What Ramp’s data tells us about AI, unemployment and more with CEO Eric Glyman | E2192

30 minutes before Trump dropped the news and sent the markets into chaos, someone took a very large short position, about $700 million, Jason, in notional value. And then after the crypto market took an enormous dump, they closed the position and made between $160 and $200 million. Reports vary a little bit.

Now, they have highlighted the person they think this is, a hedge fund manager out of Hong Kong. He has gone on to Twitter and said, "Hey guys, I have no instant information. I don't know the Trump family." But that's pretty speculative. I don't think we've locked down 100% that it was him, but people are just saying, "Hey, if you make such a strong trade so quickly before an enormous news event and close it, it seems like you had inside information."

This week in Startups is brought to you by Paper OS. Building an empire, Paper OS offers the largest library of AI-driven workflows for both founders and fund managers. Whether you're raising capital, launching a fund, or waiting through diligence, Paper OS unlocks simplicity and scale for your ever-growing empire. Claim your $10,000 credit at paperos.com/twist.

Netsuite. The business landscape is very chaotic right now. That's why you need Netsuite by Oracle. Download the CFO's guide to AI and machine learning for free at netsuite.com.

And Squarespace. Turn your idea into a beautiful website. Go to squarespace.com/twist for a free trial. When you're ready to launch, use offer code twist to save 10% off your first purchase of a website or domain.

All right, everybody. Welcome back to This Week in Startups. I'm your host, Jason Calacanis. With me, my co-host, Alex Wilhelm, is back. I'm back. You're back. Friday had a little bit of uh sick nannies, sick kids, the whole thing. Yeah, it happens three times a year when you got kids. Yeah, it was brutal. We're like 90% healed. We're over the hump and I'm stoked because gosh, Jason, it's a busy news day. We got a great guest. It's going to be a great show.

Awesome. Well, look, let's just kick off with this first show, the first um story here because it's been another 72 hours of chaos. Good time to review our rules of Trump. Number one, Trump says a lot of stuff. And rule number two, wait 72 hours. So, here we are. There was a big announcement on Friday that Chinese tariffs were going to be insane. What's happened since?

Well, after Trump said uh Chinese tariffs were going to go up 100% in addition to the prior levels, we have seen the stock market come back a little bit. About $2 trillion in market cap was wiped off the US stock market on Friday. That's an enormous amount of money, Jason. People were very worried. The crypto market also took a pretty big hit. Since then, things have come back today. Taking a look at where things are, the Nasdaq's up 2%, the S&P 500's up about 1.5%. So, a nice recovery bounce, not all the way, but I think it goes to show that the fear that we saw on Friday has come down pretty much materially. I don't think we're out of the woods yet on the Chinese tariffs issue, the rare earths issue, and everything else, but traders seem to be breathing a bit easier today, and that's good for everyone's portfolio.

Yeah, they uh the earth's rare earth metals is a key issue here. As we talked about on Friday, you know, like 60-70% of rare earths come out of China, but they only have a third of the known deposits and we keep finding more of them. So although they have a lock on it um in terms of uh distributing them right now, the truth is it's just because most countries don't want to rip up the earth and take out rare earth mineral metals because it would cost more than China can provide them for. So, if you can get uh your wheat from a farm in the middle of America, you probably don't want to stand up a grain field in your yard. Even if you could, you just buy from the cheapest person. That's called capitalism, globalism. So, it's really not going to be that big of an issue. And I think most countries are going to because China keeps yanking this um chain on rare earth metals, they're going to start becoming more independent just like China because we won't sell them certain uh chipsets are going to make their own chipsets. So this is how the markets work. If you don't sell stuff to the other party, uh they're going to find ways to route around you.

Market Yeah. Took a real dive. Crypto got creamed because you can still trade it um when the market closes. So it fell from 122 to 103. The interesting part of that was that um somebody made $200 million placing a trade 30 minutes before Trump's tariff announcement send prices falling which is perplexing but not unexpected. Lots to unpack there. Have we figured anything out or maybe just explain to the audience what technically happened?

So 30 minutes before Trump dropped the news and sent the markets into chaos, someone took a very large short position, about $700 million, Jason, in notional value. And then after the crypto market took an enormous dump, they closed the position and made between $160 and $200 million. Reports vary a little bit. Now, they have highlighted the person they think this is, a hedge fund manager out of Hong Kong. He has gone on to Twitter and said, "Hey guys, I have no instant information. I don't know the Trump family." But that's pretty speculative. I don't think we've locked down 100% that it was him, but people are just saying, "Hey, if you make such a strong trade so quickly before an enormous news event and close it, it seems like you had inside information."

And I think we have seen in the crypto world over time that the traditional financial world's rules don't always apply. And this is one of those times which people are saying, "Hey, maybe someone here was acting unfairly with information that the market didn't have." And I think u it was Joshua Devos of CoinDesk. He said the timing and scale of the positions opened on October 10th, Friday, immediately prior to the marketwide liquidation does raise suspicion of information asymmetry, which is a very understated way of saying that someone might have cheated the market.

Yeah. And it's important for people to note although um people are now putting crypto regulation in place and we didn't have new regulation for crypto uh for the past I don't know well for the whole existence of crypto. We really haven't had new regulations. The regulations have been see the old regulations which obviously sometimes apply, sometimes don't apply. What all this means is if you're playing in a global casino with anonymity and every jurisdiction in the world participating to some extent that's never existed before in the history of humanity. What that means is groups of people can manipulate markets at a scale we've never seen before. You want to place bets and try to move markets around stocks, you have to have brokers. Some countries allow you to buy shares, some don't. There's so much regulatory framework in the stock market in bonds. Even in gambling, you know, you go to a casino and you count cards, they've got an eye in the sky. They watch you. Well, we created a global casino. And the global casino still has no rules. And one of the rules that people perceive the market has in many cases, but it doesn't, is trading on insider information. In crypto, on prediction markets, they're kind of predicated on the concept that some people will have information. Information symmetry is kind of like saying, "I have information you don't have. I have an edge on you." Mhm. I know that I don't know the um quarterback was out all night in a strip club drinking and I saw him stumble into his hotel at 6 a.m. when they, you know, with a whole gaggle of partners and the game was, you know, tip-offs at 100 p.m. You kind of have inside information. You can trade on that. You can bet on the Jets or do something stupid like that. Here, you could bet on crypto. So, just know if you're not running the project, you are the sucker at the table. The people running the projects are the casino. The people running the markets uh and the um marketplaces, the market makers, they're kind of the equivalent of the casino, the bookies, the the sports book. You really should be thoughtful about what percentage of money you put into crypto and what your expectation is for that return. I would say I've always said low single digits of Bitcoin or the most known stable projects. If you can afford to lose it, you'll make it up if you do. And if it goes 100x, well, wow, you know, it's 5% of your portfolio. Now your portfolio is 5x. It's great. So be thoughtful, folks.

