#174 — AI Killing Engineering Jobs Is “the Biggest Load of BS”
November 06, 2026 · Listen · All episodes · Chat with the show's AI
▶ Subscribe on YouTube Follow on Spotify
For 15 years, entrepreneurs and investors the Morins and the Lessins — Jessica Lessin (founder & CEO of The Information), Dave Morin and Brit Morin (Offline Ventures), and Sam Lessin (Slow Ventures, ex-VP Product at Facebook) — have debated the future of Silicon Valley as the closest of friends. More or Less is that debate, weekly: the tech news that matters and the arguments behind it.
In this episode
- Parker Conrad: AI replacing software engineers is 'biggest load of bullshit'
- Election night coverage via Kalshi and Polymarket, not cable news
- Rippling's AI moat: permissions architecture and spotting problems upstream
- Sam Lessin's competing AI agents monitor his personal notes file
- Random 'lottery' compensation scheme buildable in Rippling AI
- Applied AI companies should earn 4X margins on Anthropic tokens
- California wealth tax trending toward failure; markets shrug off blue wave
"That is the biggest load of bullshit. I remember when all the VCs were telling me a year ago that nine months from now there's gonna be mass unemployment with software engineering, and you look today and salaries are higher, the market's tighter. The value of a really great engineer is even higher today than it was a year ago."
— Parker Conrad
"If Anthropic tokens are not worth 4X in your system than what they're worth inside of Claude, what are you doing?"
— Parker Conrad
"Starting a company is somewhere between completely impossible and very, very, very hard. And between those two extremes, there's this narrow ray of daylight that passes through. That's your opportunity."
— Parker Conrad
"I'm happy for all my agents to look at it, and then they literally on every minute or two check it, try to finish things that I haven't even told them to do, and then just deliver me the results back."
— Sam
Get Sam's weekly letter
One email a week from Sam Lessin — what the group can't stop arguing about, before it's consensus.
Read the full transcript
AI-generated — may contain transcription errors.
Parker Conrad: The most important mode is gonna be which companies can have visibility into the fact that there is, like, work to be done for business. 'Cause, like, right now, there's this huge gap between the capabilities of these systems and actual uptake.
Sam Lessin: I actually give a private URL key to the file to all my agents. I'm just keeping my own personal notes of things to remember, things to do, just my literal notebook for me to clear, but I'm happy for all my agents to look at it, and then they literally on every minute or two, like, check it, try to finish things that I haven't even told them to do, and then just deliver me the results back.
Parker Conrad: Starting a company is somewhere between completely impossible and very, very, very hard. And between those two extremes, there's this, like, narrow ray of daylight that passes through. Companies that succeed, they end up finding ways to do seemingly impossible things at, like, multiple critical steps along the journey.
Guest: More or less. Is it no or is it yes? We'll debate the test that's best. Will we get more or less? Dave and Brit, plus Sam and Jess. Put it all right to the test. More or less.
Jessica: Why, hello, friends, and welcome to a very special episode of More or Less. You're here with the lessons and the one and only Parker Conrad. Parker, welcome.
Parker Conrad: Thanks for having me.
Jessica: So we ditched the Morans. Now, the Morans are very busy, and so we thought, who is gonna have opinions, thoughts, insights to take stock of, uh, this week?
Sam Lessin: Who's gonna, with his big mouth, gonna get himself in trouble?
Parker Conrad: Exactly.
Jessica: Exactly, and we have both said at the same time, no, just kidding, but there was a mutual sense of, "Oh, yeah, Parker."
Sam Lessin: Well, I just... Parker, for your context also, last week we were both out, and the Morans did it, and they picked Bing Gordon, who's amazing, and his new book, and they just talked nice things about philosophy and whatever. So we're the Lesses, so you're gonna have to come up with some, some hilarious stuff to say.
Jessica: Well, we'll do it. We'll get to what's happening this week and what also is happening in Rippling. But let's start. Sam, I haven't seen you 'cause we're on opposite coasts.
Sam Lessin: Hi.
Jessica: Hi. How'd you watch the election last night?
Sam Lessin: I was asleep. I'm un-
Jessica: Yeah
Sam Lessin: ... used to it. But I, I, you know what? I'll tell you my honest answer, is-
Jessica: Yes
Sam Lessin: ... I woke up this morning to figure out what's going on, and you know where I went is Kalshi. 'Cause I was like, all these, like, websites have words and, like, I don't know what's going on, but if you actually... Like, it's very easy. What's gone to 99%, 2%, you know, or 98%, 2% or not, that's kind of all you need to know. So it is classic Hayekian logic, which is just watch these great prediction markets if you want just the news.
Jessica: Parker, what was your go-to source to keep tabs?
Parker Conrad: It was basically the same thing. Like, I was looking at Polymarket, but same idea, you know. Like, the last election, I think I was trying to figure out, like, okay, which, which county do you need to look at the returns and be like, well, if it's, you know, plus three in this county, that means XYZ, and it was very easy to just go to the prediction market sites and see. I, I guess someone's done that work for me.
Sam Lessin: Well, it's just like every newspaper has so many... They, they highlight the opinions of things, and I'm like, "Just tell me what happened in, like, numbers."
Jessica: Well, so I have an even wilder story. So I'm obviously in New York, and I was in a hotel that had cable television. So I was like, oh, I'm actually gonna turn on a CNN, a CBS, Fox News. Guys, it's so boring. I, I can, I'm happy to report we still don't need to watch TV news. Just vamping and vamping and vamping, but it was wild. Honestly, like, why people would stay glued to that, and I know many people do. I'm not that dumb.
Sam Lessin: So I taught our kids at the last presidential election that it's a family tradition to order Domino's Pizza and watch the results come in 'cause, like, Domino's is, like, what you do.
Jessica: Why? I, I was there. I was just confused. Why?
Sam Lessin: 'Cause it's like America. America, we'll eat Domino's Pizza and watch election results. So our children were like, "Can we eat Domino's Pizza? It's an election night." I was like, "No, we have much better food to eat." Um-
Jessica: They must have been so devastated, Sam.
Sam Lessin: That sounds pretty good.
Jessica: You did make that promise. Okay. So when we woke up or we saw these elections, we had a sweeping blue wave, as we like to call it. And I should say, so Monday night, I'm here in New York. I did as one does. I went to dinner with hedge fund and debt investors, really scintillating stuff.
Sam Lessin: Probably was.
Jessica: Well, it was great, actually. One of them came into the room. He's like, "Have you seen the Kalshi on Texas? Have you seen the Kalshi on Texas?" Right? And this is Monday night, was only ahead by, like, 60%. So, you know, oh my God, if Texas goes blue, the stock market's gonna crash. It's gonna crash. So I had this framework, and of course, you know, in my edit meetings in the ensuing days, I'm like, "Guys, we gotta watch Texas." You know, we're, like, all prepped for this. I wake up this morning, we're taping on We- Wednesday, stock market's up. You know, like, the hyperscalers are up. And actually, I think, and, and I wanna know what you guys think, but I actually think what's playing out in the markets the next day is, like, a great sign for the overall technology ecosystem because it's sort of nuanced. It's, like, sort of rebutting this idea that the whole AI trade, the whole economy, so much of the markets are tech now is gonna be, like, wiped out if, like, data centers, you know, don't grow. But Sam, what do you think? Yeah.
