Speed, Standards & Building Moats in the AI Era with Keith Rabois

There's two basic things about AI. It looks like AI, at least from a growth perspective for a startup, has repealed the forces of gravity. Like, you just see startups going from like zero revenue to double digit millions in a year. That's unprecedented in American history.

The Library of Alexandria was an attempt to capture all human knowledge. Fire and war destroyed it. Today we try again. Welcome to the Library of Minds where you'll learn the frameworks and mental models of the world's greatest thinkers. And when our conversation ends, yours begins. Each guest has created a digital mind on Delphi that you can talk to anytime, anywhere. I'm Dara Ledgevardian. Let's dive in, everybody.

I'm super excited because I'm welcoming our guests. One of the most distinctive voices in Silicon Valley, operator, investor Keith Raboy helped build iconic companies from the ground. PayPal, LinkedIn, Square, OpenDoor, and more. And as an investor has backed generational Companies like Affirm, DoorDash, Ramp, Fair, Delphi and more. Keith is Delphi's earliest believer and our leader at openstore. The star of JAR and I met ad and also the most effective consigliere that I know. Consistently only takes that are contrarian, but. Right. Except maybe this stretching one. That's one that I'm still.

Which one?

The stretching one.

Oh, that's actually proven now. We'll send you a link. I actually finally found research. Okay, that one was always like a Keithism. And then I finally found research this year. Yes, Stretching is not good. Stretching is terrible. Don't stretch. It actually came up. I actually was like, I'm trying to recruit this guy to help build a robotics startup. So about one o' clock, meeting, went for a hike with this guy and he's pretty athletic and he's like really about the stretching stuff. So it's very top of mind actually.

Because in the generalist piece he was like, what can train ideas? And you're like, oh, my ideas aren't.

Taken, they're all proven. It's boring. But stretching is still a little controversial, so you don't even buy it. When was the last time you were injured? Two years? Three. Two and a half years ago. Yeah, yeah. When was the last time I was injured? Basically never, because I don't stretch. 2018 is the last time I was injured.

Soccer injury.

Someone broke my ankle playing basketball. And I don't think stretching would have helped with that one.

Okay. All right. I also wanted to share to everybody that you were one of my personally earliest believers because we met on a zoom call while I was still working at Apple and I've known Keith for almost four years, so it's been nice to have you.

You even recruited me a chief of staff.

I did. I recruited as chief of staff. Correct. That seems like it worked out okay. There's so much about you on the public domain and we have your Delphi as well that the team has learned from. So we're just going to dive directly into questions that we have. I think one of the things that Dara actually says to me a lot is you have said that you can decide whether or not you're going to invest in a company when you go to their office and what are the traits of a company's office culture you look for when determining if it will be successful?

So there's easy stuff like you just show up either early or late, show up at 8am or show up at 7. Seven is not even that late and you can just tell how hard the company works. That's easy. Doesn't take a lot of brain power. Just like you look around the office, it's like the old investment banker traders. Traders rather would actually buy photos of office parking lots and use that to make trades in public companies all the time. They've been doing this for decades. So that's easy. A subtler one that actually roll off at Sakata taught me is you walk around an office and you kind of watch are people working or are they kind of fake working? That's a little bit more art. You get a feel for just walking around an office, especially open office with most companies have you get a pretty good feel for are people really dialed in or are they kind of working? And that's a softer signal, but that one's actually more valuable.

Okay, this next one is actually the one that my ghostwriter Jack wrote for you, which was better be good.

Yeah.

You joined a startup with zero business experience and you said you wouldn't have been able to get a job at McKinsey when working@voter.com, payPal or Square. What did you do to inflect your personal growth rate? And more broadly, how have you seen ICs of companies take on more responsibility and become barrels?