And um it is what it is. I'll just throw in that uh later on Donald Trump did post again that you know, don't worry about China. We'll sort this out. And that led to Coffeezilla, one of our favorite friends of the show. we've had him on the podcast. Uh said, "Imagine getting liquidated because of tariff fears on Friday only to have it called off 2 days later. People took a lot of financial hits, Jason. I saw people posting on social media that they were leveraged and lost all their assets. So, if you're going to trade in crypto, maybe don't use leverage as well. That seems like an additional risk you don't need. If you're going to dabble in exotics, um I'm just glad that, you know, AMD was off 8% and uh Tesla fell 5% and Nvidia lost 5%. I'm glad that we're kind of coming back from those concerns. Though, I do think that it shows how brittle the market is, Jason, that things fell so quickly over a Trump tweet this far into his administration. That was my takeaway.

Yeah. And producer Claude made us a little table here. We'll pull up on the screen. As you just mentioned, AMD, Tesla, Nvidia, Broadcom, Apple, and Oracle uh were are these the top declines or amongst the top declines? These are amongst the top declines. Some were a little bit sharper, but we looked at market cap and percentage decline to try to find the most interesting declines.

As my friends at Squarespace like to say, a website makes it real. So, like the sort of thing you need to hire a huge team to do, right? Well, it's actually easier than ever before with Squarespace. They have everything you need to get your domain name and establish your online presence right now. Maybe you want to showcase your work. Maybe you want to set up appointments for a service. You're a masseuse. You're a tutor. You can sell products. You can invoice clients. You can collect donations now. You can monetize your content. Any business model is possible with Squarespace. And Squarespace has the easiest to use, most beautiful templates. Plus, Squarespace has joined the AI revolution with their blueprint feature. And this like basically works as your creative partner, helping you generate original images and designs and customize the perfect website that matches you and your company's aesthetic. So check out squarespace.com/twist for a free trial. And when you're ready to launch, go to squarespace.com/twist to get 10% off your first website or domain purchase. That's squarespace.com/twist. Got it.

Okay. So we took two variables into account here. And if you look at them, some of them are directly impacted by China. Apple makes iPhones uh and their laptops and everything mostly in China to this day. M and then you got Nvidia and AMD uh who are very um fully very large companies now. So one might even say Tesla falls into that as well. Uh robust valuation club seems to get hit a little bit more because anytime your valuation gets disconnected from fundamentals in some way. Some people might call it a meme stock, momentum stock, or just visionary founders with incredible potential and people get excited about owning it. Um, or and or they have a brand name that makes retail want to own more of it. Um, yeah, when there's a when there's a pullback or a downdraft, they might lose double what the market does.

And thank you to producer Claude from our friends over at Anthropic. Hey man, if you want to do really great real-time deep research research like this, get a producer like Claude. Head to claude.ai/twist and you'll get 50% off your first three months of Claude Pro, which is what we use and we pay for here at the show. It's claude.ai/twist.

Yeah. All right, Jason, shall we move over and talk to our guest?

Absolutely. All right, so next up on the docket is someone that I've known for a long time, Eric Glimman, the co-founder and CEO of RAMP. If you don't know RAMP, they started off their life in the realm of corporate cards. They've expanded quite a lot since then, bringing AI agents to the fintech use case for all companies out there. Jason, they're a mega unicorn. They're doing incredibly well. Eric, welcome to the show.

Alex, Jason, it's great to see you both and and thanks for having me today.

Of course. Of course. I am uh I have a lot of these heavy heavy ramp cards in my in my uh little man purse, also known as a satchel. Don't judge me. Indiana Jones had a satchel. Uh but uh you know, it's great for corporate spend and expense management. I don't know if we have a promo code, but uh I do love the product. It's a great product. uh and you've been doing a lot of work around taking the aggregate payments that startups spend and you're able to without invading anybody's privacy um putting that out there very clearly tell us what's going on in the space who's spending on what products huh

absolutely and um well well first just uh thank you for for um giving us a shot and believing us and and letting us serve you and your team. It it it means a lot to me and and all of us. And uh you're exactly right. We're uh now uh RAMP customers over you know 50,000 organizations are spending more than hundred billion dollars per year uh across the platform. Um and through that um uh it turns out it's an incredible index um uh in an aggregated and uh anonymized way to get a sense of what's actually happening uh in the economy. Um you can see this um at any point. just go to ramp.com/data uh and you can dive into and see you know um spend increasing decreasing where um people say you know uh growth is happening in the AI market um uh how is it happening at the model uh layer and and dig in um in any way but um that's been a really fun uh project open source.

All right, let's leave that up for a second here Alex because we can review it and explain it to the audience if who's listening primarily. If you're listening and you want to watch the show, uh, we have video up on Spotify and you can go to YouTube.com and search for us. So, um, well, we had that chart of the leaderboard. I think the leaderboard was kind of interesting. If we can go back to that one, Alex.

Absolutely. Here we go. Perfect. I make it two times bigger if you don't mind. Uh, new customer count, OpenAI in the number one spot, Intuitit, uh, which makes QuickBooks, I believe, Anthropic, which makes Claude, Canva, which makes, and Adobe, which make creative software. And then by new spend uh you got HubSpot, Carta, Vanta, Pipe 17, and Avaler. I don't know if I know Avalara, but Carter obviously for cap tables, Vanta for your sock 2. And by new spend, that's interesting. So these are the top SAS vendors from last month across all of your customer base, which is startups, right? Or mostly startups.

It's um you know, that's how we started, but actually um it's really not anymore. um you know technology where it's a little overindexed in but you know this is everything consumer goods healthcare manufacturing um you name it and and some of these are are temporal so Avalara for example is uh is sales tax uh automation software and uh there's a big tax deadline uh I think actually uh on Wednesday of this week and so uh folks kind of uh bolstering all all all that side of it but um it's an interesting look um even at just what AI adoption is um or software adoption even outside of uh typical um software world.

Yeah. And this information people used to trade on information like this at hedge funds where it was available for purchase. So there would be companies that would aggregate credit card data. They would pay the credit card companies for the aggregate data. They would clean it up and they would sell it to hedge funds, you know, just like satellite companies sometimes would look at the number of c the cars in a Walmart and then they would literally back in the day, Alex, count them and then they would show the trend of how many people are in the Walmart parking lot and for how long or whatever they could and then you could maybe make some trades on how Walmart versus Target are doing um and make a couple of basis points. Again, back to that information asymmetry we talked about earlier.