Sam Lessin: Well, you don't know this yet, but I did try to put in a sell order on a bunch of stocks that were quite negatively affected by this result at about 2:00 PM yesterday 'cause, like, I don't need these that aren't... And-
Jessica: Okay
Sam Lessin: ... hilariously, JPM didn't call me back before the cutoff to confirm it And they're, like, really apologetic and being like, "We'll trade these after hours for you. We're so sorry." And I was kinda like, "You know what? Don't worry about it. It's not that big a deal. I'll do it in the morning."
Jessica: Ugh, Sam. Is this part 10 of Sam not knowing how to trade?
Sam Lessin: Parts of the trade, because this wave has cost us real money, is what I would basically say, because-
Jessica: Well, so what? It make us money in other ways. But, like- Like, Parker, when you're, you know, as all founders do, refreshing Google Finance every five seconds today, and I, I'm laughing 'cause you probably don't do that, but, like-
Parker Conrad: Yeah
Jessica: ... what is your read on how the markets are digesting this?
Parker Conrad: Man, I mean, I'm, I always tell people I'm the last guy you wanna ask about investing advice, 'cause my, my strategy is buy high and sell low. Um-
Jessica: There are two of you on this podcast, yes, now. It's you and Sam.
Sam Lessin: To be clear, if you got that consistently right, please do call me, because that's a very useful-
Parker Conrad: You know, I, I actually think someone could make a lot of money by just doing the opposite of whatever I do in the markets. I don't know why the markets reacted with, with the way that they did, but it, it seems, like, really positive that, um, as someone who has supported Democrats a lot in the past-
Sam Lessin: Just keep saying that now.
Parker Conrad: Yeah, I'm, now I'm gonna really say it before I try to keep that hidden.
Jessica: We have to talk about the A16Z memes, guys, about the land acknowledgments.
Parker Conrad: Yeah.
Jessica: I love it.
Parker Conrad: Yeah, that, uh, well, but it, it's, it's nice to see that people at least believe that the Democrats are not going to, like, try and crush the technology industry.
Jessica: Or they're voting for them because they believe that they will.
Parker Conrad: But the markets seem to believe that that's not gonna happen. It would be obviously counterproductive.
Sam Lessin: Well, I think that the story I've heard in various Telegram chats of people who claim to be credible, but who knows, is the good news is the Democrats have won by so much that the moderates won't need the extremists to get things done, and therefore, this version of winning for the Democrats is actually kind of net good. Because, like, look, no one likes the crazies on either side, right? But this version of the world where, like, there's enough majority they don't need people to pass things is actually kind of what you want.
Parker Conrad: End the Iran war, and oil prices go down, and tariffs get lowered, it's, like, seems like that's great for, you know, everyone.
Sam Lessin: And just think of all the Mexican electricians who can now come to the United States. Like, data centers galore.
Jessica: I think that's absolutely in the big picture. I also think that existing infrastructure is more valuable. I mean, I don't know the movement this minute, but, like, Amazon was way up. But, uh, my, I just took note, because the, the meta narrative has been, we're propping everything up with this AI bubble. We're propping GDP. We're propping everything else.
Parker Conrad: Well, we are doing that.
Jessica: Okay. But- ... if, if there's a wobble in the system to that narrative, it's not necessarily disaster. Now, I wouldn't wanna be Neocloud that's subscale. I wouldn't wanna be second tier chip company or something like that, right? But so I don't know. I think it's like-
Sam Lessin: Well, like, it's also just hard, 'cause, like, the ne- with the, with the hyperscalers, like, again, we'll see how this all sorts out. There's, like, the Meta thing, which is huge amounts of debt problem. And, like, the reality is, is that what, the next two years, if the Democrats spend it going after everyone in the administration and making no progress, hopefully we get Rubio and not... Like, which seems likely, 'cause seems like Vance is now, I don't know, not that we know anything, but, like, hard. That's potentially what happens. But here's the, the short term interesting thing is, like, you're SpaceX, you're Meta, you're Google, you already have a shit ton of data centers.
Jessica: Yeah.
Sam Lessin: They just got harder to build.
Jessica: Yeah.
Sam Lessin: Like, way harder to build. SpaceX, by the way, is the only one with some story where we're putting them in space, which sounds insane unless you truly can't put them in the US. It's very unclear what, like, the net is for the hyperscalers, where to your point, a bunch of the stocks I tried to sell yesterday at 2:00 PM that failed to sell were, like, things that are, like, downmarket from there, where, like, there's a compelling story, but only if they can build a lot.
Jessica: The other big election result is that Californ- the wealth tax did not pass, that was on the ballot. You've been a vocal opponent of this. Do you feel a sigh of relief, or do you feel like the next one's around the corner?
Parker Conrad: I, I think it hasn't been called yet, and so I think it-
Jessica: Oh, okay
Parker Conrad: ... it's trending in a particular direction. But so, you know, knock on wood, hopefully.
Sam Lessin: But as I said to Parker, it's trading on Polymarket, I'm sorry, on Calshi, at 2% likelihood of passing. So at this point, it's so low that if you really worry, you should just buy a bunch of yeses, and it'll make up for whatever you have to pay.
Parker Conrad: Yeah, yeah. That seems like a good strategy. Um- I, you know, look, I think that this not passing, obviously I'm mindful of the fact that no one wants to hear a billionaire, even a fake paper one like myself, complain about, about wealth taxes.
Sam Lessin: Parker, I have to break to you. There's-
Jessica: Our audience is okay with that. You can go.
Sam Lessin: This is probably the only podcast in the world where the audience is actually probably not that.
Parker Conrad: I do think that, like, California, like, the state dodged a real bullet, 'cause I, l- looking at this, I think there, there just wasn't an appreciation by the people that drafted this bill or people that supported it of just how different it was in terms of, like, how hard it would be for someone, particularly the people, like, earlier in their entrepreneurial journey, to stay in the state with a tax like this that... And no one believed that this thing would not be repeated, just there was no credibility-
Jessica: Yeah
Parker Conrad: ... to the idea that it was one time, and I think there was no credibility to the idea that it would only be people who were actually billionaires.
Jessica: Yeah.
Parker Conrad: I think everyone expected it to be repeated and for the, the levels to come down. And it just means that for a lot of founders, like, that raise significant money at significant valuations, sometimes the outside world, I think that looks like real wealth in the sense. But the way these rounds come together, as you guys know, is that the valuation is something that you end up solving for, because you're, you're really negotiating about how much capital are you, is, are you gonna provide the company to grow, and, like, how, what percentage are you gonna get in exchange for that. And sort of, like, the valuation sort of falls out of that, but it's not... You can't take that and, like- Use that to pay your bills, right? It's not... The stock certificates are still just paper for a long time, and so it creates a really, a really big problem for the exact types of businesses that you wanna retain, which are the businesses that are growing very quickly, that should someday lead to a lot of tax revenues for the state of California, that have historically been located here. It just makes it impossible for them to do that in California.
Sam Lessin: Yeah. And for what it's worth, Parker, I mean, this is only more true in this era of cuck financing, right? Which we had, we had a Slow Ventures announcement on this today.
Jessica: Sam, are you gonna explain that? Are you just gonna leave that hanging? Like, what are we doing here?
Parker Conrad: I wanna hear what that is.
