And I think one thing that's kind of interesting is if you think about it economically, if investors believe your terminal value, your core business is zero.
What's funny is, um, that's not funny, it's sad, but you have zero to lose in the pivot because your core is not worth it anymore. So why not, right? Unless you believe magically it's going to come back someday soon.
What does hyperrational thinking, you know, sunk cost? It takes a lot of emotion. There's a lot of emotion underneath. Yeah, that's right. So, but I just go back to what I would think a CEO might consider. It's look yourself in the mirror. Is your business coming back? Is it really worth your core business? Is it really worth anything? Is there a logical agentic opportunity next to you? And that's easy because you can sort of see like what markets are taking.
Welcome back to the Get Paid podcast. I am your host Manny Medina, the co-founder and CEO of Paid. This week I sit down with my friend Nick Mehta who spent 13 years building Gainsight into an over billion dollar company. He recently stepped down and now he spends time with VCs and trying to figure out what's next for SaaS, for compute, for robotics, and for health. This one goes deep. We talk about why so many SaaS founders are wasting their lives, and what it takes to actually make a pivot into an agentic company, why durability and growth are way disconnected right now than any time in the history of SaaS. Nick also gets really honest about his personal type of things, including some significant loss he's been navigating. And it makes for one of the most human conversations I've had in this podcast in a very, very long time, probably ever. And honestly, this is one of my favorite episodes. If you're a founder, an operator, or anyone trying to make sense of where SaaS goes from here, this episode is for you. Enjoy my conversation with Nick Mehta. Like and subscribe, and as always, let's get paid. Nick, it's so good to see you. Thanks for coming to the podcast. Every time we have a conversation, it's an amazing conversation.
Oh my god, it is awesome to see you, too. And I think we're like in the Sequoia studio now.
We're in the Sequoia studio.
So, I believe I'm in the chair that like Jack Dorsey did his podcast in. So, yeah. So, it's pretty good. I feel like pretty big.
That's right. That's right. That's right. Well, you have to, you have big shoes to fill.
Exactly. Um, well, I need a big beard. That's the thing. Jack's beards are always pretty amazing.
They are pretty amazing. He's all very calm too, which is not you at all. Exactly. Exactly. All right. So, let's go back to like you're advising VCs and when you look at all this, you know, bright-eyed, bushy-tailed, hot new startups, what are you seeing? What are you not seeing? Like what, what has that taught you about sorting the wheat from the chaff in this new world?
Well, it's interesting because, you know, I'm, give a little color for people. I'm, you know, spending time with a few different VC firms. Eventually want to start something new of my own, but I've learned a lot about just like what's in the process of creating companies now versus let's say when we did 15 years ago. And one thing is it's insanely different, right? Like, like not even a little bit. And you went through fundraising. So, you know, blows my mind how the power is all in the entrepreneur's hands now, right? So, literally raising money is just almost like a bake-off where they bake off the investors if they fit the mold. Yeah. I'll talk about a little bit if you're in the mold or not, right? Um, so the mold is either what you're just getting started and you all came out of, I don't know, Palantir or you all went to MIT together or you all worked at Decagon or whatever and you're starting a company or you're research at OpenAI and those people like they're pre-seed or seed, just the insane competition.
Um, or, um, you are somebody maybe it's seed and your traction's off the charts, right? And that's amazing. Now the challenge is, we'll, we can come back to this later. If you're a series A and you raise money, uh, and you don't have crazy traction. And when I say crazy traction like what we thought of as crazy traction like triple, triple, double, double, all that stuff is not even, doesn't even get you in the game. I know, but I'm not saying any of this is right. I'm just like stating what I, what I feel.
Yeah. It's just what I observe. I think some investors are seeing there's a gap like there's some regular kind of SaaS company, AI companies that will grow more like older SaaS companies but they'll be more durable and maybe they should invest. So there is probably a counter market there but there is definitely this pattern and this kind of factory happening and because the other thing that happens is like, and this will get back to like the challenges from a go-to-market perspective, when a category gets hot it's incredible how all of a sudden there's like four or five companies, six, seven that at the same time get funded, you know, Sequoia does one, Andreessen does one, Benchmark does one and so then the competition is like off the charts in these deals and the competition for talent and then eventually for, you know, business as well. Okay. So, how do you judge? So, like, you know, that you're like, you're in the VC, you see one company, you have to assume that there are seven others that look just the same that are about to get funded. What, what would make you say yes or no to somebody like that?