That's basically true. I was a litigator, so almost no relatives. Corporate transactional attorneys at least have some business fuel. Sometimes they're negotiating deals and they work with business people. Litigators only deal with broken relationships in the business world. So I didn't really know anything I had taken. I was crazy enough or whatever, insightful enough, were lucky Enough to take corporate finance theory. My first year of law school, and basically everything you need to know about business is baked into corporate finance theory. You can read a book. We have a textbook by Marvel. It's pretty good. It's better if you have a professor, but that really sets a foundation of understanding pretty much everything you need to know about businesses or how they're valued and what people care about. Learn alpha, beta, all that kind of stuff. The second thing was I was very conscious, and this applies to everybody, whether or not you have business experience of. I jumped when I felt I found somebody I could learn from. So the guy who hired me is this guy named Jeff Donaker. Jeff was more like a friend of a friend, but a friend from Stanford that had jumped into the Internet bubble early, did a startup that got acquired, was running a small startup, got acquired for really good price, stayed on this relatively large search engine at the time and learned a lot. And then he was doing a second startup. And so Jeff recruited me, and I sort of believe that he pretty much knew as much as anybody could know about the Internet. Internet was like four years old, and Jeff had been doing it for four years. And smart guy, very logical, pretty German. And so I was like, he can teach me everything I need to know. And that basically was true. So two months working for him, I absorbed how do you negotiate deals? What matters basically, how do you get distribution? What are the terms? What are the moving pieces, levers. He tried to teach me how to do Excel. He ultimately failed at that. There's this podcast where Delian talks about it. He's never seen me open Excel spreadsheet, and I'm hoping that I moves fast enough that I will end my career before ever opening an Excel spreadsheet or Slack. Slack I definitely don't use, but that's like an intentional decision. That's not like I don't want to learn it. It's just like, it's bad for you. It's bad for companies. We talk about why, but I never created a PowerPoint at the time. Now I only use Keynote, but I never created any dax. And Jeff taught me all this stuff in, like, a couple months. I did so well that actually, Jeff's still a good friend. He became the COO of Yelp. Kind of retired, runs a seed fund. But I did so well that they promoted me over Jeff. It was a long, complicated situation. I didn't really want it. It was not great for our friendship in some ways, but it was good. It worked out okay. In any event, we Both did fine, but that was the key criteria. I was like, I need to learn this stuff, I need to learn fast. The other thing I did is I started. I mean I was always a voracious reader, but I sort of tuned what I was reading. I had never read business books at all, business biographies even. I read history, political science fiction, everything, pretty much everything. But. And then I did want to learn. And so I did try to find business books. There's not that many in the genre. There's a big genre, it's one of the best selling, second best selling genre beyond self help, which is relevant to you guys, is actually business books. But they're usually pretty bad. And so trying to find the five or 10 ones to read was an exercise in and of itself. I did find a reasonable number. I found some that I actually found were pretty good, could get up to speed enough. For example, one that was actually pretty good is I found an approachable introduction to accounting. And it's actually moderately well written, moderately engaging and solid enough that I went from literally understanding nothing about accounting to at least being able to follow along. And that's kind of where you get, if you get some following along, then you can learn things yourself over time by osmosis. But if you can't even follow along at all, it's pretty hard to learn even if you're surrounded with smart people. So I just kept reading and definitely kept searching for good, interesting business books. It became a little too transparent actually when I first started doing this, I think maybe Peter at PayPal, hopefully not other people were smart enough to pick up that I was learning some shit from business books and like make fun of me. I got a little smoother on that later.

Well, I love the framework of picking a place because there's someone you can learn from. Because that's one of the reasons why I joined Upstore. I met you and I was like this is someone I can learn from. And an interesting trend that's going on now is that there's all of venture capital is so flush with cash that so many people are opting to start companies and join YC versus do something where they would learn from somebody and we actually lose more hires to people who are starting companies are going to yc. And so what do you think about this trend and the opportunity cost of making that decision?

Well, I think first of all, choosing to join an early stage company and betting on the future of the company is really challenging. If you're a really great VC, maybe kind of hall of fame, you get about 30% of those calls. Right. So if you're an employee, think about which company to join. The chance that you're going to be more successful in that decision than the best VCs ever and beat the odds is pretty ridiculous. But you can evaluate whether you can learn from people in the company or a person in the company pretty accurately. And so if you make that bet, the economic upside of a startup early stage is all upside, because if you learn things and you get good and proficient on things, they come out. This is what happened. So my first startup failed. It's actually a good idea. And actually another mutual friend of ours, Evan, has basically recreated what I was trying to do in 2020 that failed. It's not exactly the same, but it's 80% and he's doing it really well. So it took 25 years, but eventually, like this old idea I worked on in 2000, the state affairs. State Affairs? Yeah. State affairs is literally right down the middle of this company called voter.com, which is my first startup that failed anyway. But when I jumped out of voter.com, i joined PayPal. And PayPal was a mess at the time. Elon had just been fired. Peter Thiel had taken over as interim CEO. Interim CEO explicitly. September 25th. I joined six weeks later. And the only reason I was able to do well there was because I've actually learned all this stuff in my failing startup. So when I joined Day one, I was ready to go. Like, I knew as much as pretty much anybody in the company. Like, the Internet was only five years old and I'd been working on it for one, and. And it had a good immersive learning experience, so I was able to keep up.