Yep. it you I I I think you raised a really good point Jason that you know a lot of this data was out there but it was you know the highest bidder to go and get this and uh a big part of why we um publish this at the same time every month we make it available to everyone is you know it turns out for most people just you know small business owners finance teams people you know just trying to uh you know make improvements um have very little visibility both into what are others doing to improve their business and so we just try to open source this and let people see what you know um right or wrong uh what are people moving their businesses to so you can have the latest sense of uh what actually might be creating value not just who's marketing but what are people buying uh and then more interestingly you can uh this is even broken down in a product I love in the uh in RAMP's product called price intelligence uh and then accounting automation where you can see you know maybe you get a a quote from a vendor like Salesforce they tell you it'll be $300 per seat you can upload that contract and see here's what the rest of the market is paying Uh and so just as you can go on Zillow and see what your home might be worth, um you can figure out if you're paying uh market rate or getting charged a little too much and make your business a little bit better. And so we we love just making data available to uh people building businesses.

Well, this is just such a great startup tip. If you can create data that comes out on a regular basis and people cite it on podcast or journalists do, that's how Zillow with this estimate. And we've had the founder of Zillow on here a couple of times and uh and I've actually had the CMO who created it on and it infuriated people when they launched it. They did an estimate which then forced everybody to talk about and how it was wrong and so then everybody engaged with it which then created more press because people were like my home's worth 2 million. You're saying it's worth 1 million. This is terrible. It's like okay well we can fix it. Just tell us what it's worth and we'll adjust it. And they did that even on a very granular level Alex. they would uh do it by market. So then they created a marketing strategy in the 2.0 of that to go after the local newspapers, go after the local news programs, go after the local radio shows. And this is what's called earned media in the space. Paid is you're paid for Google ads or Tik Tok ads. Earned media means you created something of quality content that gets you on a podcast like this and we talk about it and there's an implied like, oh well this person Eric is smart because he has data and here you are and now people know RAM. You get a couple more customers so well played and uh you know Carter does this. Everybody does it. You got to actually have a thoughtful good data set and I always appreciate Eric when you tip me off on which trades I should make with this data before in our group chat. So I do appreciate that everybody gets access to it after we make our trades and place them.

Eric's shaking his head. We don't do and we'll cut that off the show immediately. But you know what? I will tell you it would not this is not financial advice, but I don't believe and I'll have a lawyer vet it with us. But um it's not actually inside information knowing processing data or whatever. That's not inside information. Inside information is inside the company or with their partners. I don't think this would fall into that. But hey, uh if you're if you're a RAMP employee or partner or you build the website, don't don't do it. Don't do it. Don't doubt don't insider trade. That's a terrible, terrible thing.

My social media feed is filled with all these experts giving me stock tips, investment advice, but we know nobody's got the crystal ball. Everyone is just guessing. They're making their best guesstimate. But now you can take the guesswork out of your business planning and strategy when you use Netsuite by Oracle. Netsuite is the number one AI cloud enterprise resource planning software on the market today. That means every facet of your business comes together in a single easy to use fluid platform giving you a single source of truth and the data you need to make smart informed decisions. And like all great businesses, Netsuite is constantly updating their product with great new features. For example, their new AI powered intelligent payment automation system allows you to not just automate, but optimize your payment process, which is going to lower your cost, accelerate your payments, and it'll build stronger relationships with your vendors. So, find out why over 43,000 businesses, including a number of my own ports, have futureproof their operations with Netsuite by Oracle. Plus, our friends at Netsuite, want to give you a free gift. Download the CFO's guide to AI and machine learning for free at netsuite.com/twist. That's netsuite.com.

But you also know how the economy is going, I think, Eric. So, there's been a lot of discussion and and we came up with a term here for it. Um, what's our term for oh, static team size. So, we was talking, Alex and I, about this trend for the past four years. Uber, Airbnb, Google, Meta, Microsoft all having the same number of employees this year as they likely did four years ago or modestly up or modestly down. So static team sizes, team size does not change. You probably know when team sizes change because they would issue more ramp cards. So what is the data saying there? Are companies hiring or not? And then what does that say about the impact of AI? Because you're seeing a lot more AI spend. So we got really two interesting points here. Let's just go with the first. What are you seeing in terms of the size of companies that are already large? Are they staying the same, getting bigger?

Uh so first of all, you're exactly right. uh company the the revenue scale and also the valuation the market cap scale of of companies uh per employee uh has gone uh either just up generally for these these mega cap companies to even you know you look at companies like um you know cursor um which are maybe these extremes you know a couple years ago we're bringing on their first customers to today you know I believe they have something like you know um 50 employees you know ballpark work for um you know a 20 to rumored you know 30 billion valuation um I think that the sheer leverage per per employee um you know uh events in particular uh industries or not I think has gone up um and while this is going on the the backdrop of this I think unemployment in the US um for the labor force was something like 4.1% um which I I believe below 5% I think is the target that the Federal Reserve um keeps as a target um when they kind of make their estimates for um you know is inflation low or not and so it's both these these companies are getting smaller well uh unemployment is is actually within uh target uh and and even below um and so I I I think to me um this is both interesting I think sometimes people focus on the fears of like do you need as many people to build companies I think the other way to look at this is actually um maybe there's going to be more companies um uh maybe there's going to be more people who are uh not stuck in um you know mid-level uh hell absolutely working at these giant organizations but instead whether it's at startups or more lean and highly leveraged companies people can just get more done with every dollar an hour and so I tend to be fairly hopeful in the near and midterm around this just for everyone's information the natural rate of unemployment or the Fed target is between four and 5% and currently we're at 4.2. So, Eric, you're dead on.

Yeah. And this is the 50-year low for our lifetime. And if you look at recent college graduates, however, they're having a heck of a time getting jobs, which I I attribute to entry-level jobs are being taken by AI because they're easy to automate. Now, Eric, you make such a great point. Whenever uh a complex system has uh some unique variable introduced to it, things can get weird and you have reactions and then you have second order effects. So if you look up the broad concept of you know cognitive biases and how and systems thinking you can jump into a rabbit hole where a bunch of you know Malcolm Gladwell type people spend and Bill Gurley spend a lot of time thinking about thinking but the truth is if there's no jobs for graduates and they're smart then what three or four of them will do is apply to Y Combinator or launch accelerator found a university for try to find something to do with their time because they'll be frustrated which is what we did When when I was a kid and we graduated school in the early 90s, there were no jobs. A huge recession. I think we were probably at like amongst young people uh mid- teens. I think in that time period, it was mid- teens. So, you know, most of your friends had jobs, but probably one in five didn't, one in six didn't. And what happened then was people started zeds or they became freelance photographers and they joined what was called in Wired magazine freelance nation. It was really interesting this concept that you didn't have to have a full-time job and stay somewhere.