Sam Lessin: Uh, cuck financing is what's happening now all across Silicon Valley. This is the new, the new hotness, which is, it always kinda happened, but now it really happens, where for hot companies, if you're the lead investor, you're like, "I'll give you money at a 20 post. I will also invest $12 at a 600 post," and then everyone else pays the $600-
Parker Conrad: Yeah
Sam Lessin: ... 600 post price.
Jessica: Yes.
Sam Lessin: So this is, so you are the cuck-
Parker Conrad: Cuck financing
Sam Lessin: ... to do the high price. It's happening all over the place. And in fact, Slow Ventures, I had an announcement today with Kevin, very exciting, which is, going forward, for any company that wants, we are willing to invest .0001% of whatever we put in at a $1 billion valuation for any company. So if you are a seed company and you would like to set up a cuck financing, call us first.
Parker Conrad: Like, you guys will do it. That's great.
Sam Lessin: Yeah.
Parker Conrad: That's, that's a good thing to do.
Sam Lessin: We'll, we're happy to cuck other people.
Jessica: So Sam, I think there's like, there's a undercurrent to your kind of question, which is sort of that, like, things have gotten wild. It's fun to be in New York and talk to, like, the money people out here. I would say the mood here is very much AI financing is gonna dry up. There's more and more skepticism that the $8 trillion in CapEx that is penciled out, if you add it up, will be justified by the AI revenues. And whether it's the debt side or the equity side, no one's saying they're necessarily gonna stop, although some people are saying, like, sitting on their hands. But just that financing is gonna get tougher and tougher.
Sam Lessin: I was in New York last week meeting with an event diagram over-to-slapping set of people in big boy finance land, and-
Jessica: Big boy finance
Sam Lessin: ... they're all like, "Look, here's the deal."
Jessica: Big person finance land, yes.
Sam Lessin: Big boy finance.
Jessica: You met with some women. Okay, moving on.
Sam Lessin: Oh, sorry, I forgot the Democrats wing. Big they finance.
Jessica: Oh, my God. Parker, this is what I put up with. Okay, continue.
Sam Lessin: I was spoken to many theys at big finance, and like, their basic thing is like, "Yeah, yeah, you wanna understand what's going on?" It's like literally all the hyperscalers are just, there's only so much money, like we're out of money, and so everyone's racing to get the money now because it's gone or going away. Everyone's already loaded up with what they're gonna load up on. And so yes, the spigot, it doesn't even matter, like one, we've just run out of money in the world, right? And then two, it's like there's, this cycle is kind of over at this point. Even the money that's been committed, there's a lot of skepticism that it'll ever be used.
Parker Conrad: It'll be interesting to see if, like, you do get these types of supply constraints, who benefits? I mean, obviously anyone with that existing infrastructure, but then I've got to imagine that, you know, it starts to matter what you're doing with this at the application layer, and that becomes really differentiating in ways that, you know, are less true today.
Sam Lessin: Sure. If you actually... Yeah, 'cause if you ha- It basically goes, it's simple. Price goes up, and if price goes up, you get access. You can only use it, right, or get it, if you have a better use of it. It's not that complicated, right?
Parker Conrad: That's right. It means that like, I mean, one of the things that's always shocked me is how investors sort of have disregarded gross margin on tokens inside of these companies, and-
Jessica: Hmm
Sam Lessin: What is this, what is a margin, Parker? Am I... Is that a...
Parker Conrad: It's, it's a cuck metric. It indicates-
Sam Lessin: Oh, of course.
Parker Conrad: You know, it indicates that no one's actually using your thing if you have good gross margins.
Jessica: Well, I think there's a difference. I mean, this, and look, obviously Anthropic gets to do its IPO against the backdrop of all these questions, and-
Sam Lessin: Are they gonna do it? Like, Jess, what's the deal?
Jessica: Yeah.
Sam Lessin: Because this seems like-
Jessica: November se- 16th, there or abouts. I mean, look, it would be fascinating. Who knows? The S1 did reveal a pretty efficient, fast-growing business. Certainly, they get hit by compute going up, though, so that's on people's minds. But I think they do well. But OpenAI has a very different approach to margins, as in they don't really care at this point. So I think that is gonna have to change pretty quickly, especially if, uh, they wanna get on track. By the way, side note, I broke the news yesterday that OpenAI hired Jill Hazelbaker, who is a very well-respected policy and brand and comms leader coming from Uber, who they had pursued, I, I think I can say for years now. So actually, that's a good win for them in, in getting them to button up their messaging. But on the margin side, it'll be very different. I mean, on the financing side, Parker, so you guys have raised gobs and gobs of money. Anything you wanna share about IPO thoughts?
Parker Conrad: There are no, like, immediate plans that we have. There are still, I, I think, a lot of advantages that it seems like exist in the private markets. You know, you obviously need to be, like, profitable right now in today's world, and so, uh, in order to IPO. And so, like, like right now you kinda look at the public markets and it's sort of this place for, you know, sort of a retirement community for slow growth but profitable businesses.
Jessica: Is that true? Is that what the public markets are? I mean, I'm not an expert, but, like, I mean, are the Bag Eight slow growth profitable business?
Parker Conrad: Well, certainly from a s- from a software perspective. You've got-
Jessica: Yeah
Parker Conrad: ... Palantir that's growing very quickly, and everyone else is growing south of 50%.
Jessica: See, I'm over here in media land, so just slow growth is south of 50%. Okay, good to know. Good to know.
Parker Conrad: Well, I still, I mean, I think that's right. In the public markets-
Jessica: Yeah, yeah
Parker Conrad: ... there's different definitions. So I mean, it, it's something that we look at, and that it's certainly something we spend a lot of time on. It's not something that's within, like, the next 12 months, for example.
Jessica: And you feel good, like- This New York sentiment, again on a Tuesday, could change on a Thursday, that financing is gonna dry up around fueling this huge CapEx. Does that make you nervous, even across private markets? Does it seem like that's talking about targeting a very different type of company?
Parker Conrad: Well, I think, I mean, my experience having done this for, like, now two decades, is that there's always financing available for companies.
Sam Lessin: There's always money in the banana stand.
Parker Conrad: Yeah, exactly. Like, even in the depths of the downturn, great companies can raise. It obviously gets harder. It impacts a lot of stuff. And so I think you just have to show that you have something enduring that's really successful, and then things tend to work out.
Jessica: So I'm curious, like, th- 2026 was supposed to be the year of the SaaSpocalypse, right? The narrative going in was that AI would erase all SaaS, that user interface wouldn't matter, and we're all gonna... You know, Databricks is gonna win, and so on and so forth. And, like, the markets believed that for a bit, then they didn't believe that. You never really believed that, right? So, like, tell our listeners a little bit about from, like, your seat watching that narrative versus, like, watching, actually operating a very large SaaS company. Like, what's been the difference?
Parker Conrad: Well, so I would argue, Rippling, not a software company at all. We're an AI company.
Sam Lessin: Oh, thank God.
Jessica: Wow. I thought you-
Sam Lessin: Of course
Jessica: ... you were making the point I don't wanna make. Okay, fine.
Parker Conrad: You sort of, you really have to be, though, right? And I think m- maybe a- another way of saying this is I actually think that, like, look, going forward, those distinctions, like, don't really matter. The answer is, like, you need to solve problems for customers, and there are a lot of things that you can do with AI. And even if you're an AI company, there are a lot of times where things look like a nail, and you need a hammer, and you need soft-
Sam Lessin: So are you saying that they're just companies, Parker? It's just companies?
