuries of having surplus resources. I totally understand where you're coming from, Rashad, by wanting to hire for these skills, because ideally, you hire someone who teaches you new things, who comes in and brings something that was missing in the organization beforehand.

Speaker D: Right?

Speaker E: You're not pulling them all up and having to deal with, uh, I don't know, a friendlier word than Luddites, but people who don't want to update their processes, and then you have to do performance reviews, and they're like, why are you in this conversation now?

Speaker D: Right?

Speaker E: All these, um, it's expensive. This is really expensive. And it's, uh, a luxury to only very profitable businesses. Right?

Speaker C: But I think this is true always of, like, A players want to be around A players, and right now the A players are using this stuff in. In, um, In. In their work. And so, like, I don't think that that's. I agree with Joe. It becomes a performance conversation in the same way that it becomes a performance conversation with any B player at the company. So you just have to hold. You have to hold a high performance bar like the. It becomes a performance culture, and you need to not shy away from those conversations.

Speaker A: Can I offer two things that. And you tell me if you agree or disagree. I think there are two differences to the previous world of performance expectations. One is that the difference between an A and a B player used to be 30%, and now it's 4x. Especially in, like, development, like, the differences that you observe because of AI, because it's, uh, an execution tool, are magnified a tremendous amount. Right. Like Joe just said, like, people are just producing slop. They used to produce slop anyway, actually. It's just that they produce now a hundred times more of it. And so you're like, oh, my God, that's a lot of slop. Okay, so it's a magnifier. So that's one problem that I'm observing. And the second is that the bar. You used to be able to hold a static bar for at least three years, and you could, like, describe the bar. But the problem is that with the technology changing, the bar is shifting. And so it's like you're constantly being graded on a curve. Right. And, uh, that curve is shifting because the best people are just moving it further and further up. And I think that those two dynamics are not the same as two years ago, or at least my experience of it is not the same as two years ago. I don't know how you would react to that.

Speaker B: So two statements that you're wondering whether or not we agree with. One, that the difference between A and B players has grown. Yeah, Yeah. I think something that, like you said, like, before, it was only 30%. I think before it was also still multiples. Um, especially we talked about, like, dragging down the morale of superstars, Things like that. Um, I think it was still also, like, pretty big. But. Yes, yes, I think it has grown. Um, and then the second part was the expectations are changing quickly. Also agree. I guess what I would say, like, I. I do think that communicating the expectations clearly and coming to a decision of how long you're willing to wait, and it's like, that's important in the context of your business and the organization, the team, your ability to replace the talents. Like, there's many variables here that, like, I. I can't dictate for Anyone? Right. However, something I do strongly believe is that if you. If you create a culture or a, uh, setting where the expectations can change and you'll act on them before essentially, like, talking to people or they have, like, an understanding, it becomes very challenging where people will be like, hey, it's like there's just an axe hanging over my head at any moment. And stress. Like, if we were just talking about the science of creativity and things like that, like, there is a very clear, documented pattern of stress pushes back on creativity.

Speaker E: Totally, totally.

Speaker A: This was excellent in ways that I did not expect. Olivia, you brought your chief scientist, but he's like your chief people officer.

Speaker C: That's how I said he's my Sherpa. Uh, yeah, this is. Joe is my. Joe is my AI Realist. Uh, I think he's very much at the cutting edge of what's possible, but he's very realistic about how we make it work inside of our organization.

Speaker B: Now I'm feeling that. I hope I didn't. I hope I didn't dominate the conversation. If there's anything else that we wanted to talk about, you know, I'm. I'm able to go over our whole

Speaker D: thing is that we just kind of, like, kick it off and see where it goes.

Speaker A: So I also don't think that there's anything more important for operators to hear about than how to manage their organization in this moment of change. So I don't. I can't think of a more important topic.

Speaker E: I'm not sure if you guys noticed, but I think that there was a bit of a through line through the whole conversation about quality of information.

Speaker D: Right.

Speaker E: And that's kind of like what House is about.

Speaker D: Right.

Speaker E: But, uh, you know, I love that

Speaker C: Jack knows what we do. Jack, thanks for knowing what we do.

Speaker E: I got to. I got to spill the beans. My full interaction with you guys. I'm sorry I've never been a customer, but I've opened up this blog so many times, using House to do exactly what House is meant to solve and just saying, hey, Jabby Coffee did this thing, and maybe you should try it too. But really, you need, like, an individual answer that's based on your actual business. But, uh, this article, I can't tell you guys if you look at, like, your Google Analytics for, like, the number of times this blog page has been opened. It's me.

Speaker C: Oh. I've got to tell our marketing team, this is one of my favorites too. I still talk about this one all the time. Uh, it's like the most clear example of the power of incrementality is the Javi coffee case study.

Speaker E: So, yeah, that's a good one.

Speaker C: Awesome, Jack. One day, maybe.

Speaker A: I'm sorry, I actually have to hop.

Speaker C: Okay. Thanks, Rashab.

Speaker E: Thanks so much, guys.