Let's go to the next piece which is you are a SAS based business and it's it's got a per employee component to it because each employee gets a ramp card. So some amount of your revenue is based on headcount. So how does this impact you if you're the land and expand concept as a SAS company doesn't work? Do you have to spend more time trying to find new companies?

Uh I love that you asked this and and so even um you know SAS apart we started the company um about 2,400 and and I guess one day ago um with this this this sort of counterintuitive mission which is we actually want to help our customers spend less money uh not more um um and we would get all these questions of you know but don't you make money when businesses spend more and we would say you know yes that's true but um turns out if businesses um stick around for a um um and spend less. Maybe they'll spend less this year, but I think there's going to be a lot more uh their health span uh will increase. Um uh maybe I'll I'll make 5% less on the card spend, but you might expand into more of the business and the business might become larger over time. And so, you know, in general, we're actually totally okay um if our customers um spend less on software in in one given year. Uh we think kind of doing right by by businesses will earn us more businesses for for the long run.

Eric, just to be clear here, you're talking about interchange revenues that you make when people use their ramp cards, and that drives a large chunk of your revenue. So, you're happy if they spend a little bit less as long as they stay with you and grow with you.

That's right. And I think the same is is true. Uh we do have a component um where you can add on uh paid seat-based software. Uh it is an extraordinarily fast growing business line. um you know it's it's uh you know two years old and uh you know already the second largest uh component of what we do um but what I would say um is you know we're very happy um actually if people are are downgrading the the number of seats um at any particular point the goal is um you know we we just want to be a partner that helps businesses be more profitable. I I think that approach in aggregate while we make um may make less than any uh individual customer has is has worked over the last year. You know, Jason, we we we passed over a billion a year in revenue. Um the business is just about doubling and um you know, we're doing it while generating cash, which is uh amazing, you know.

So, are you profitable or you're not trying to be profitable now?

Uh we're generating free cash flow right now.

Congrats. What What do your investors think about that? like uh I guess pre-IPO that's a good thing because it sets you up but aren't they also some board member saying hey listen I got in the seed round I got in the series A why don't we acquire more customers here Eric like what are you doing with this free cash flow we don't need free cash cash flow we're not here for a dividend how do you manage that?

it's it's very funny um that that you say that yeah when your product is selling money to companies you sometimes want them to burn money and to go spend more and so one um uh to your point we definitely have have uh you know investors and board members were saying, you know, uh that's great, you're self-sufficient, but um maybe this is, you know, a bug, not a feature. Um can you go find more ways to spend more money and grow? And I think they have a point where, you know, something like 2% of all corporate and small business card spend in the US is is happening on ramp, but 98% is not. Uh and so I think there is a good point of we want to find ways to efficiently uh deploy more capital to

Let me hit you with an idea. Let me hit you with a couple ideas because this is what you know seasoned board members like myself do. We send you on side quests just based on our own personal experience that has nothing to do with reality. No, in some cases it's based on some reality. But I assume that um you know many people like our company uh the executives have a an American Express Platinum or Centurion, they got a United Business and then they got a ramp card. So when I am out and about in the world doing stuff, I'm like, "My rank and file employees, go ahead and use the ramp cuz you're not spending a lot." But when we have big spending, I'm like, "Get me those United points. Get me my platinum. Get me my Centurion Lounge because we'll use those." So I guess first question, where do you stand in terms of like benefits and competing against the American, you know, wonderful the platinum card is absurd. Not that I'm optimizing for these things, but man, the United flight points is a really incredible program. So, how do you think about AMX points, this incredible United, and then we were doing Bonvoy for a while and getting I mean, I didn't pay for a hotel for a couple years there. So, so tell me how you think about your value prop versus theirs. Obviously, the world's greatest moderator and the world's greatest angel investor needs the world's greatest solution for managing his funds.

So I want to tell you about Paper OS and why we use it here at our own companies launch in the syndicate. Paper OS has helped over 10,000 funds, founders, and investors, including me, automate their workflows. Their tools will help take you through every stage of the process from handling the operational details I don't want to worry about because I'm busy meeting founders and looking for great companies. For example, Paper OS has made the onboarding process so seamless for limited partners. Their investor intake forms, automated subscriptions, it just takes all the busy work out of managing our syndicate and it automates capital calls, investor accreditation, tax filings. This is tedious and important. These are the chores of running a fund or a syndicate and they have saved me and my team hundreds actually now it's probably thousands of hours which means we can focus on founders. Our friends at paperos want to give you a complimentary $10,000 credit to claim it. Just go to paperos.com/twist. That's paperos.com/twist. And if you prefer price discrimination, you can go to paperos.com/discretized.

Yep. Great question. So, um I I'll start up with like our general philosophy, then I'll hit to the the the specifics. So, so first for for most business owners, I think the average American business um has a profit margin of 8% um I think roughly last year, which you know, if you just think about the math of that um and if you're valued, most businesses in America are profitable and valued on a you know, a multiple of profits. Um a dollar saved is not equivalent to a dollar earned. You know, a dollar cut of cost is mathematically equivalent to $12 earned if you were trying to get more profit dollars. And so we think that actually reducing cost and helping people spend less, ramp helps businesses spend uh more than 5% less every year is just much more powerful um than you know the points and rewards. And so we focused on how can we take what at launch when when Alex first covered us we we were um you know help companies cut their expenses by 2% per year um you know now it's upwards of five and I think that's too low. I think it should be uh closer to 10. Uh and so I think it's it's more leverage. And second, I I actually would argue I I think that the the the luxury um in today's world in 2025 um it's not access to a lounge or or anything like that. If anything, when I go to JFK, um you know, the lines for these lounges are too long. Um they have screwed them up. This is like become a literal thing that they they the velvet ropes they just they're not building enough velvet ropes behind the velvet ropes in my experience.

Um I agree. So that makes sense. I always felt Alex that like this was a bit of a grift because I remember when I had my last job working for somebody which was Sony in 92-93 everybody was trying to figure out not how to get the cheapest flight in hotel but trying to figure out how to get the most points so they could take their summer vacation. The company just kind of turned a blind eye to it because like yeah whatever we're making billions of dollars so but I think that's probably correct. And then you have new entrance like uh Robin Hood. I have that. I was an early investor in Robin Hood. They sent me one of the first gold cards and they pay you back money like 3%. So I do think there's this trend towards that. I was going to tell you a ramp lounge that when you go to the ramp lounge, you you buzz in with your ramp card and it's like um yeah, here's a bottle of water. Get the out. It's like you're you're that's not why we're here. We're not here to get two lawn chairs in a large open room. There is something to be done here with the ramp card set to go. Be be an anti-lounge. I mean, if ramp's all about saving you money, here's where we're here's where we're cutting.