Yeah. Okay. And so talking about your time on PayPal, that sort of dovetails well into the next question, which is there's stories of people at PayPal sleeping in the office, pulling on nighters to get shit done. But I know that you care a lot about sleep now. And we also give everyone on the team eight sleeps.

Oh, good. Great. In retrospect, that's why you're doing well.

Yeah, exactly. Uh, we inflected once. We gave our own.

I want to see that graph. I'll probably. I'll tweet it. If you have a good graph that shows H sleep and, you know, AR growth, we actually might, because we treat it.

We rolled it out and then we inflected. In retrospect, was that needed for PayPal survival and. Or would you.

Yeah. Oh, definitely. Like, Max. It was. It's just Top down. Max Levchin would be in the office with the engineering and fraud teams constantly solving very real problems. He's talked about a few of them.

At Coastal Summit, he was talking about the cryptography.

Talked about. Yeah, that's just one. But like there he even did one at a firm. There was one issue at a firm a few years ago that the first couple nights of trying to figure it out, he actually stayed in the office with the team. So leadership by example works. But we had constant fires and threats and problems. I generally got pretty good sleep at PayPal. That doesn't mean I work seven days a week. But I was typically asleep by like call it 11pm Max. And then I would wake up early, check my BlackBerry, literally at the time. And there was definitely days I'd wake up at 6 or 7 and I'd be like, oh shit. Really important things. But typically my kind of stuff, there wasn't that much I could do between 11pm and 7am like, it was mostly dealing with a lot of external parties. They were typically asleep, but I was definitely on call seven days a week. I would be working Saturdays and Sundays. I missed every wedding I got invited to. But my sisters for three years, it was just impossible to take that time off.

I mean, I think I remember you told me that you went 13 years out of a day of vacation.

Yeah, pretty much. If you like what you do, you don't need a vacation. By the way, think about vacation has different connotations, but one of them is escapism. And so why would you want to escape? My general view about vacations was. And I've changed it a little bit, but at the time, my general view of vacations was like, fix your life. You want to be happy and thrilled and challenged every day. Taking a week off and then being happy. Escaping makes no sense if you live the other 51 weeks or the other 30, whatever weeks doing things you don't want to be doing. So like totally the prioritization is completely wrong. The power law is all to every day. You should be thrilled with what you're doing, with the people you're doing it and challenging yourself and building the life you want. Not going to Mexico to escape all the things you don't want to deal with, certainly.

I mean, the one thing that I have wanted to put time against is. What is it called? The place in Arizona?

Exos. Oh, exos. Yeah. So that was actually one of my first vacations. In April 2010, I took a week off to go to this place where people want to Train. Mostly play in the NFL, but train professional sports. Go. It's a combination of training, recovery, and food. They make all your meals for you prescribed certain portions, carbs, protein mixes. Great. I just wanted to see really, what do these people do? It's the first time I really had the opportunity to see what elite people in other fields do and how do they act. Now I do this intentionally. I have the opportunity to meet people who've been successful in other fields, whether it's politics, music, sports, and maybe some other fields too. And look for common denominators, basically. But this is the first time I did the sports one, and it was great. When I did train twice a day, ate food, went to sleep at 8:30, and read a book every day, it was like the healthiest thing ever. What was interesting about that experience was at the time, these are the kind of things you check in on Facebook, your feed. So I checked in and I'm on vacation. Half the feed thought it was the greatest thing ever. Literally. Half my friends were like, oh my God, I'm so jealous. I want to go. The other friends were like, what the fuck's wrong with you? You're taking vacation to work out twice a day. It was like this controversial debate on my feed wall or whatever at the time.

That sounds awesome.

Yeah, it was great. So I've been doing it. I go back pretty much once a year. Not for a week anymore, though. I've never been able to carve out a solid week again in the last 15 years.

Fair well. So one of the things you also just mentioned when talking about PayPal and a firm is constant fires and dealing with bugs and issues, which we've had our fair share of that. And one of the things about Square is that it was a financial product. And so trust with your customers, customers is so critical. And so you also said managing product velocity at Square was an obsession. And so what did that look like tactically? Weekly roadmap reviews, direct calling, firing people quickly. And then especially with Square, because of the trust, how do you balance that velocity with.