Parker Conrad: I think, I think it's just companies and businesses, and AI is a tool, and it's a tool that radically expands the perimeter around, like, what you can do in particular domains for customers. And I think that is, like, the thing that, that certainly companies like Rippling, and I w- I think, like, many software companies could do this, and, but many of them have not. You know, certainly the thing that Rippling focuses on is, like, historically, the, the goal... You know, the goal's always been to solve problems for customers. And historically, though, that was always understood to be circumscribed by the fact that computers had this, like, really limited set of capabilities. And, you know, now they can do a lot more, and it means that what we can take on for customers is so much bigger and deeper than what we used to be able to do.
Jessica: But so can everybody else. What becomes the moat to the broadening, circumscribed amphitheater, so to speak?
Parker Conrad: So I'll give you an example of, like, something that I do in Rippling. So sometimes, you know, a lot of times people ask me, "What is Rippling AI?" And I think a big part of it is we're trying to, like, automate a big chunk of, like, what people in G&A functions spend their time on. And there are lots of, you know, many little examples of this, but one example is we're a reasonably large company, and every month we get a ton of stuff from the government in the mail that, like, notices things we need to fill out and send back, audits, questionnaires. And what I do today is whenever I get one of these things, I take the scan or a picture of the form, I dump it into Rippling AI, Rippling AI takes it and reads the form, figures out what it needs to do, fills it out, because all of the data for this is in Rip- and then sends me a signature link. Like, I just get a click to sign kind of thing to sign the form, and then it faxes it in to the government agency.
Sam Lessin: That's funny, 'cause when I get those, Parker, I just throw them out.
Jessica: Sam throws them out.
Parker Conrad: Yeah. Well-
Jessica: This is not good
Parker Conrad: ... so, so here's why you shouldn't throw them out. A lot of these forms are notices about unemployment insurance audits. By the way, I think we should, our tagline for Rippling AI should be, like, AI for boring. Like, everyone else is building sexy AI, and we're building AI for all the boring stuff, like government notices. But, like, I would say I get about 30 of these a month, and probably five to seven of them are fraudulent.
Sam Lessin: I would think more, 'cause this is, like, the classic, my favorite is I get, I've gotten for years notices about money the government owes me, like missing money, and the problem is most of them are fraudulent, stealing money, and so you just have to throw them all out.
Parker Conrad: Those things have a cost for you guys. Like, they have a cost for the information because if you don't dispute them-
Jessica: The information has a separate pi- Sam has nothing to do
Sam Lessin: No, I don't do the information ones. These are separate. I don't do slow either, just for the, for the auditors here. It's just random shit.
Parker Conrad: When those types of things come in, if they're fraudulent and you don't dispute them, your rates go up. Like, you end up paying more, right?
Jessica: And Rippling can build a better product because you have, like, my corporate address? Like-
Parker Conrad: No, it, you, you have... So these things require all of the payroll data, you know, all of the employment history. Like, you need to attach their, in some cases, you need to attach someone's severance agreement. You need to attach all of these details.
Sam Lessin: So can I just take all these and take pictures and send them to Parker @rippling.com, and they'll be taken care of?
Parker Conrad: You send them to me, I'll hand them for, I'll handle them for you.
Sam Lessin: The, um-
Parker Conrad: In fact, we'll give you, you can forward it to ai@rippling.com, and it will just do it.
Sam Lessin: No, no, but I trust you. I don't trust the AI.
Jessica: Well, but they, I'm told that I think they're the same. I think we're g- I think it's a great strategy.
Sam Lessin: The thing that I think is interesting that I think you're both hitting on is, like, I was, I've kind of been on this hobby horse, that if you wanna understand the agentic future, the best thing to do is read Ari Emanuel's book about actual agenting, right, in Hollywood. Because-
Parker Conrad: Yeah
Sam Lessin: ... the kind of net of it is, to this point of you do everything, is, like, everyone's like, "How is the agentic world gonna work?" I'm like, it's gonna work exactly the way actual agents work, which is you have a few trusted relationships, right? In business wise, when you go to Ari and you're like, "I need this done," it doesn't matter what he does, he will do it, right? Like, it doesn't, like-
Parker Conrad: Yep
Sam Lessin: ... so you basically do everything. The problem, which Jess is pointing out, is that when business shifts from, like, I have a core lane of competency, and I'm the best in the world, and you come to me to do that thing, competency vertical wise, which is historically how it's worked with software, you built like a ver ... To, oh, no, I'm just one of three key relationships that exist for you, and my asset is trust and knowledge of you, and I will do whatever you ask me to do, right, to service you. It does become a very competitive market where like, for instance, like Hollywood agents need a terrible business. You get fired all the time, you get replaced all the time. It's a trust-based thing. There's no competency. Like, the question for me is like, and I guess this is what Jess is hitting on whatever, is like, totally buy, this is the way business shifts, is like it goes from... Like, I'm seeing startups now that they barely have a strategy. What they do is they go get customers first.
Parker Conrad: What can we do? Yeah.
Sam Lessin: I'll do whatever. What do you need, right? Like, and they just build it all, and it's kind of a wildly backward strategy from the history, but it makes sense. The question is how it all sorts out in the limit.
Jessica: So Sam, these are agents that are probing and gathering information on people and asking them what they want and building it and see what sticks?
Sam Lessin: No, just imagine the, the old way to build a company. I know how Parker started. You had conviction about where to be.
Jessica: I call it vision. We need to talk about Parker, the best entrepreneurial sterling.
Sam Lessin: He does. But, like, Parker started beyond having revenge co, which is revenge cos are always the best co. But, like, you started by being like, it's IT plus a bunch of other connectors. Like, there was a fairly specific viewpoint you had on what you did, and then you expanded from there. New founders are going and saying, "Hey, everyone wants a VAI in some form or the thing. Let me be your guy, effectively. I'll do whatever. I'll make it happen. This software's so cheap, I'll just build whatever I need. We'll figure it out later," which is just a very different pattern of company.
Parker Conrad: Yeah. It's interesting. I mean, and the question will be about where the moats are. I have, like, actually, I think a, a somewhat interesting or crazy sort of point of view on, like, moats in this new world, which is, like, the most important moat is gonna be which companies can have visibility into the fact that there is, like, work to be done for business. 'Cause, like, right now, there's this huge gap between the capabilities of these systems and, you know, the actual uptake, because most people, like, don't know what they can hand off to the AI. And so the ... I think the answer there is that you need AI to bring the capability to you, which is, I think, what some of these assistants do so well.
Jessica: So they're problem solvers.
Parker Conrad: Well, it's, they're problem identifiers.
Sam Lessin: Yeah, they find the problem before you know you have it.
Jessica: Yeah. Well, but then they have to fix it, yeah.