Well, it's interesting because of course I'm not like an investor, so I'm learning. I'm watching what people are doing. I think one quite fascinating is like how much is based on the team. Um, and a lot of that is like their ability to recruit amazing people. And so much of this is like can you recruit amazing people to build like the differentiation. You know, you think about a lot of these businesses, the beginning, let's say if it's a AI native app of some kind. Um, the beginning is honestly stuff that's some level of a wrapper around OpenAI or Anthropic or, or Gemini or whatever, right? But then the eventual plan is, hey, we're going to build, we're going to train our own models for many of these examples. Yeah. So are you going to be able to hire the people that can do the model training, custom, you know, open source model training, right? And are you going to be able to get those people where they want to work for you? So, so much initially is just like this team, they have a good idea but the idea is probably not sustainable on its own. I'm going to go turn it into something bigger. Obviously the best success story here is Cursor where like, you know, they build this amazing thing but it's still dependent on the models and then they build their own model, right, Composer model. And so that's one thing that I think is, is super interesting is like team. The other one that I find really intellectually fascinating because we always dealt with it and actually I, in Gainsight, I kind of dealt with this like with all of our customers which is the moat, the defensibility, the retention, all that kind of stuff like, and you don't see it short term because all these companies are growing so fast so you can't see in the metrics. But long term the question is okay, all the different sides of competition, you know, I honestly, I think when we were in our era there was probably like, okay, who was our direct competitors and there was also probably like, you know, for me it would be like a Salesforce could be a competitor, right? Now, I mean we all got the question what if Salesforce decides to do X, I do that, whatever. But it was like the Salesforce, it was an amazing company but didn't have the capacity to do all those things. And now the big companies can get in your space, the older big companies, because, because of all, you know, coding. And it's like do the model level companies move up in your space? Yeah. Like let's say, you know, obviously Claude getting into design, right, is a good example, right? Or do the model level companies make it so easy for other competitors to get in your space so even if they don't get in it they like almost like level playing field for everyone else.
Right. And then, and used to be that part of the stickiness was it's hard to leave. Part of that was data and configuration, but now that's actually really easy to move because you just tell Claude, move it, right? Like to go do it.
Yeah. And, and Claude Code or whatever. All right. So the canonical example here is Harvey, right? And you saw the, the news of, about that like vibe coded alternative to Harvey.
No, no, no, no. The other news in which they, they ditched their own models to just jump on Claude.
Exactly. And now they're just like harnesses and Claude's and they're using managed agents in particular which really locks you in. Yeah, right. So is that a thing or is that a thing? Like is that a real model? Like that does have durability over time.
I mean I'm not going to jump into this Harvey versus Lora de, holy crap. People strong, walk away slowly. Walk away slowly. Actually I thought like it's pretty good. Winston from Harvey posted his actual active user stats and they were quite good. Yeah. Um, so it was a pretty good response because I think a lot of people have had this narrative that Harvey's, you know, dead because they're now built on Anthropic models and also Lora is, everything people say is it's grown really fast. Yeah. But, um, I, I do think that, um, you, one of the things that's interesting is you have to run so fast in this new world. Right. Right. And like I think there was, software is kind of good in a way actually, old, old software like on-prem was great. You run fast for a little bit. You build something, everyone's stuck and you don't have to run. You have to run anymore. The way that showed up was companies could just like grind down R&D over time, right? And that drove up margins, which is great, right? And then SaaS like it wasn't as good, but you could kind of gradually reduce R&D and, you know, you look at public SaaS companies, right? A lot of the margin gain they got was cutting down R&D, getting more efficient sales marketing.
Yeah. And now it's kind of like in these SaaS companies today, it's almost like they're building a new product every year and they, they're like starting over every year or every month, right?
And so it's like a perpetual game of R&D plus sales and marketing, you know, G&A will scale down over time. And so it does make you wonder like if you have to have, be on that treadmill forever, what is the steady state margins for? And not talking about gross margin, operating margin for these companies. So the gross margins are probably going to be lower because of tokens.
Yeah. Right. You see it. And then I think their operating margins might also be lower because you have to just keep fighting every single day. This, this mindset of like, you know, people are, they sometimes wear it as a badge of honor. They're like, 'Oh, it's so good. You have to build a great product and every day you do it.' But I'm like, look, look at the Fortune 500. The majority of those companies have built products that you just, or services or whatever that are impossible to defeat and therefore they don't have to be good anymore, right? And we look at them and say we say they suck. And those CEOs are like, I look at my margins and my profit and my private jet or whatever, right? And so I wonder what's gonna happen to these native companies. Is it a perpetual fight forever? Is there a point at which they can let, let go? I don't, I, I can say for sure right now. I think everyone is fighting every day right now.