Eric, can we go back to the 2% 5% thing, though? Cuz I recall when Ramp was young, you were helping people find double spend things they were paying for twice and excising that. How have you managed to two and a halfx the amount of money you save on average? Where is that coming from?

Yeah. So, a couple things. So, so first we'll we'll we'll kind of build it from the basics to the really advanced stuff. Um, you know, on your consumer card today, one of the most frustrating experiences people have is they sign up for a a gym or a subscription to a service and they want to cancel it and you can't do it. You know, you call them, they won't pick up the phone, you wish you could go to the card, um, you can't turn it off. With RAMP, we were the first uh in the world and still one of the only companies on the planet where you can one click whether it's on one merchant or 10,000, you know, merchants 10,000 cards. Um you can say, I don't want to pay for um this gym anymore and every other merchant in in the world can charge your card except for that one. Um that should exist on other cards, but somehow it doesn't. And you know, when you're running a a company, um things like this happen all the time. You you you you have engineers paying for software. You're trying things out. um and they just add up and this gives you kind of a you know a kill switch at the central level to kind of turn off spend um that is not

Yeah. So that's I had my own I I built I rolled my own ramp experience in this way when I would have a I'd have like two cards created under my card one for like media subscriptions New York Times whatever and then one for SAS subscriptions and I would just say cancel them. I would literally cancel them in September because I know all these things are coming and there trust me if you're lost on an island like literally uh Wilson what was the Tom Hanks movie cast you're you're literally with that soccer ball and Salesforce and HubSpot they'll come rescue you to get your payment for the next year they will find you there's no way for them not to find you because they want that renewal so bad um and so I would just turn these cards off and everything would be ding ding ding ding ding. We make phone calls. They call everybody. They get on LinkedIn. They would DM everybody, email everybody to find out what's going on. How do we get this thing renewed? Um, and man, that really works well. That's my favorite use case for what you do, Eric, is to just ramp the cards down to $1 a month and just watch people lose their mind because of these dark patterns.

I wanted to share two things that I just think would be interesting for our discussion while we're here. you know, this crack uh team I have here doing live research. To just go back to our discussion, young male college grads are now jobless at the same rate as non-grads. So, just take a minute, Eric, to think about this. If you look at the college grads on the left, the men there, seasonally adjusted, three-month rolling average, uh, 22 to 27 year old by education type, non-ol, which I'll call generation tool belt. That's what we call it here on the program versus college grads, uh, in some cases getting like weird degrees. Um, the college grads spike up. Nobody needs them. Now, for women, the gap is, um, not as bad. I think there's more women in college, but for men, it does seem like maybe men men are uh not as necessary in the business workforce with their college degrees, or maybe they're getting the wrong ones. And here's uh from the Bureau of Labor Statistics, 2019, recent college graduates were at 3.25, second cell down in the second column. 2025 average, same time period, January to dece uh this is January to July versus January to December, but 4.9%. It's up 1.34, but that's that's not that's 1.34 is the point change, not the percentage change. That's like a 50% increase in unemployment versus uh those folks. There's something going on here, huh, Eric?

Well, I uh have a lot of thoughts about this one for me. So, a couple things. Um, last week OpenAI had their their dev day and they highlighted it was only 30 companies ever um that have consumed more than a trillion tokens uh on their model and ramp was one and so um we're a very very heavy user of these models and one of the things that's very unusual about these large language models is um you know I guarantee the you know all the latest models have read more about these specialized skills that one might learn in college uh than any person alive for it to specific. These models know more about accounting in aggregate than any account on.

Oh, this is such a great insight. They know more about health, you know, diagnos because that information is on the open web because these are careers and people are searching out career information. So, content producers, universities, they put all this stuff online. So, therefore, the LLMs get what a great insight. Wow. the the thing that I I think is going to be very strange for people to reconcile with is I believe over the last hundred years uh the way to wealth uh in the US was specialization. You would go to university and you would pick up a craft. Um Yep. Yeah.

That's that's uh this is the if you're watching the audio version I just pulled up. Before you go into this thing, what about the ramp? What what are you using tokens for at ramp? Are you using it to identify spend and categorize it?

That is exactly it. You know, I think one of the very tedious areas of work for companies is let's say you've you've gone you've you've booked that flight or hotel, you've you know, paid for that SAS subscription. Uh there's a lot of work that goes into go and get that receipt. Um go you know um put it into clean readable um uh form and then put it into your accounting software. You usually have controllers, finance folks, accountants kind of tagging these. So it's very tedious and monotonous. um and are um you know today RAMP is not only faster but more accurate um you know really than uh almost all accountants um you know using the platform and so just as you can kind of autocomplete your sentence for um you know things you're you're writing you can have like a faster form um rough draft essay you can have your books virtually almost done before you even open them uh to go review them and so we use a lot on accounting automation bill payment automation procurement automation so we use a lot of this but I I I think and and we can go a lot deeper to the macro point. Um you know I I think that in a world where you can through a query or an API call uh um call on this knowledge base of um you don't need to be an accountant but you know how to interface with a digital accountant. You don't need to be a you know a lawyer but can integrate um interface with with um a model that's knows more about law and how it relates. I actually think um there might be a good reason why non-ol graduates um are doing just as well as college graduates which is if you know how to use these tools um you may not need to have the specialized knowledge levels the playing field exactly what you learned in college is so superficial and light compared to the depth of AI if you just spent one year using AI tools exclusively you would be so much further ahead than trying to remember that you know top 10% of the knowledge

interestingly um to spend a if you've spent over a trillion I'm just looking at Claude explaining to me what that costs um looks like you're spending tens of millions of dollars on AI spend to do this ballpark correct with open AI uh it is uh I think that estimate is a bit high um there's a lot more efficient ways uh there's calls and then there's also what's the amount of data you send through um for the the query which which lowers the the cost. So if not tens of millions, you're certainly spending millions on AI with open AI just as a founder um you know open AI came out first but are you looking at the other models and load balancing beside them and actually thinking like people did in year five, six and seven of you know their cloud spend years 1 through five you're like this is amazing I don't have to

Stand up, my servers. Then you get to year six or seven, you're like, "Wait a second. I wonder if, you know, Google Cloud is going to beat Azure, if Azure is going to beat AWS, or Oracle's cloud." I feel like we're now in that moment where people are going to start price comparison. And then there's always Deep Seek, open-source. Uh, and there's another open-source competitor in America now, uh, that's doing pretty well. So, how do, how much do you spend your, which one was it? Together AI, I believe. Together. Yeah.