Parker Conrad: And so the companies that are gonna win are the companies that can identify the problems first, because you solve the problem before anyone else can get there. Like, if you're waiting for the customer to bring you something, it's too late. Someone else has already solved it. And so I think you can look at a lot of businesses out there and be like, who has, like, the furthest upstream visibility into problems? And that's where I think you see the assistants are like, well, email is a great source of this. You can look at someone's email, and you can be like, "You forgot to respond. You forgot to handle this." But I would actually argue that, like, not to sort of promote my own book, but, like, Rippling is, like, an unusually rich source of this information because so much of the work that needs to happen from a GNA perspective is downstream of things like employee life cycle events. You're like, "Hey, you terminated this guy, and, like, I noticed that when he joined six months ago, you paid him a signing bonus, and his offer letter says that, like, he needs to pay it back if he leaves within a year. But, like, you've probably forgotten to do that. Do you want me to just handle it for you?" And, like, that, tho- all of those things become triggers that can allow us to sort of reach out and be like, "We'll just take this on for you." And you can kind of like, yes, could other people build things that do that, too? Yes. But, like, we'll solve it before they even see it.
Sam Lessin: Parker, it sounds like you have the wrong business model. The business model should be, "Hey, you forgot this guy who owes you a bunch of money. I've already collected it. I'll give you 10% if you want. I'll keep it."
Parker Conrad: There we go. Well, that, to your point about the agent business model, that's the agent business model.
Sam Lessin: I got your royalty check first.
Parker Conrad: That, that's what you get. You get, like, people who are like, "Hey, there's government money that you're owed. We will collect it."
Sam Lessin: Yeah, so you're in that business now. But the, the, the, the funny thing, Parker, to your point, y- I'll tell you a funny personal story which mirrors that, because I agree with you, is I, like a lot of other nerdy Silicon Valley people, have recently just, like, rebuilt my own personal version of Apple Notes. Like, Apple Notes is fine, but there's a few things about it I don't like, including the fact it doesn't continuously screenshot and keep all of my change logs. So, like, I have one a notes file. When I delete it, I don't want you to forget what I deleted, right? So I built my own, effectively, iOS and Mac synced up. But here's the key thing I do with it, is I actually give a private URL key to the file to all my agents. My personal agent has it, Muse has it, GrokBot has it, et cetera. So I have them constantly looking at every update to my personal notes, which is actually the top of funnel source for them solving problems for me, and I let them compete, right? Where I'm like, "I don't care who solves this. You all can see. Forget what I ... I didn't ask you to do anything." I'm just keeping my own personal notes of things to remember, things to do, just my literal notebook for me to clear, but I'm happy for all my agents to look at it, and, and then they literally on every minute or two, like, check it, try to finish things that I haven't even told them to do, and then just deliver me the results back, right? And the ... To your point, though, I've set it up as it's not just for one. I have a whole fleet of different things that have access to the same root, effectively, note file, and they compete through, like, who can do it first, right? Effectively. Or who's got the best idea.
Jessica: This is where, like, getting it right matters. Like Instinct, which I used for a hot second, it's, like, trying to solve problems for me I don't have based on the limited data I've given it. And so it's constantly asking if it wants me to start a group chat with Brit, because I asked one ques- I was trying to get information from Brit's instinct. And I finally like, "Don't ask me if I wanna start a group chat with Brit." So I know it's learning and we're in, but like the bar for not just annoying people is
Parker Conrad: To me, it's like I don't really care. Like, the way I look at it though is I think that's directionally correct, and like these bots that I... Again, like I just said, "Muse, every five minutes, check this URL, and like if you see things you can complete off my task list, just do them and like, like send me an email of what you've done or whatever." The thing is like most of the time it totally fails/gives me useless feedback on it, but it doesn't matter. So there is some dynamic to play out there. But I, I agree, the bots right now are like, they're like shitty assistants that are constantly looking at your task list and being like, "Oh, I can solve it." You're like, "That's stupid. Don't do that."
Jessica: Or it's stuff that you failed at previously, so I have done, and I don't feel like telling you, and so you keep asking to wanna do it.
Parker Conrad: So in, in the business world, I don't think anyone's really figured out, like how are you gonna have this type of assistant architecture in the business world?
Jessica: I thought Slack figured that out. Didn't Salesforce, Slack, Anthropic
Parker Conrad: Well, so that's a great like communication plane, but like-
Jessica: Yeah
Parker Conrad: ... I, I think that, you know, in order to do this successfully, these things have to really understand your company, and they have to really understand permissions. Like who can... You know, different people are allowed to see different things, right? And even just like, hey, I, in my, my boring example of a government notice, that it needs to find someone in the company to sign the notice, it's like, who can it send that notice to? And it's like, well, that notice has a bunch of payroll data on it. It has a bunch of information about whether that person was fired or not, and the amount of severance that they got. You need to make sure that you're sending that to someone for a signature who's allowed to see all that.
Jessica: Totally.
Parker Conrad: And to do that, that's ac- that's actually quite hard to figure out, right? And so the advantage that I think that I will say we have, but by implication, it's like, you know, software companies more broadly, is that all of that permissions architecture already exists inside of Rippling. And, and like, yes, companies, like it's their data, they can get data out, but it's like really hard to export the permissions architecture of the system. And you kind of have to get that right when you wanna have agents that are very like multiplayer in a business context. Because like when they're talking to one person, they're not allowed to know the same stuff as when they're talking to another person. That gets really complicated. The irony is identity and permission management is the entire business model of like most of the world.
Jessica: Yes.
Parker Conrad: Right? Like that's for sure-
Jessica: Yes
Parker Conrad: ... what's going on.
Jessica: Parker, how... Are people willing to pay more for this? So the other question is just like margins, rev. These could be the best features in the world, but if they become commoditized, you don't have a great business. What are you seeing?
Parker Conrad: No, his business is taking the money to get back from the person.
Jessica: Right. Your primary business, not the new one we've invented.
Parker Conrad: He'll do the entire thing for free. He just wants the right to go collect that money. The way that we think about this is there are things that Rippling AI does that you can do in Claude, and we are going to be zero margin on that stuff, and there are things that Rippling AI does that you can only do in Rippling AI. And in those cases, these are things that are gonna save you enormous amounts of time and headache that would be far more costly for you to do yourself manually or through other systems. And in those cases, we're gonna make money and make margin.
Jessica: Are, are you gonna just charge for those specific functions, or you're just gonna have a subscription, like new tier plan? Like, what's the-
Parker Conrad: It's all usage-based, and it's just in terms of like how we think about the usage consumption for something simple. It's like, yeah, yeah, you can do that in other system, versus if we're doing something like filling out a SUI form or helping you recapture a signing bonus that you're owed or, you know, imagine that there are 350 other little things like that that we can do, m- many of them because of like the data in Rippling, the permissions architecture in Rippling, because of the visibility of stuff that we have, and because of the capabilities of the system, our ability to, to move money, to send you a signature request, to file your taxes, to ship you a laptop, to provision access to Slack or Salesforce. These are things that Rippling can do that other systems can't. And in those cases, there's like a unique ability for us to save you a lot of time. And our goal is like, look, if you're an HR person or a finance person or an IT person, can we just eliminate the bottom half of your job? You can move up into the top half and do all of the other stuff. You know, and I think that top half will expand a lot. But the bottom half, there's just an enormous amount of like kinda junk work that people have to do in G&A functions that I think it's now possible to just delete. So Parker, in the spirit of like, look, if you think about we're being honest about business software, there's like basically what you sell is getting promoted, right? People buy things to get them promoted. And on the, the converse to that is you definitely don't not buy things that are gonna get you fired. So like, what's, like, is basically the story, look, you're gonna have 50% more time to figure out how to get promoted is like, if you work with us, like, what is the like 'cause we're taking all the shit work out? Because the junior people are like, "Oh shit, I'm gonna get fired." Well, I think it's, I think it's two things. I think, one, it's like, look, you don't have to hire the next person in that function, like as the company grows. You're gonna go much longer and much further with the team that you have. And the second thing is like, if you are someone who works in HR, o- I mean, one of the things you wanna do is you wanna be spending your time helping the company think about talent, helping the company think about manager effectiveness, helping them think about sort of retaining people and the culture that you're creating. You don't really wanna spend your time filling out government forms, dealing with a bunch of administrative work, adding wage garnishments to payroll for child support. These are all things that are, that like are not sort of at the ti- And you, and they take a lot of time. Or, or frankly, like administering parental leave, you know, dealing with all the paperwork around someone's parental leave.