Like, I mean Harvey, Lora is just a great example and Sierra, Decagon's a great example and there's a million examples, right? Right. And I think that that feeling of it's only getting harder, not easier, is something to like watch out for a little bit.
The durability used to be in the model.
And, and, and by, and durability I'm using as a proxy of like it gets easier over time.
Exactly. That's totally related. So, so, but so it's not the model anymore. Harvey just literally just ditched their own model. Jumped on the new one and saying like, I, you know, we're going to do the work for you. At that point the durability is what is it? Go to market. It's the understanding of the customer. It's like the, the, the, like the harness and the context of, you know, a law firm, like so where, where is it like what can you grab on to or will it just change and it's just staying ahead of it.
I think one part of durability is actually the friction that the customer has to change, right? So putting aside data and configuration because that's easy but like, oh, I trained all my people, they're using it this way, they like it, this other thing is a little bit.
But have you seen the Harvey interface? Dude, it's a Claude, it's just like, but I mean good for them like, you know, it's a, it's a new paradigm anyway.
Definitely. Yeah, they have to because they want to make it look like what people are used to, right? 100%. But I mean, I'll say for my own experience, right? Like, so I use, you know, Claude Code a ton. And what's interesting is, you know, uh, GPT-5 came out. Code, everyone says Codex with GPT-5 is awesome. Some people say it's better for some long-running task. I'm like, no, I don't want to have to set it all up.
Exactly. It's not like, it's not like I'm religiously tied to Claude Code. I really like it, but I'm like I don't want to like spend all my time switching between back and forth.
Is it because you're all in Chromogyny or is it because it's truly hard to switch them back and forth?
I mean, it's not all, it's some are all in Chromogyny for sure. Look at me. I mean, but, um, but some of it is like I have so many projects in Claude Code and I could definitely switch them. I could definitely create some kind of Codex thing to move my stuff over. I'm sure that's super easy, but it's just, and that to learn it and it's going to have slightly different user interface.
Do you build your own agent yet? I built a number of agents to do like, I'm using managed agents from, uh, Anthropic to do a number of like, like daily tasks, calendar things, like that, right? So like, so right before the managed engine came out, like I built my own, right? And like the longest thing that actually took me was setting up the context files and the harnesses for the engine to understand me and like the one thing that was important to me that took me like a day, maybe a day and a half, I'm not going to do that again, you mean? Like now the files are kind of portable and stuff but like it's still a pain. Yeah. So like there, there is some of that.
There's some of that and, you know, I think obviously, if I were Anthropic and OpenAI, you want to build more and more into the harnesses that's differentiated, like, like managed agents, so that you get, you get less likely to switch. So the question is can you do this at the application layer or not? Can you create, uh, tooling, context, um, user interface, learning, whatever that makes it really like a pain to switch from a user perspective. I don't think it's going to be pain to switch technically. That part I think is solved.
But, um, I think one moat is companies feeling like it's pain. Now, now that ties not just to the vendor but the type of company. So like for example developers in general are happy to switch all the time.
Oh dude, like the moment Claude Code dropped, like people dropped Cursor in my shop. Oh my god. Like a, like a, like a bad habit, like it is gone. It's crazy. And obviously Cursor still do great, like, but, but people switch at the developer layer so much in my experience. Um, people also switch at the marketing layer a lot because they're kind of doing, running campaigns and they're just like it's all kind of one to one. One campaign, we're done. Here's another one, we're done. We try something new. MarTech was always pretty churny because of that, right? Um, and then what categories are, are like where the buyer psychologically is like, not actually but can psychologically is old and cognitively like you're saying and they don't want to switch. So for example, I think that like, you know, finance stuff is likely to be more, more moat-y. I mean lawyers clearly.
Lawyers, I think, I mean the only thing is lawyers are also project based so law firms are, so that's the only like end of the, like you could be like I'm going to use this for this matter, I'm going to switch.
You know what I mean? Like, so if you want to come in with a wedge you can say well for these types of cases.
And so I think a little bit is the nature of the buyer, the nature of their work, but like I think a CFO in accounting is going to be like very durable, unfortunately makes it hard for a native startup to come in, right? But, um, those might be more sticky, right? So you look at the AI native accounting solutions. Yeah. Um, you know, Campfire, Ramp, stuff like that. Everything I've heard is like they're not like growing like crazy, like they're not going Cursor level. Yeah. But I would be surprised if they have some stickiness long term because who wants to switch off their accounting?
So is that, is that a rule of thumb then, durability is, durability and pace of growth or initial growth are not related?