We take an engineering approach to it, which is we had what's called money moving code. Kind of like what sounds. It moves money from point A to point B. That had to be zero defect, perfect, perfectly reliable. And every single line of that code was pair programmed. You were not allowed to ship anything with yourself ever. Then there was things that are not money moving code, like think your conversion funnel, like signing up for your account. That was all done by individual coders. And so those teams could Be like this, this, this. You know, look, if your conversion breaks for a second, it's not ideal but like no one's business is going to fail because your conversion funnel is broken. So we just literally divided the world into things that have to be zero defect and point perfect and things that don't. And some things are a little artistic, but that alone. And different teams, different people, different styles can work on different projects. So you get the velocity as fast as you can with the perfection you need to move a lot of money around makes sense.

Okay. And so I think this also talking about how you organize an engineering team leads on to what we were kind of talking about a little bit before this, which is that you tweeted the Primary topic of 506 Board of Meetings in this last week is how to redesign a startup org in the age of AI. And also we've talked about how org design is destiny in the past. And so what were some of the takeaways from this conversation? Biggest mistakes you're seeing people make because things they should be thinking about differently with AI.

Well let's start with the basic. There's two basic things about AI. It looks like AI, at least from a growth perspective for a startup, has repealed the forces of gravity. You just see startups going from zero revenue to double digit millions in a year. That's unprecedented in American history. Or you see companies going from commonly going from 15 million this year with a very viable plan of 50 next year. Basically almost never been done in history. I recently saw some data from A16C because they have a big portfolio and they have this top 25%, so top quartile which isn't even that impressive from launch. One year from what, six months from launch. The median revenue in the top 25% is 1 million ARR. One year out is 5 million. That's only in the top 25%. Those are numbers that would have been like astound like in the Internet days you could have gone public as a public company like some did actually like loud cloud like Yahoo and things like that actually did or broadcast.com or you know, all the like Mark Cuban stuff. So this is just unusual. And then once you see the laws of physics are changing, you definitely want to adapt to that. Imagine literally gravity, we'd all be dead. But imagine there's some law of physics that we all took for granted that just was proven to be false. You would obviously change all your behavior. So that's just one, that's one driver. The second thing is the technology is Rapidly emerging things that are literally impossible today will not be impossible in three months or six months. And that's also unusual. Even in major technology revolutions. They kind of evolve once you have the first step function. It's kind of like if you've studied evolution, you know how there's these kind of species creation moments and it's kind of a step function versus like a linear evolution. Usually you have a technology evolution and then things are kind of slow again and then people keep expanding. The breakthrough AI is constantly rapidly emerging. So if that's true, you have to build an organization that can take advantage of speed. Basically not speed of execution, but speed of adaptability, which is a little bit different. Those two things try have a lot of consequences. So what's most interesting is I think the companies that have to make the biggest adjustments are actually successful scaled companies like call it Ramp. Ramp's the best company on the planet that was built post 2018, but it was built post 2018 doing everything the perfect way. According to how do you build a technology startup the best way possible. They're a perfect illustration of it. Nobody's done it better. But it's a 2018 or 19. It's actually 2019 shipped in 2020 officially company which is old. So they have to think about with these two principles, what do you want to do differently? You've done everything with a plus level. You won four Super Bowls. What do you do next? But they have to start again. The world is changing.

Well, so some might argue and I'm curious to hear your thought is is the law of physics changing temporary Unclear.

I don't know if there's any way not easy to decompose that because there's maybe verticals where it might be more temporary and there's verticals where it doesn't feel as temporary. You can kind of look at like churn data but like maybe unclear. It also it may not matter. It depends how long, right? Like if it changes and it changes to five years, that's kind of irrelevant for people who have to run a company now. If it changes to six months, okay, you'd be better off. Not like you know, they came radical adjustments. But I don't know what the early warning signal would be. I mean I guess it would be if companies stopped doing that. But as far as I can tell, there's more companies now that are on true escape velocity trajectories than there was three months ago. And it feels like it's definitely more than six and nine months ago.

I remember this conversation so distinctively because it was like right before we did our seed and you're like, it's unclear whether AI is going to be sustaining or disruptive yet. And I'm wondering what your updated thoughts are about that.

Well, it's still open at the macro level. Like let's take Google, right? The classic illustration. Is AI going to displace Google or sustain them? Personally, I think more likely disruptive right now. But if you look at their earnings and the data they marshaled yesterday, it doesn't show up quite yet. But on the other hand, ChatGPT is the fastest growing consumer product ever. It's likely to hit a billion users and almost sure they will pass at least Google Search. Not the constellation of all Google products, but pass Google cross search in relatively short order, like definitely by next year. And so somehow or another that has to be disruptive to Google. Maybe it's incremental somehow. But like, I have a hard time believing that ChatGPT being a place where people start things doesn't disrupt Google to some extent. But that wasn't obvious two and a half years ago when we were chatting.