Jessica: Do you need AI to do that, though? I mean, I feel like some of the stuff Could, has been, or you've been on the path of automating.
Parker Conrad: I think one of the really interesting things for me about AI in software, a lot of, like, when you build software companies, one of the things you, you really wrestle with is the last mile problem with software development, where basically it's really easy to build a software. Like, I had this moment 15 years ago where I talked to someone, like a customer, about, we were building a performance management system, and she was like, "It's so simple what I want. I want this thing," and she laid it out for me exactly what should work. She's like, "I don't understand why no one builds this thing." And then I had conversations with a dozen other people, and they all gave me the same story, and they all wanted something completely different. But it was, each thing was so simple, and so the problem is, is like, how do you solve the problem for not just one customer, but 100,000 different customers? The way that we've solved that historically is you have a zillion configuration knobs in the system around all the different options. And what's great about AI is this possibility for, that software can be written at installation time, where the software that you build for a company is downstream of their particular company context, their policy preferences. And so you build the last mile of it in a way that's, like, just right for them. And I think, like, Sam, a lot of times when you're building your own software, that's what you're looking for, is like-
Sam Lessin: Yeah, 100%. The small differences that I really care about and, like, it's just, at this point, so cheap to build and do it yourself.
Parker Conrad: But the problem that you create is, like, now you've built this big old shotgun for customers to blow their own legs off. There's this multi-year maintenance burden that they don't appreciate. They turn around and they've vibe coded three security holes that are gonna get all of the data exfiltrated. And so I think the right answer here, you don't wanna vibe code authentication and permissions and analytics and approvals and all of the core things. You want just the kind of last mile. You know, you look at something like parental leave or where the long tail of what, how companies do this stuff is so long, and the ability to sort of customize that long tail in ways that's downstream of someone's policy, I think it dramatically expands what businesses can do.
Sam Lessin: Yeah, I mean, like, the, the story that I've talked about with other people, I think is the way I would kinda characterize this is, like, it just turns out that, like, the biggest problem with software is that people are bad at using software as users. What AI is actually good at is AI is excellent at using software. Like, and so if you think about it, it's like AI can write software and AI can use software, and the fact that AI is just very, very good at using software, there's all these capabilities already exist. Most of these things, we're not inventing new capabilities really, but, like, to your point, in your Photoshop, when you're 87 menus deep trying to find the fucking toggle for blah, blah, blah, blah, blah, turns out that, like, humans are very bad at that. It takes a lot of specialization to get good at using it, but AI is excellent at reading all the documentation and just hitting the right button.
Jessica: Are people, I guess your point is, like, people don't have to be good. Are people good at using AI or d- does it not matter?
Parker Conrad: No, that's why AI has to read your notes and also tell you what to do with AI. The goal is the AI comes to you and is like, "There is a problem that you have, and we can solve it." And that problem in a consumer context might be, like, there's a bunch of emails that you missed. In a work context, it's like, "Hey, you know, there's a signing bonus that you forgot to recoup that we can help you with." And then it goes out and does it. And so the companies that are gonna win are the companies that have the customer business context that can see the problem that exists before anyone else and bring the customer the solution as the penultimate step. I've already set this up. I can, all you have to do is click approve, and, like, mischief managed.
Sam Lessin: Why do you have to click approve?
Parker Conrad: If you're making changes to payroll, the customer can say, "Well, I don't wanna have to click approve going forward." But I think the first time-
Sam Lessin: I mean, but realistically, like, why would you even have, like, y- shouldn't the AI just negotiate with your employees for how much they get paid? I buy everything you're saying and-
Jessica: Sam, have you ever done a comp negotiation? I don't think employees wanna talk to an AI.
Sam Lessin: Oh, I've done plenty of comp negotiations, and I'd actually much prefer them to talk to an AI.
Jessica: No, I know you would, but I don't think they would.
Sam Lessin: Here's an interesting thing. If you're in a small startup, being like, "Look, you should talk to the AI for a comp negotiation."
Jessica: There are laws against this, right, Parker? Like, there are laws against determining comp based on AI, right?
Parker Conrad: Why? I, like, hiring decisions, there are states where you have to have human in the loop.
Jessica: Yeah.
Parker Conrad: I'm not a lawyer, so I don't know about the comp conversation.
Sam Lessin: But you need to be like, "Look, I don't know your comp, but we talked to the AI." That's kind of the future, I think. Most people, a lot of people are very bad at negotiating. It'd be much better to do it in AI.
Jessica: I had a problem, when you were describing surfacing problems for you, I had the mistake this week of uploading the seven-year-old's picture to the fourth graders' school class place mat. So the face of our seven-year-old-
Sam Lessin: That's awesome
Jessica: ... will be, but they caught it. But that's the kind of thing where it could have used AI to be like, "Wrong child, wrong grade." Like, that would've been helpful.
Sam Lessin: I don't know. I'm not sure where any of this lands us. Like, I'm sure Parker will build, like, you'll build a great b- you have built a great business. And the interesting, again, tell me if I've got the history wrong on this, but, like, didn't you guys originally start by raising a shit ton of money and, like, disappearing for two years to build a bunch of connectors?
Parker Conrad: There's like, uh, it was a shit ton of money at the time. I think now it is an infinitesimal amount of money.
Sam Lessin: Fine, but is that directionally correct?
Parker Conrad: We raised a seed round that was about $17 million, and then we spent, like, two years to try and get-
Sam Lessin: But this is, uh, interesting. In 2026, how many days would you need to build the same connector suite?
Parker Conrad: It's inter- I mean, it wasn't just connectors. It was, like, payroll and device management software and-
Sam Lessin: Okay, fine, all of it. I mean, the question is, like, you basically had to spend $17 million, whatever it is, to get off the ground, and two years of your life-
Parker Conrad: Yeah
Sam Lessin: ... in revenge mode to get back to the starting line, right?
Parker Conrad: Yeah.
Sam Lessin: And then, like, build this thing. And it's just an interesting, you think about, like, how the AI world has changed. It's, not today's version, but that version. Is it a day, a week, a month, a year to get back there? It's not a year.