There could be like some inverse correlation. That's what I was, right. You know, there could be some, you know, developer stuff is going so fast, right? And it's maybe less durable. I, I don't know. This is, I think that was definitely true in the older era. I think the easy come easy go is real, right? Now, can you, something's obviously that the two fastest growing business in the history of mankind, OpenAI and Anthropic, also probably are going to be really sticky. So, there clearly exceptions to this, right? Yeah, but I think that it's also, it's the combination of the growth versus, um, durability and then also the buyer's psychological reticence to change. That's my hypothesis and therefore if you're building, let's, therefore honestly, like I see, I've seen people look at, uh, development stuff in the agentic world, the whole, the whole platform of agent stuff and with great metrics, people are so reluctant.
Yeah. Because it's like you have this, like, buyer willingness to switch.
Yeah. Yeah, we saw that a lot in open source by the way. So many open source, I mean the one that's very well known, Docker, right? Which, you know, created the whole container concept and it's like that and, you know, that's an amazing founding story, never turned into a gigantic business that all matched its impact because developers are just so mercurial and willing to switch, right? And then, then you have the second order effect which is how likely is the model, uh, the, the model companies to get into that space.
Anything in the agent stack, the, the, the likelihood is super high. Yeah. Yeah. Legal probably low, right? Honestly, like I don't really, maybe eventually massive market. Yeah, it's a big market, but I think that like I'm just saying relative to like development, right? If you're, I think, I mean even for me, I, when I create stuff in Claude Code, I'm like, I'm again annoyed at signing up for third-party services. I know you, it's like, oh, use this thing for email, use this thing for, I'm like no, just please, just do all of it for me. Add it into the managed agent. Give me an email service. Give me a database. Give me a whatever. Right. And, you know, they're kind of creeping up into that area.
Well, they get to see all the requests. They can just prioritize based on like what gets most requests and like, all right. So you're, you're thinking of starting something now.
Oh, yeah. Yeah. Very early, just yet. Very, no idea.
Okay. So right now you're a little boil in the ocean, right?
100%, just meeting people, learning, that's still kind.
What's in that bag of ideas?
Well, I mean I think that there's so many different things that are interesting, right? Yes. But that's what's kind of cool. It is like the most exciting time to be an entrepreneur. Even though it's like kind of crazy and competitive, it's also the world is everyone's oyster, right? 100%. And so I think there's some things that are like have always been interesting purely for an impact perspective, right? Like things like, you know, digital healthcare, right? Which is I think continues to be interesting, hard, complex.
You know, not as hyper-growth, but there's a lot of, you know, everything longevity to different types of specialty care and all that. And I think that like that's an example where you're like, yeah, you look 5, 10 years from now, there's no way healthcare looks like the way it is now. Right. Right. It's got to, it's going to get better. In so much of it is identifying problems, right? And what's crazy, I don't know how much you track this, but the like research on new solutions to medical problems is insane, right? Like the this famous trial on pancreatic cancer, huge remission rate, used to be a death sentence. Yeah, like curing deafness, earlier efforts to cure blindness, early onset dementia, different trials for that. So there's this interesting thing where it's like, okay, you take all these new diagnostic technologies, everything from a Grail to like full body MRI, etc., you combine them with different virtual care and then you get people ready for when those treatments are available. And so I think that's really interesting, you know, like, like the, the GLPs which are certainly going to be like one of the biggest businesses ever are, are only like the tip of the iceberg for like all the amazing advancements. So.
It's funny, I read a sign the other day that everybody in San Francisco is micro-dosing.
Oh my god. Micro-dosing specifically.
Yeah, of course. Yeah, of course. Yeah, totally. Like, and that's like a, that's a good example. And so, I think there's just so much that's going to come in mental, in physical health. I think also, and I have personal connection to this, in the mental health world, so much more that we can be doing. It's complicated, but clearly like this like talk to a therapist every week and nothing in between and have a psychiatrist sometimes only very late intervene. It's not the future, right? And like measuring, I mean, I, I, I have a lot of experience with mental health and like when you go through it there's a questionnaire you fill out and they're ridiculous. Like they were written like in the 1970s. It's like the PHQ-9, PHQ-9, which is a depression screener, and it has questions like, uh, do you have trouble, uh, concentrating while you're reading the newspaper? And I'm like, yes, because I don't have a newspaper. Yes, I have trouble reading.
I haven't seen one in ages.