So take me through how you think about load balancing and/or comparison shopping and negotiating for tokens versus, "I'm going to just stand up one of these open-source models." Have you tried standing up an open-source model and just saying, "I'll just do it myself"?

Yes. Um, the, the, the short, like, to be direct and, and, and quick, the answer is, you have to do this. You're a great guest, Eric, because when I ask you a question, you actually listen to it. You're like, "Yes." No, it's a really good question. And, um, I, I think like one of the lenses to sort of understand this is, um, you know, whenever, um, uh, so I remember when there was this release where, um, OpenAI went from Chat like GPT-4 to GPT-4 Mini, um, or the 40 model, and people, you know, um, investors saw, "Wait a minute, this, this, this task is, um, to call the mini model, it costs only 10%, um, per call of what it would take you to call the main model," and they said it was 90% accurate. Um, and people said, "What does this mean? Are you using this? Are your costs going way down? Um, how do you deal with the inaccuracy?" And it's like, "No, no, what you do is you send, you, you have, you invest a lot in, uh, benchmarking, uh, and different tools to kind of go and see what's the accuracy of the models for certain tasks." And it turned out that for 90% of tasks, um, roughly, it is 100% accurate, and for 10% of tasks, it is completely inaccurate. And so once you learn, once the models are good at what you do, is you take this 90% of traffic that works really well and you send it to the low-cost model here, and this last 10% of traffic, you send it to the expensive model. Uh, and, uh, models are kind of like that. What, what's strange about these new models as they jump out is things that suddenly didn't work do work. Uh, things that worked before, you might be able to find a much more efficient model architecture is able to take. So you got to be on top of this. Your tech team's got to be on top of this because it's a, it's a major expense and it's a major opportunity.

After OpenAI, which obviously you got a major partnership with, who's most impressive to your tech team?

It depends a lot on the function. I mean, I, I think that, um, you know, well, who do they keep bringing up? Like, who do they keep saying, "This is impressive"? Which one?

Enthropic, in particular, for, uh, for coding and engineering. There's just something in the model, uh, I think in the way that it's developed, which lends itself to be, um, I think extremely compelling, uh, consistently for, uh, software engineering, uh, in particular. I think it's really good. I think that, um, the, the, the latest Gemini model has also, because of the, the much longer context window for very complex tasks, um, heavy research, I think has been, uh, extraordinary. And, uh, even Grok, as well, I think for physics and math, math, uh, related.

Yeah, they're doing great on math. I, I was, uh, I was at the XAI office and I was meeting with the math team specifically and they had, yeah. You know that, uh, Elon invited me there on a Saturday, parking lot full, ordered in steaks, hung out with the top people. It was very impressive to see their commitment. And they were working on that humanity's last test. Is that what it's called? Humanity's last test. Y. And they were like walking me through the problems that are like the hardest things in the world to solve. And they didn't want to like have known the answer just from like, uh, you know, I got, I stole the teacher's, you know, quiz book and I got the answers. They wanted to know how to actually do it. So they introduced a demonic AI agent into the group of agents solving the problem. And they said the goal of this demon is to try to give the wrong answer. And then these five agents have to explain to it why it's wrong. And it was really, really interesting.

What are you showing here, Alex? Explain to this is the, uh, humanity's last exam, the test we're talking about. And this is the leaderboard of current winners. And Grok 4, GPT-5, and Gemini 2.5 Pro, the models that Eric just mentioned, are at the top of it.

Eric, can we talk about agents, though, for a little bit? Because you guys rolled out agents for controllers in Q3 and you rolled out agents for, uh, accounts payable in Q4. How strong are these tools and how are they different from, I think you rolled out Ramp Intelligence back in like 2023? So to me, it feels like a reprise, but I presume they're doing something different.

The, the big, um, I, I would say when you think about 2023 with intelligence, large language models could go. And I think that the dominant design then was this idea of a co-pilot. Um, you could feed a questions and it would suggest kind of the outcome. What's unique about, uh, agents is, I just, I think there's a lot of jargon around this. As I think of them as models plus tools. Um, you know, they're not just the model response, but you give them permission to go do something on your behalf. Uh, whereas intelligence might have said, "I suggest you categorize it in this way. I think this might be fraud." An agent will go, uh, and can automatically approve that report for you. Can go, uh, actually initiate the buying purchase, um, process from an advisor to an assistant, essentially. Instead of telling you what you might do, it just does it for you.

That's right. Um, is I think one of the big things and some of that has to do with just the, the sheer level of improvement in accuracy and predictability, um, coupled with the ability to handle more generalized task, going on a web page, completing some outcome. And, and maybe to explain what's so useful about the policy agent, ever since, you know, uh, Enron, uh, happened and that failure blew up, there was this act called Sarbanes-Oxley, which says that for any transaction, um, you can't buy the thing and review the transaction yourself. Someone else needs to do it. Makes sense. Um, good idea for people keeping books, but what it's resulted in, um, is for anyone who's worked at a large company, um, you know, decades of, you know, pardon my language, but just like corporate where like, you know, if you buy like a $5 coffee, your boss needs to sign off. Was it appropriate for you to buy like a coffee or a hotel or something like that? And it's just this cottage industry of an unbelievable amount of work where, you know, today most people, they get an expense for their report and they don't review it because it's a waste of their time. Uh, or they do, and it's, you know, um, is do you really want the boss review for doing, uh, is it deep human intelligence to go and do this? Functionally, what we built in the, the, the policy agent was, um, you know, we built an AI that knows your expense policy in detail, can see all the context around the, the transaction, and with 99% plus accuracy, is able to approve, flag, or deny transactions, um, on the manager's behalf. Um, we've seen, uh, leaders, um, you know, like a Notion or a Cora or, you know, today, um, you know, thousands and thousands of other companies adopt this, and they're able to automatically approve 90% of transactions that are in policy. I can show you all the reasoning for why that is. Flag the last 10% you catch like 15 times more, uh, out of policy spend, and you, you save a whole lot of time, uh, that, um, just would have been, you know, people doing low-value tasks. And so it's a bit of an example of, it's not really in anyone's job description to do this stuff today, but it's a perfect use case for an agent to go and just make work feel a lot less cludgy. Um, and so, yeah, I'm curious about adoption of agents inside of the Ramp customer base because you mentioned earlier that your customers are now much more than tech startups. So when you look outside of the realm of tech, do you see a similar adoption curve for agents amongst your more mainstream customers?