Parker Conrad: I think, 'cause keep in mind, Rippling's business operates with the permission of regulator. You need more than just software. You have all these places where we need to interface with the real world, and so it's not just about the code that you write And then I think also, like, what we have found is there are parts of our software stack where we can move very quickly from a code generation perspective, but it's the places where it's okay for us to push something to production that we then revert. You know, that we can see there's... There are many parts of Rippling where you can't do that, and the reason is, and it's p- in particular the places that connect to other systems. Because if you have a bug, and, like, money has moved as a result of that bug, you can't get the money back in all cases. If you've, if you've changed something about someone's insurance enrollment, it's not always possible to, like, undo it. And so, the, the companies that can move where you can just really let AI run wild are the companies where it's everything is reversible. There was a big, long Twitter thread that some guy who runs a game thing, like a world thing, posted that went super viral last week, and his point was the entire workflow of software development in AI is wrong, with, like, checking in commits and da, da, da. What you wanna do is just constantly push to production, constantly, and fix on the fly. That's great for the gaming world, where things go to production and they don't work, you reverse them. But in, the more that the business interfaces with the real world, the more things become not reversible, and the more careful you need to be in those parts of the system. Yeah. I mean, it goes back to, like, when I worked at Meta a bajillion years ago, it was really fun. Part of the reason it was really fun, we used to talk about, is, like, we're not working on nuclear missiles. And because you're not working on nuclear missiles, right, you have the license to experiment and move super fast, which made the job really fun. Whereas, like, you're like, oh, if you're working on missile systems, you have to be so careful that this is, like, a sad job, right? Like, you have to go slowly, you can't experiment, like, mistakes are too expensive. So the most fun jobs in the world were the jobs were, like, sort of highly iterative, experimental, and quick. What's interesting is that extreme is obviously gonna be taken over by pure agents. They can do the same game much faster. And then, like, interesting thing is, like, what the businesses you're in, which is effectively more towards nuclear missile end of the spectrum, like, you can't fuck up, or, like, parts of it you can't fuck up. Again, what you're actually paying for and doing is trust. It's an agenting job. It's not the software. The software is, like, whatever, right? It's literally just, you pay Parker because Parker has a real balance sheet. He's not gonna fuck up 'cause it's too expensive for him to fuck up. If you sue him as a real company, you can actually make mon- Like, there's all these, like, reasons that you trust platforms. Please, please don't. No, you... But people will. I'm sure they do all the time. That's all the email you're g- That's all the forms you're getting.
Jessica: How's the Deel lawsuit going, speaking of lawsuits?
Parker Conrad: Uh, so I mean, there's a, you know, I... The approved statement is that, you know, we feel very comfortable about sort of the lawsuit and ultimately prevailing. Something about justice. You know, there's, I mean, I think, as you guys, I think you guys reported on today, recently got a subpoena from the federal prosecutors in ND Cal asking us for everything that Deel has produced to us in the civil litigation.
Jessica: Interesting.
Parker Conrad: And so there, you know, there's some active process going on there that w- we don't control, and w- wheels of justice grind slowly on this stuff, but they are grinding. And the crazy thing that I was shocked by is that Deel's opposing the subpoena, and so they're fighting the federal criminal subpoena or, or the federal grand jury subpoena, you know, asking us to give the government the material that Deel has given to us and say, "No, no, no. We don't-"
Jessica: Does it feel like you're in a spy movie?
Parker Conrad: It's this sort of small part of my job buzzing off, you know, over here that I kind of check in on every once in a while, where there's crazy stuff that's happening. You know, and, and that's just, that's not normal, as I understand it. Like, companies are usually falling over themselves to cooperate with government on, on these types of things. So Parker, here's a question for you. As an outsider, your job does not seem that fun to me. What is the most fun part of your job? It's like all the things that I like least in the world is what you do. Like, what are the things you're having fun with? And don't say it's supporting the customer or something. You're really screwing me on my re- re- my recruiting goals here, Sam. I'm not saying... To be clear, I'm not saying that-
Jessica: Um, your company is that
Parker Conrad: ... I mean, I've gotta convince people on this. I'm saying unless, unless you're trying to replace yourself personally with a different CEO, whoever's taking Parker's job-
Jessica: Sam thinks CEO jobs stink. Sam is on record about... No, he's like, "Who the hell would wanna be a CEO?"
Parker Conrad: I do think being a CEO is, like, there's a lot of things, you know, boo-hoo, but there's a lot of things that are very hard about it, is I think that, like, in a lot of ways, the goal at Rippling has always been to just delete all the boring stuff on everyone's plate. And so one way of looking at this is, like, you are just deleting all of the boring, and you're working on all the boring stuff. But, like, I have always taken this sick, twisted pleasure in trying to do that, and the thing that's really interesting about AI is, like, how much more boring stuff you can delete in a system like this, and, like, that's what's exciting to me about it. And I think, you know, Sam, you mentioned, like, oh, well, you know, if you're working on the nuclear missile stuff, you can't actually, you know, have a lot of fun because, like, you know, the consequences are so high. I think you're wrong about that. I think it has to work in a different way, but, like, you know, one of the things that we're doing is, i- interestingly, like, when you think about payroll systems and you think about tax engine, the underlying things that compute taxes and, you know, take... 'Cause payroll, you know, payroll systems, the way to think of them is that they are data pipelines of organizational data that end in money movement and regulatory filings. And so the fidelity and the integrity of those data pipelines needs to be very high.
Jessica: You're not convincing Sam at this point. You're talking about... Keep going, keep going.
Parker Conrad: All of the rules are out there. And so there is this interesting thing where payroll is one of these domains that is actually externally specified. And so it's very hard. The rules are not always clear, but they are ex- The rules are externally specified, and so if you get them into the system, you can have payroll systems that are not built, but are grown out of that underlying- ... rule set and data set of what the regulations are, and just, you know, there are not a lot of domains where you can do that. I think there is room for some really sexy stuff from, like, you know. So engineers, you should come work on this. Boring problems are the ones that always make money.
Sam Lessin: Wait, you're still hiring engineers? You're not hiring Vibe Code product generalists, like actual engineers? 'Cause if you're hiring those, they need jobs, Parker. So, like, you're gonna, like, hire all of them.
Parker Conrad: That is the biggest load of bullshit. The idea, I remember when all the VCs were telling me a year ago that nine months from now, software, there's gonna be mass unemployment with software engineering, and you look today and, like, salaries are higher, the market's tighter. It's just, like-
Sam Lessin: Mm
Parker Conrad: ... not true. The value of a really great engineer is even higher today than it was a year ago.
Sam Lessin: Well, that's for sure true. I completely agree that great engineers have never been more leveraged. The question is the average engineer, right? And, like-
Parker Conrad: Like most software companies, we only hire the great engineers.
Sam Lessin: Of course, of course.
Jessica: Yes.
Sam Lessin: So what's the most fun payroll product idea? Like, I'll give you a fun one. Is there a world in which you can sign up for a company, and you're like, every pay period, an AI will just decide how much to pay me? You know what I mean? Like, based on how I did that two-week period, like, instead of quarterly bonuses.
Jessica: I'd like to go on record saying this is a bad idea, but yes, go.
Parker Conrad: You can do that today in your Rippling. Like, in Rippling AI, you could go in, you could tell it to build you an automation that runs twice a month, that goes through, like, create a custom skill that, that it just, you know, runs twice a month, that goes through every employee, assigns them a random number, and, like, adjusts a bonus based on... You could do that right now, and you could really, it would be a lot of fun for, for your team. It would just be random payments every month.
Sam Lessin: I've always wanted to compensate people this way, which is, I'm gonna give you a little bit of money every two weeks, and then minimum wage, and then there is a 5% chance you'll get a million dollars, right? Like, random reward. You know, b- like, whatever it is. You just come up with a number. Could I do a random reward compensation scheme?