Exactly. And so I think that like that, that like self, that self-answered, um, question on mental health, which is a lot of the diagnosis, are just so old school. And like there's, you know, people have done research around like listening to your intonation on your phone, uh, with using the microphone to understand how you're doing and is this person activated into a bipolar state or whatever, right? Like so I think what's cool about health and mental health is there's so much research happening and it feels different than before where we had this long period of like not much progress on cancer, not much progress, nothing really on Alzheimer's, etc. And now maybe we're getting more like close to like the changing point. Yeah. Um, and so, you know, that's kind of one cool thing. By the way, this is like in some ways Ray Kurzweil is the, um, the sage of all everything we're talking about, right? He predicted everything, uh, in AI. And one of the things he's talked about is like you just need to live a little bit longer and then the things are going to get way better, right? And so I think there's a lot of healthcare stuff that's really interesting. Um, for me, I'm super fascinated by robotics. I have zero skills in it. So I'm just like, I sound like I haven't started robots. Every, every software entrepreneur that I know is fascinated by stuff that is not digital or like the.
Well, that goes back to what we were talking about, like there's the moat, ultimately software, the cost of building software is going to trend to zero, right? So the moat, one of them is having physical stuff. So, you know, I think that's, there's no, no surprise that investors gravitated towards, you know, robotics, defense, space, right? Um, semiconductors.
And so, um, so I, I just found robots as a nerdy kid like always super interesting. And so I think it's cool just to see the amount of innovation happening there. It's interesting there though because you get into these other areas and something you, I don't know, we probably never dealt with is seeing these companies that have legit science risk, like legit, like it's not possible risk, not like, like I don't know.
We haven't seen that in, in ages. No, I mean like venture capital.
I don't, but Gainsight, we didn't do anything like novel that was worth a research paper, right? So there was nothing that like what if we failed, it was our execution. It was not that or there was a market risk like people didn't want it but it was never, oh, that's not technically possible.
And it's like, oh, it's kind of interesting to see these like new businesses that are like at the edge of what's possible in science. Yeah. You know, which is actually honestly for investors much harder to read then, right? Because they're in, they're making reads on things that are like much more difficult, but there's a lot of upside, you know, and so I think it's just super cool that so many smart people are going after robotics now. And I think similarly space, defense, I'm just not as close to those, you know. So I think that's all interesting. I think that in the, um, in the, um, kind of AI native software stack, um, what's interesting is like this whole trend that, you know, Sequoia has written up and others about just services and software together and, you know, the different ways you can prosecute that from like building stuff and then buying, you know, services firms or, you know, taking a software company and turning it into a BPO. You see how General Catalyst bought that traveling.
Yeah, exactly. Long view, General Catalyst entity bought MX Travel which is huge, you know, and then you've got OpenAI creating the JV or with, with TPG, I think, around, um, basically building, um, a services business.
Anthropic's doing the same thing. I think that stuff is really interesting. Now the question is there, the AI going to truly be able to address all the use cases so you can make it more of a software business over time where, you know, you buy the services business, you automate a lot of it.
Or do you end up with just like a services business in which case, like, you can, what, you know, today what happens is the software entity that buys the services entity, software entity has a much higher multiple. Yeah. So the deal is fine, right? It works and maybe it's dead, too. But over time, if that combined company has a profile more like a services company, it will eventually be at the, at the multiple of a services company. The market eventually catches on to that, right? And so, so much of it depends on how ready is AI to automate that particular business process, you know, but I think that's the whole thing is really interesting because those can be a little stickier.
They're just deeper in the customer's workflow. They kind of, you know, it's like being, being the law firm versus other law firms, right? And I think that's, that is really interesting. So I was talking to the founder, the co-founder of Promera.
And, and, and he, and he was like, well, like, is, is, is the impact of AI real or like, you know, what's, what's going on? And, and like, and like, I'm like, you know, what are you seeing in your portfolio companies? And they tend to buy a lot of businesses that have a sticky customer base.
And his play was to like replace the entire back office with AI, right? And, and I don't know that that's the play that General Catalyst put out, but.
I think there's, but there's more, more, more of like the whole thing becoming businesses. But like I think that there is an opportunity there of like if, if you have like say a wealth management firm.
Which are easy to roll up, right? Because they're all just assets under management.
And, and, and as long, you know, it's akin to a doctor, right? You're going to sit down with somebody, understand the particular case and create like a portfolio strategy that is unique to you.
All the back office doesn't matter.
You see what I mean? Like, like has zero value to, to the company. So I feel like there is something there and like, and I'm sure that there's other businesses kind of like that in which, you know, I don't know, maybe like chains of, like, you know, hospitals or, or beauty salons or like whatever that is, but interactions are what matters and everything else could be automated.