Yeah, totally. I mean, ChatGPT hadn't even met.

It hadn't probably hadn't launched. It hadn't launched. It wasn't any good.

Yeah, yeah, no, definitely. This was. Yeah, this was like two or three months before ChatGPT.

Yeah. So it would have been like August or maybe July or June before ChatGPT.

Well, so talking about iconic consumer companies and how you like product building principles to choose what to build, I would say one of our personal favorite frameworks or anecdotes that you told us is Shoot the Movie. Christopher Nolan doesn't talk to people about what they think of the screenplay. And we also hate AB testing. Yet often having data to drive decision making is also important. But so how have you seen effective companies balance being data driven and tuition led?

It is tricky to do both. The good news is payments companies like Square, you have to do both because as I've talked about other places, you can't be reckless about where the money's going. And so there's great movies like Superman 6 that show you what happens if you miss small fractions of money at scale. So you have to run any company. Square was top down design driven on steroids, but we had to be obsessive on lots of metrics because literally every cent or every fraction of a cent has to be perfect all the time, infallibly. So there's a way to stitch them together. However, the movie Metaphor, I think works also somewhat instructively, which is you produce a movie. Christopher Nolan, he's an artist in some ways. There's a top down vision. He asked us to sell tickets and he does measure himself by first weekend release sales and that's how the industry mostly measures him. So ultimately, if you build a product, you have to have a vision, but you've got to sell the tickets. And if you don't sell the tickets, you can recut the trailer. And sometimes that's the easiest fix, right? What's the messaging, what's the marketing, what's the channel? And they do do that in movies. It's hard to recut the movie. It's not never done, but it's really expensive, really painful. You have the benefit or disadvantage, but you can actually recut the product too. So you can start with the top down vision. If you can't sell tickets, try to identify as fast as possible what the root causes are. Ship fixes to the root cause and you get the iterative cycle. But ultimately you're measured by do you sell tickets? There's no Academy Award you get as a consolation prize if it's artistically great, but you don't sell any tickets.

Definitely. I mean, so another thing that's been a powerful, I mean, even though, you know, say 20 to 100 is easy, I'm sure it's going to be easier than whatever comes after. It's been a stressful year. But stress is good for you.

Yeah, exactly. There we go.

And so one of the things that, you know, inspired from the Sequoia Crucible Moments podcast, we asked all the people in the speaker series is like personal Crucible Moments, what's been something that's been the biggest point of stress in your life where you had to sort of, you had to come to terms with the fact that you had to handle this stress.

Well, being a VC doesn't cause stress, truthfully. It's like, so I've had to kind of find stress elsewhere, like artificially manufacture stress, because I do believe stress is good for you. The we could talk about why VC is that way, but like it really just doesn't drive true challenge and stress. So I think the areas I've done that is I did try to start a company that caused lots of stress. I used to describe, like, fortunately I'm no longer CEO is to say like 99% of all the stress of my life comes from open store and everything else. All my venture stuff, family, everything is like 1% else. So like starting a company is a good way to get stress. I did find like athletic versions of stress. Like I want, well, I just want like the lowest rest of your heart rate. Best two minute recovery, be in better shape than any of my friends, even if they're half my age. Like just find ways to challenge myself in other ways. Teaching Berries classes in some ways was way like, I'm kind of introverted, I don't really like to do this stuff. But like the first one or two or three, after about 10 definitely gets easier and there's less stress. But the first 10, there's a lot of adrenaline and stress before every class. So you can find orthogonal things that create that spike. Maybe I should learn to do something new. There are things like, yeah, there's stuff like that. The problem is you have to also think about how much time commitment to get to the point where you really can challenge yourself. And so the problem with DJing is I think it's thousands of hours to get to a point where you could actually challenge yourself. And I'm not sure I can make that commitment.

Okay, awesome. I want to open it up to questions that people have. Nora. Hey, Keith.

In the early days of the company we would have monthly meetings. And one thing I very much remember telling you and your response, I was like, keith, it was when you were raising the second half of our seed round and I was like, Keith, all these VCs are asking about what our tactical bout is. And your response was, you should never talk to those VCs. You said that the only moat that matters is a user moat which is built by good product. And I think today AI makes it easier than ever to build a product. And so I'm curious, with AI, are there any different vectors of moats that we should be thinking about?