I do. Um, and in some ways, it's actually been, um, you know, almost faster. Um, I, I think one, one of the lenses to think about is like, there's this revolution happening in, in, in the world of, of, of AI. Uh, and people know that this technology is out there, but most businesses don't have, you know, a single software engineer working at their company, let alone, you know, an engineer working just for their finance team. Uh, and so for our customers, they're not saying like, "Hey, I'm coming to you for, you know, um, you know, go, uh, sell me the AI product." They're just saying, "I want to close my books faster. I want, uh, you know, convenience to get, you know, the expenses in quicker." Uh, and so if it's easier and it's quicker and intuitive and it's embedded, they'll just turn it on. Uh, and so, so there's no concern from them about hallucinations or mistakes because if you go back a year ago to AI, people were talking about the flaws more than the productivity.

It sounds like in this case, because you've packaged it up in a way that's like, "Save time," people are just willing to, to go with it.

That's right. And, and, and what's so interesting about in our model, there are hundreds of millions of transactions that occur every year on, on Ramp, and it goes at the end of the month to a controller who, you know, they are quite literally hired by companies to review and ensure the expenses are accurate. And so, um, rather than before, you know, they're tagging every transaction by hand, you know, and then reviewing it and then pushing it over, the transactions are all categorized, they, they, they, they review it, um, and based on their, what they approve or deny, um, uh, it's functionally a large-scale context engine to learn not just how companies keep their books, but how you specifically do this. And so with every progressive run, uh, less and less needs to be reviewed. You gain trust. Then you see these companies move from, um, heavy review to, you know, "I trust the model to go take this through on this 90%." And so that training step has been helpful.

How long does that take for them to go from, "We'll try this out" to, "We're confident this is taking care of 90% of the work for us"? Because that seems like a pretty important, uh, time spend to understand AI agentic adoption.

Yeah. Um, not that long. I mean, I, I think even for the first month, um, that people go and take, you know, do transactions, I think we, we, you know, one-shot see, um, you know, it's 90% plus accuracy on our recommendations are ultimately accepted. And every progressive month, that that goes and teeters up to the 95, 99, and goes from there. And so, yeah.

All right, everybody. Eric, you are an amazing guest. You got to come back soon. I would just like to have you on and talk about like the news with you. Great guest. You know, I have a four-quadrant guest expertise and candidness. And you're like, in my top right quadrant. You got great expertise and you're candid. That's how I cast Freedberg, Sachs, Gersonner, Gurley, all these great people I cast into shows previously. Are they candid and are they really competent? You're in the candid competent quadrant. Great job, Eric. Everybody go, uh, try Ramp. It's awesome. I use it. Yeah. Not an advertisement, just authentically, I use it and love it. All right, Eric. We'll see you soon, man. Thanks for your time. Thank you, guys. Upload that audio file. Um, yeah. Thanks, pal. Uh, what a great guest, huh, Alex? Just, I, I love a guest who just was like, "Yes, I'll answer that question." Not the question my PR department asked me to filibuster in and shoehorn into the discussion. Eric has been like that since the very earliest days that I knew him because I, I covered Ramp back when it was in its early rounds. Not, not trying to brag, just I haven't. Part of the reason you're here is that you have such great industry knowledge having been at TechCrunch as a high schooler. Um, basically, but he was always that candid. I mean, even, even back in the day, he's managed to maintain it, too, which is even rarer, I think, amongst founders who get to the deck of corn, you know, stage, they tend to get a little more closed off. Not so.

All right, we're in, uh, we're in our docket. If you want to follow the docket, you can watch us build the docket starting the night before. Thisweekinstartups.com/docket. And then as we're doing the show, you can see me in real time looking at the docket and I do strike through when we've covered something. Uh, and I really want to cover this story about the broken handshake deal with YC. And God, it seems like every day is another YC drama. Let's go to the drama. Tell me about the drama. This, this week in YC drama.

Yeah. Well, you know, YC is, is very large. It's very well known. It's well capitalized. Has a lot of founders. You put all that together, Jason, you got to get some drama. Now, here's what's going on this time. There's a founder by the name of Daniel Jung. He is in charge of a company called Omen, which calls itself the first agentic investing platform. Tagline: "Trade anything." Pretty standard. Why not go through YC? This company applied late, got into YC, used that, um, impremature, that, uh, that label, that YC, uh, credibility to go out and hire people, and then backed away, turned down the traditional $500,000 YC SAFE investment, and essentially just left the program after taking their whipped cream off the top. This led to a lot of folks being a little bit concerned because handshake agreements are pretty important in early-stage investing and especially in accelerators like YC. And so the founder was heavily criticized. His point is, well, hey, you guys say drop out of college. Why can't I drop out of YC? And folks are pretty mad. So I want to start, Jason, by asking you, uh, explain the importance of handshake agreements in early-stage investing. And then I want you to give this guy a grade from, he's being the good kind of trouble to, he just torched his entire reputation in Silicon Valley. Uh, this kid's a genius. Um, total genius. If you want to do something punk rock, that's what...