I think this is where you get into some interesting questions because you alluded to one point about where are people seeing the impact of AI in their business. Yeah. I'm on the board of this company Layered in, um, Russ, you know, Russ Rien.
No, yes actually I do. So Andre funded and, uh, they have built this super cool technology that analyzes how you're using AI. Not the, not just like the fact that you're using but how you're using it, looks at your prompts, all that and then gives you like, helps you improve productivity of the business using AI. Um, but what's interesting is today, Fortune 500 companies, uh, usage of AI, they have a lot but their business impact from it is very, like, it's in general pretty questionable. Right. Right. I'm a huge believer but, um, we haven't.
People just like, [__] around, like such a random thing and it's, I, what, there was a really good post on X just like two days ago. I, I'm going to get this guy's name, I think it's like Akash Gupta, something like that, and he's on, I posted on my, my feed and he had this really good point which is, um, just because like you have a bunch of ideas, let's say it's in your backlog and AI lets you do more of them, doesn't mean they're all good ideas. And so I think.
I saw that, like most of your backlog is, it's garbage, it's garbage anyway, I know. Why are you working on it?
So, the thing that, that we have to really understand well is for these Fortune 500 companies, you know, who are ultimately the adopters of this stuff, right? Um, what is the thing that we're going to help them do better, right? Is it making things better revenue-wise? How exactly? These are big companies. Moving revenues is really hard, right? Procter and Gamble, like how do you grow revenue for Procter and Gamble? It's like it actually literally grows at CPI, like inflation.
And then, then, then it's like, oh, you know, you're increasing margin. Well, then you have to get to like, you know, real stuff, right? Like, okay, we're going to replace these people, whatever. Right.
Well, you have to replace materials.
Yeah. For CPG, it's like, you know, you're going to make the diaper out of.
Or materials if it's a margin thing.
Yeah. And so, you have to get really tangible about the, the direct business impact and not just have like, like one of the statements that's made on X all the time is there's unlimited demand for software and therefore AI will help us build more software, therefore everything's going to grow. I'm like, that is just a thing that sounds nice to say. There's unlimited demand for puppies too, but I think there's actually more demand for puppies by the way, but like there's demand for software that helps drive business results, right? A lot, unlimited demand for that for sure. It's not always obvious what that is and then picking a specific example that there's a lot of problems to solve. I went to an event on, uh, voice AI yesterday. It was really good. Yeah. Cerebral Valley had an event, Eric Newcomer. And, um, we're, they were talking on stage. They had all the, all the company, all the, all the labs and all the voice companies, Sierra, etc. And one of the things that was discussed is a huge question for me is, you know, so much of these, um, business process replacement, uh, AI companies are essentially going to be the voice call center for a doctor's office or, you know, a gym or whatever. And how much will consumers want to interact with those? Um, you know, right? I like if you're inbound, if you have to call in, fine, that's good. You know, you can call in and talk to it. But if you're a business that needs to call your customer, reach out to them.
Will the consumer interact with an AI agent?
Uh, when they don't have to, and if they don't, if it doesn't work, they're going to go back to their humans. Right. So, as a little simple example, I built this tool for my, a little fun project for my EA. It's in Claude Code and it basically like we used to like have private dining room, uh, restaurant reservations for like customer dinners and stuff. You probably do that, right? And so, um, the tool lets you type in a city, radius from where around that city, how many people, and it'll find private dining rooms. Uh, which is a little bit hard to find.
Yeah. Yeah. You have to call them.
You have to call them and they'll email them for you with a form thing. But then I added in with Eleven Labs the ability to call the restaurants. Yeah. AI agent. They all hung up my.
Wait, wait. At what point did they hang up?
They hang up on like, for me it was like 10 seconds in. You know, once they figured out it was an AI, I said, 'Hey, I'm an AI agent. I'm calling on behalf of, you know, my assistant.' And they would.
Oh, interesting. So, you had them disclose it was an AI. Yeah. I wonder if you didn't.
But I, they're not good at, they, the one at least the Eleven Labs one I had was not good enough to, and you know the, the latency is still pretty significant between turns. Right.
You sound really slow. Exactly.
Sort of like a short bus.
So I don't, so anyways, I just think there's a lot of technical problems to be solved for these mainstream companies to really realize the value of AI. And it's not simply, hey, let's have them all build more software. Um, that I don't think that's enough. Yeah. Yeah. They need to build software that actually drives revenue or cost savings or something else, you know.
So, have you heard of the, um, the work lump fallacy?