I think there are. I mean, I think that's right for consumer product, like at the end of the day, consumer product that consumers love and use. That alone is a very serious vote. But are there other things? Sure. Proprietary data, like a truly proprietary. Sometimes there is actually proprietary data available for a particular challenge and there may be a scarce number of places to get it. And locking that down can be a significant moat. Maybe not permanent, maybe not ever, but years. So that can work. Secondly, I think there probably are some ways to lock in. This is probably less true for businesses, consumers, but businesses ways to lock in customers. So at least the foundations you're building, like if you have 10 customers this year, they're not going away. So you're just, it's just a question of how much you add in the future versus constantly. And you know there's. You're locking them in because you've trained their entire analyst class on using your product. Do they really want to retrain 50 people? Their analyst class? It's customized to the workflow like so in enterprise, like if you're Goldman Sachs, are you going to buy like are going to buy like a financial services product from a startup you've never heard of? Probably not. So like can ramp sell to Goldman Sachs now? Probably. Maybe we'll see. But like you know, there are things like brand that do matter. Credibility, the way you phrased it. So that that can be pretty important. It's usually hard to do from day one, so it's usually not phase one of a startup building the brand itself. You can leverage other people's brand early. Arguably you guys did that by using like celebrities, well known people. That is actually what you do is you start sometimes with someone else's brand and substitute more into your brand. And how to do that artfully is a skill in and of itself. So I would go through that book like rigorously. There's an extreme blog post if Trent, Trent, you want to send it to him that it takes that argument to its extreme and says there's no like lock in or no like advantage, no durability and advantage in AI. I think what's interesting about this blog and we're going to debate it, it's interesting enough, we're going to debate it at our MD off site this week. There's a lot of really impressive thinking behind it, but it probably is like 70% right and 30% wrong. But it's worth definitely worth reading and kind of grappling with.

Why is Slack bad?

Oh my God. Okay, so I can actually distill this pretty quickly. Slack changes the parody actually turn this into a basketball metaphor. But Slack changes this. The paradigm of communication to my responsibility is to post something and everybody else has to read it. And I feel like I've done my job if I've posted something into Slack. Communication though is about the person receiving in my view, if I say something to you, but it doesn't make any sense. I have not done my job communicating. So I learned this in basketball. Actually we had a coach at Stanford back when I was in college. He's a very good basketball coach. He's in the basketball hall of Fame now and he had this concept he would teach his players, which sounds rudimentary but it's actually really brilliant. It's like your job when you pass, the ball is not ended until the recipient successfully catches the pass. By reframing it that way, all these excuses kick in. For those who play basketball, it's like, oh, well, I threw him this pass, but he fumbled it. Or blah, blah, blah, blah, he should have caught it. He's like, no, you have them all. You have correct possession of it until it's transferred successfully. It's your problem. So I think communication is that if I have a brilliant idea for you guys and if I cannot express it so that you guys actually hear it and grok it, understand it, I have not done anything. I can blame you guys. There's the stupid ass founders. They don't listen. But that is not really me doing my job. My job is to keep trying to figure out how to frame it in a way that works so your eyes light up. And so I think Slack does the opposite. Like, oh, it was there. Okay, I did my job.

So that'd be more meetings, sorry, instead of Slack meetings.

Well, it depends. I mean, sometimes you can text somebody and it's very clear they understand what you're saying, but if it's something that's important, you want to make sure that whoever really should know that information actually both received it and understands it. And Slack inverts like that and says, well, you didn't read my Slack. You didn't read it. And that's just not very good.

I think another thing that I've also heard that I thought was good, that's kind of similar is the barrier is too low because you just fire off a bunch of messages. And if you're doing an email, for example, the social contract is almost like, I'm going to sit down and I'm going to think about this before I send it. Because it'd be weird to send 10 emails.

That's true. And then there's another dynamic that actually pulls. Peter taught me a PayPal a long, long time ago. Yeah, maybe even before 2001. Maybe it was 2000. But right around then was Peter was adamant that every email could only have one chew. You have ccs, but every single email could only be addressed to one person because it's very clear who's supposed to act on it. There's no confusion whatsoever. I, I would admit I don't actually always, I don't often do that, but it's a really good discipline. Slack and creates a dri. It effectively creates a DRI without the formal infrastructure to create a dri.

Cool. Okay. Keith, thank you so much for coming.