Parker Conrad: You could set that up using Rippling AI immediately.
Jessica: Okay, I would like to go on record saying that'd be horrible for your employees and their predictability about, because not everyone's a rich startup employee, Sam. Most people want the paycheck reliably.
Parker Conrad: We can do this. We can, it'll go huge on Twitter if the s-
Jessica: Yes, that's right
Parker Conrad: ... the, you know, the team will love it. They'll, they'll really enjoy getting paid minimum wage and with an off chance of getting a million dollars. I think you should do this. I'll help you do it. You can set up in Rippling. It'll take, like, 10 minutes.
Sam Lessin: The way you assign bonuses every year or compensation increases is you change the percent likelihood of the payout rather than the payout amount.
Jessica: As someone who works really hard to reward my team in what reliable ways they can understand, this makes me very nervous, so I'm gonna move on. Okay.
Parker Conrad: Jessica, everyone's playing Polymarket. Like, everyone wants these random rewards like that.
Sam Lessin: People desperately wanna be randomly compensated.
Jessica: Okay, I have been accused of being not Gen Z enough. I may be showing my stripes. Okay, Parker, what's, like, overrated and underrated in Silicon Valley? For our listeners who aren't here, like, what should they tune out that's, like, they hear a lot about, and what are they missing?
Parker Conrad: I think what people should tune out is the sort of day-to-day, like, the AI trade. It's so clear what the direction is on this stuff and how important it's gonna be long term, and the demand is gonna continue to grow, and I'm sure there will be air pockets and w- like, what the right investment level is right now and next year, like, I don't know. But I think there are people that read those stories on the sort of horse trading on these stocks, and they think maybe AI is not real, and that's, that's clearly not the case. That's what they would tune out. I do think that the application layer is, like, tremendously undervalued. I mean, obviously I have, like, a book to talk up there.
Jessica: Fair enough.
Parker Conrad: But, like...
Sam Lessin: You're an AI company.
Jessica: I know, I'm confused.
Sam Lessin: What are you talking about, Parker?
Parker Conrad: We're an applied AI company.
Jessica: Oh, I have a newsletter on that, yep.
Parker Conrad: Yeah, and I think that there's so much, you know, and then, like, maybe there were, there was a brief sort of fight, I, I, don't know fight, but, like, tiff that I got, got into on Twitter where I think that, like, what you wanna be is you wanna be the applied AI company with really great gross margins on token spend. And, like, and the reason is not because, like, you need to prove something to your investors. It's because you need to prove something to yourself, and you wanna be the company that is adding enough value to, for your customers with your AI product that you can charge a multiple on this stuff. If Anthropic tokens are not worth 4X in your system than what they're worth inside of Claude, like, what are you doing?
Jessica: Are you rolling some open source Chinese AI over there? Like, what are you doing?
Parker Conrad: We don't today. I mean, we're using frontier models, like, mostly, almost exclusively OpenAI in Rippling AI, but we experiment with all the open source models and, you know.
Sam Lessin: How many lines of code would you need to change to flip it to Anthropic or to flip it to DeepSeek? Like, two?
Parker Conrad: You get different behaviors in the system. It's not hard, but it's not quite as simple as you're saying because th- these models behave differently, and sudden- and so you look across your eval set, and you switch the model, and suddenly it, it's just behaving differently than it was before, and you need to go tweak a bunch of things and change your guidance and your skills and something about your harness to make it work. And so there, there are some switching barriers, but you're right, you can switch it. You know, it's just, for us, like, the priority is, like, building the ability for the AI to take on the next quantum of work from our customers, the boring stuff, but the stuff that we get sort of, like, weirdly excited.
Jessica: Okay, last question. Parker, did you run the New York Marathon last weekend? 'Cause Sam did, and we have to congratulate him.
Parker Conrad: That's awesome. Congratulations. I have to say, I saw the pictures of Sam running the marathon, and at first, I thought maybe this was, like, a Halloween costume. And then I was like, no, that's right. Sam actually is, like, a big deal runner.
Sam Lessin: Not a big deal runner, but I had fun.
Jessica: It was awesome. Shout out to what? The 60,000 runners. So cool. I had my app, and I was following some friends as blue dots. It was a real win for technology meets real world.
Parker Conrad: What was your time, Sam? I, it was crazy good, wasn't it?
Sam Lessin: No, it was okay. I, I ran 3:09, which I will clarify is not even close to my fastest marathon, but it was really, really fun, and I will- I don't know. Uh, the day, when you finish a marathon, you say, you, in your head, you're like, "I'm never doing that again." One day later, all the good drugs kick in, it's like, and like you're like, "Fuck, let's do it faster." So I'm in that part of the phase, and I'm not sure how it applies to Rippling.
Parker Conrad: That, I mean, 309 sounds pretty amazing.
Jessica: I think Parker's been in the fuck let's do it faster approach to building businesses, too. So I think that-
Sam Lessin: Again, I, we're talking to one, we didn't talk about it on this pod, I think of you as one of the great revenge company entrepreneurs of all time. So congrats on that.
Parker Conrad: Well, thank you. I appreciate that.
Sam Lessin: Yeah.
Jessica: You also, I wanna end with this, and I really hope it was you or I'm gonna have no ending, but you had a quote about being a founder or CEO is something about convincing people to do something right at the edge of the art of the impossible. Is this ringing a bell? I, I remember, and it was probably a throwaway thing in your interview, but it was very motivating to me. It's your job to convince, as a founder, right, to convince people that sort of, like, right at that edge where you can, you've some possibility that the impossible is possible.
Parker Conrad: I borrowed this, actually, from David Wein, but it was very, I remember a very influential conversation I had with him where he was like, "Look." So it's, it's his line, but it was basically starting a company is somewhere between completely impossible and very, very, very hard. And between those two extremes, there's this, like, narrow ray of daylight that passes through. That's your opportunity. And so the implication of that is that it, it is, like, really hard, and you've gotta ask your team to do impossible things. And, like, companies that succeed, you know, they end up finding ways to do seemingly impossible things at, like, multiple critical steps along the journey.
Jessica: Yeah.
Parker Conrad: You know, yeah, it's, it's, it's really hard as a consequence of that, and it's why it's not for everyone. Being a CEO sucks.
Jessica: Yeah.
Sam Lessin: Amen.
Parker Conrad: As the only non-CEO on the call.
Jessica: Okay. Parker, thanks for joining us. You're a busy man. We had a lot of fun chatting with you, and glad that the-
Sam Lessin: He's not busy. The bots are doing it all. He's just here to hang out on podcasts now.
Jessica: The bots are doing it all.
Parker Conrad: That's right. That's all I do.
Jessica: Okay. Well, then we look forward to having you back since you're not busy. And, uh-
Parker Conrad: I'd love that
Jessica: ... dear listeners, thanks for tuning in, and we'll see you back here next week for another episode of More or Less. Bye.
Guest: If you enjoyed this show, please leave us a virtual high five by rating it and reviewing it on Apple Podcast, Spotify, YouTube, or wherever you get your podcasts. Find more information about each episode in the show notes, and follow us on social media by searching for @moreorless, @davemorin, @lesson, @jlesson, and as for me, I'm @brit. See you guys next time.
More or Less Podcast · Episode archive · Spotify · Apple · YouTube