Okay. So, why don't you explain to the audience what is it and like why is it not happening this time?
Yeah. Well, lump of labor is a whole, now you're going to get me started. Oh my god. We're talking about jobs. Crank, crank. Jobs and AI. Oh my god. Okay. So, there are two equally, um, intellectually bankrupt things that you'll see on X about jobs. Um, I, I will make the case on both of them. Go ahead. Go ahead. Go ahead. Okay. Case number one, uh, is the, um, is the, 'Oh my god, everyone's going to lose their job. AI's going to be better than everyone. It's only a matter of time. You're going to lose your job, too. We're all screwed.' Okay. Uh, and I'm like, like, starve 'em. Yeah. Starve 'em. And I'm like, uh, it's kind of like maybe, I don't know. Like, there's definitely no, no, we don't know for sure. Yeah. Um, and then the counterargument, which has some truth to it, but I think it also has the problems, is nope, everyone's going to be fine. Stop complaining. You're all being doomers.
Um, and, um, you know, actually David Sacks, Andre put out a pretty good list of the arguments.
Oh, like everything is moving in the opposite direction of, of what the prediction says, right? Like more developers.
There's all, and so one of them is lump of labor which is this feeling like, oh, there's only a certain amount of labor needed and, um, people don't realize there's a thing called Jevons's paradox where when you make things cheaper to build sometimes customers want more of them and you build more of them that creates more demand for labor, right? But, um, what is missed is the nuance in between. So, for example, I think it's very possible that, I think it's already happening, that certain people will lose their jobs or lose job opportunity, you know, whether it's a call center worker or a Gen Z coming out of college. In fact, I'm, I'm, I co-founded an organization, nonprofit to help young people find work in the AI because it's, it's a real issue, right? We might.
Is that what you're doing with Clara? I'll ask Clara. That's what it was. Okay. I, I saw the announcement. I wasn't exactly sure as to what it was, but that's cool.
It's cool. It's a real thing. Yeah, it's real. And the thing that's going to happen is we're going to have lots of fits and starts. In fact, I think this ties to everything I've been talking about. Um, we are going to have so much flux in the economy trying to figure out what's durable, what's not, should we do services, roll up, should we not, what jobs are there, what's, which ones are, do you need entry level people or not? And, um, humans are going to be kind of, you know, thrown back and forth, right? And, you know, it's easy for us to all be like, oh, it'll adjust in the end. But I'm like, 'Yeah, but in the middle, somebody has got to pay for their kids' college, you know, tuition. Somebody else has got to pay for their mortgage, right?' Like, there's real humans that are in this process, you know? And so, lump of labor is just an idea that you shouldn't always think that every technology and every innovation is going to destroy the demand for labor. And I think there's truth to that. But at the same time, we shouldn't also think that everyone will be doing fine all the time because I can say for sure that that's not true. There are lots of jobs that are not coming back.
Lots of people looking for jobs. Yeah. And, and so, so what do you do with those displaced? Like is it, is there a, like I know you're not the president or benefactor or whatever but like what, what, what is the, the, the right answer? Because I agree with you that this is not, like neither panacea nor doom, right? Like there is dislocation in some areas and there is, you know, boom in some other areas.
Totally. So how does that work like society or you just let the capitalism just sort it out over.
I mean that's, this now gets into such highly debated, you know, active like Keynesian type philosophies versus high and just like whatever happens happens.
I mean there's, I read really good arguments against, you know, raising, um, what is it called.
Minimum wage, no, minimum, UBI, UBI, yeah.
Like I heard really strong arguments against it so like now I'm, I'm totally confused.
The, like, you know, if folks are interested in these topics like the two people I think that are super smart on them are Noah Smith who's an incredible economist and Substacker, like just the best one to follow and then right behind him, Alex Tabarrok, um, who's another economist at University of Chicago I think and they both have good Substacks and what I, I found really interesting about Noah's writing, I, I don't know him personally, but, but was, um, this argument that like UBI is universal basic income where everyone gets, you know, some stipend is not the only approach and that like it actually might not be attractive to people because of, for example, the fact that it could be taken away at any point since it's like a political handout and also there's some question about dignity and work and all that. Um, and so there's other proposals out there like for example, um, you know, having companies encourage through, through some tax subsidies to keep people on, on staff, um, and maybe have them do other jobs or whatever, right? And, and, um, I went to a dinner just a couple nights ago where it was a journalist, a journalist, um, a VC, four economists and.
You always end up in this really interesting.
This was the best. Are they, are you putting them together? Like who's.
This, what I did, no, I just joined, it was so cool and, uh, the nerdy side of me definitely got, got awakened. And the journalist actually said she was in, in China and just talking about how there's so much less fear of AI because there's so many things the government does to maintain full employment including a lot like what people might call fake jobs. Like literally people in the park.
Well, they just announced that it is illegal to replace a human.
Yeah. You can't replace, exactly. And so I think that basically I, I, I think that a pure just let whatever happens happens is a little bit dangerous. And frankly what it is, it's almost self-defeating because for the, all the people that are like, everything will be fine, let whatever happen. I'm like take a couple years of major job loss and people will just revolt.
And then by the way you will end up worse off.
100%. Yeah, you're going to have to move in the community.
Google French Revolution if you don't know. Exactly.
I know. Let's change topics a little bit. So, what are your... you just like me were from the SaaS era.
Yeah. We a lot of our friends are still there.
What's the feeling? What's the mood? What are you telling them? What are you hearing?
I mean, what I'm hearing is obviously some people have moved on to do the new thing, which we both are so lucky to...
Yeah. I feel sometimes like I abandoned...
I know. I know. Like I have a lot of reservations about that, but like you know...
Oh, that's interesting. Yeah. Yeah.
We have a Gainsight amazing successor, Chuck Chutkow, who's doing an awesome job. And he was...
He's my president for a couple years and he's so great and he's doing a great job of just saying how do you turn this into an agentic company and what are all the directions to make that possible. And so what I'm seeing is, you know, part of it is a capital structure question, right? So if you've raised a lot of money, a lot of preferred or if you have a lot of debt and when's the debt coming due, you know, you look like Medallia is a good example, right? Well, you saw what happened there, right?
Dude, I'm still in awe that that can just happen obviously.
I mean, and it's just like...
That can just like be... yeah, bankers take it, right?
And it's just like... and so you have this situation where it's so much depends on what's your current situation of cash flow.
Do you think that was in the model when... was it Thoma Bravo or Vista who invested?
TV. Yeah, TV. Okay. So was that in the model? Like was that one of the scenarios in which like the equity just gets wiped out to zero or like...
My understanding, again I'm not an expert but I have a lot of friends in private equity, is the whole business model is you don't have zeros. Zeros kill the returns, right? It's like it's not like venture where you have lots of zeros. This is like I'm going to have very little zeros, I won't have the crazy upside but...
I mean, Rory, you talked about this in how... who was it, the founder of Vista came, you know, a few years ago. He was pronounced like the most brilliant businessman ever because like this business will never fail, right? Like so and people used to remember... people used to talk about like software being like, you know, even though it was trading like equity was really more like debt because it was like people were so depending on it.
So there was like an arbitrage, you know, in terms of like, you know, they will never be taken out so they're more secure, you know what I mean? That even that was... and now it's just falling apart.
Totally. People used to lend against ARR which essentially is really leaning into that, right? And you know there's still some of that but mostly that's gone. And so yeah...
Factoring used to be done against like actual goods. I know. Now it's done against software. Like now the market's moving back to IATA, like which is easier because it's like do you have IATA worth of cash payments, right? And also the other thing that would happen, and this I don't know if this happened but might have, is very simple. You do a loan that... okay so you take a company private. Yeah. To maximize the return for equity. Yeah. You basically have some of the purchase price be debt. The debt's taken on by the company. This is a leveraged buyout right from KKR. And the debt is then secured against that company. So it's the... if all hell comes to high water, the debt payers get paid first. Right. And that's what happened here. And so, but the debt people think it's never going to be a problem because the company has enough cash flow, right, to pay off the debt, right? The other thing people did though was, I don't know if they did this here, is they do deferred principal payments, right? So, normally like in a mortgage, you're not... retired. Exactly. In a mortgage, you typically have your principal as you go, you can do interest only, then you're deferring it for some period of time. And so what they would do is a deferred principal payment. So the interest... the payment was really interest rate payment and the interest rates obviously when the Medallia deal were done were much lower, right. So you have this like perfect storm which is okay potentially, I don't know how the structure was, but the interest rates go up obviously that all happened for everyone and the principal payments kick in.
And the business probably doesn't do as well as they wanted. Therefore there's less cash flow.
Therefore now they can't pay that principal payment when it comes. Right.
So whether that's what happened or not, that's what's happening to a lot of companies right now.
This is apocalypse. It's real in a little box, right? Because like why would you take that is because the contracts are long, customers are steady, you're really sticky.
Exactly. They're really sticky. They're not going to take you out.
And then they're not. That's it. Like the moment they do, like the entire edifice of the argument just falls apart.