Pause. Pause. You stop screen sharing. I'm going to pull this up. I have a better version of it. Okay, great. Yeah. Yeah. I was just showing people the docket. By the way, if you're looking at the YouTube video, you can see our docket here. That's why I encourage everybody to go to the docket. Um, thisweekinstartups.com/docket because you see the notes that we're actually reading from and our research team did and producer Claude did. But yes, you share and I'll talk. Um, so here's what I want to say. Boohoo. Y Combinator complaining about this and using the, the, um, YC brand to say it's a YC dropout. Harvard doesn't complain when Zuckerberg does it. And in fact, Y Combinator is known for asking that question. "Tell us when you broke some rules." I don't have the exact question, but they ask people and they sort for people like Sam Altman, who are rule breakers, who do, you know, crazy things like take a nonprofit for open-source, you know, LMS and make it a for-profit. Yeah. You know, like that's what they're optimizing for. They're optimizing for punk rock. And then they want Daniel, the Daniel Young, uh, on, uh, Twitter, they want him to be well-behaved and stay in his lane. I mean, F off. This kid's punk rock. He can say, "Hey, you know, I did the, um, handshake, but I didn't sign the SAFE. I'm out." In fact, he can sign the SAFE and say, "You know what? I don't like this. Give, I want you to let me out of the SAFE." Now, they don't have to let him out of the SAFE, but he can be punk rock. That's like the whole reason, you know, uh, founders, uh, win is because they're willing to be a little punk rock. And I, if I'm YC, the proper response from the YC people, um, you know, with this, I see Pete Kuman, who I guess it looks like from his Y Combinator logo. Actually, this triggered them. Wow. Oh, yeah. So, you have multiple Y Combinator people responding to him. Yes. And they, they were very, very, very unhappy. And, uh, Daniel later on said to Mr. Pete Kuman, "Uh, Pete, respectfully, blah, blah, blah, blah, you let us into YC. We're grateful that you were willing to bet on us. We think you were right to do so, and we want the rest of the world to know why, even if our stint at YC was shorter than initially anticipated." But YC seems really, really mad about this. And that's why I was, I was curious about the handshake element because I didn't realize signed. It's a handshake. That's why they call it a handshake, you know. Is it, um, rude? Is it unethical, immoral? All right, whatever. Yes. Yes and no. Um, but the deal's not signed. Until the deal's signed, you, you have the right to back out of it. You know, you can say like, "Okay, I want to do that." But if like on the way to your car, somebody's like, "I'll just put a million dollars into your company directly at a $10 million valuation. You don't have to give 10% to YC for, you know, $200k." Well, okay. Um, YC should be happy for them. The reason why YC is overreacting here is, well, one, they're, they're super dramatic. Everything they do has to have this drama, but this is anti-founder, you know, and they're like really concerned that this is going to become a trend. I think they've been very threatened by some of the new, um, Speedrun from A16Z. Yep. Arc by Sequoia. Pair has their summer program. We have Launch Accelerator and Founder University for a long time. We're not a new entrant. Um, Techstars is coming back. Antler. All of these programs are better for founders in my mind than going to YC. Not that YC is bad. YC is, you know, as good, but I think these other programs are better because they give better terms, okay? And you're not one of 500 founders or 250 founders. They're more bespoke. So if you go to Speedrun, if you go to Sequoia Arc, if you go to Pair, if you come to our program, and I'm talking my own book here, obviously, it's less of a factory like Y Combinator, and it's, you're not going to get lost and have give a one-minute presentation on Demo Day, right? Other programs like ours, two, three minutes, you know, you get a little more time. In ours, you're one of 12 companies, not one of 200. Or in Arc, I think they take a dozen. So they're more bespoke. If you can get into one of the bespoke programs, I think you'll have a better experience. And I think that's what Y Combinator is feeling is like they have competition now. So, and also there seems to be something with young founders. And I don't know why this is, but there seems to be some pent-up. And maybe it's just a nature of being number one in the space and having such a great reputation. So it's, you know, it's actually in some ways a compliment that dropping out of YC is the equivalent of dropping out of Harvard. If I was Tyler, who says, "Imagine breaking a handshake agreement and bragging about it on social media for likes. That's a terrible tweet." What they should have said was, "We appreciate the founder. We think they're amazing. We wish them great luck. We wish they would have come to Y Combinator. We hope that when they raise their next round, maybe we could, uh, participate. We wish them all the best. If the program's not for them, we want them to do what's best for them." Yeah, that's the right response. Tyler's response, not correct. Peter's was, "For everyone wondering, dropped out of YC. It's just an edgy way of saying broke a commitment and contract. They're attacking a new founder. You should have some grace for the new founders. They're going to do things that are spicy on the margins, Alex. They're going to do things that could annoy you as a more senior executive or somebody who's been in business for 30 years." Sometimes founders do things. I've had founders do, um, like multiple times. This has happened with like 10 founders. I've had 10 founders do a round of funding and not tell me when we have rights in that round. When we have information, they didn't get your Prata. Oh my. Well, no. You could then have to go back and reverse it. Or they sign a deal without telling us. Or you know, and then we're like, "Well, no, but did you talk to your lawyer first?" Or like, "No, no, I got this great deal. I signed it." And you're like, "Oh, okay. Just, you're supposed to do that." So I always tell founders now, like, before you do any deal or you give somebody three board seats, like a founder did recently, give two or three board seats, I'm like, "Please call me first. Please call me first." Because I've already made my money. I'm already micro-famous, as everybody knows. And I'm slim now. Like, everything I want in life, I got a family. I'm, I'm Schfelt again. I made my money and I'm a micro-celebrity. And a micro-celebrity, let me tell you, it's a pretty fantastic place to be. I get a lot of great invites. I can go to F1. I can go to all this stuff and hang out in the pits. I don't have to buy a ticket. It's fantastic. If I tell you anything, it's in your best interest. I'm only doing it to help you and be a good participant in the ecosystem. Like, literally, that's 100% of my motivation. So, please don't give two seats to somebody. Please don't give two different investors or three different investors three different terms and side deals and, you know, like, just keep it standard. And if you're going to sell the company, let's have a process. Don't just sell it to your friend and then, you know, not do the process. Don't give yourself shares without telling the board members you want to give yourself a new equity grant. Like, there's a process here and let's not get you in trouble or, you know, cause reputation damage. But here, I think Daniel Young is, um, punk rock. And I'd like to have him on the show on Wednesday.

Okay. Well, we'll reach out to him and see. And I believe the YC question you mentioned is, "What social hack did you do?" Something along those lines, asking founders how they managed to circumvent and kind of short-circuit the attention economy. I think, uh, the founder of Cloley, who we had on the show back in the day, is a good example of, uh, the current young founder archetype, Jason, willing to kick sand in people's faces to make a lot of noise. And here we are one more time. Um, so there also was another comment. There's this account called Spec, S P E C, that is obsessed with Gary Tan. I love Gary Tan. I've known Gary Tan. I think he's a great human being and I think he's a great founder and great investor. I know he had a bad breakup with Alexis Ohanian, who I also have a lot of respect for. So I don't know, sometimes founders can break up, but I think Gary's great. But this account, Spec, which is open CV with underscores on either side on Twitter, keeps CCing me in this because I guess they want me in on the drama. Or they, they CC a lot of people. Yes. But, um, they had an interesting tweet. Yeah. So the tweet reads, "So you can't neg YC, but YC can neg you." If you're not familiar with the phrase neg, it means to just diss somebody essentially, Jason. Uh, and so in this case, there was a girl who quote lost her full-ride scholarship because she dropped out to do YC. And then it shows an email. Uh, and the email reads, I'll just read it out for folks. And by the way, neg means negative in this sort of space. So if you neg a girl in these, like, you know, those crazy pickup artist, um, pickup artist stuff, I couldn't find the neg is like, "Oh, wow, you know, one of your earlobes, Alex, is longer than the other. That's kind of weird." And Alex was like, "Will you date me?" "I'm already dating you, the co-host." Okay, let's read this. I'm self-conscious about my ears. All right. The email reads, and this is to a founder who is perfect earlobes. Thank you, Jason. Also, I, there's one person in the world who's allowed to break my phone silence, it's my. Sorry about that. All right. The email reads, "I wish I was emailing under better circumstances. It's become clear that you three cannot operate as a functional team. We funded your company under the assumption that you could, and as a result, company name can no longer participate in YC. You have three options: Shut down, give us the money back. Keep the company alive and give us the money back. Or keep the company alive and keep the money. At which case, we'll give you back your shares and cancel our SAFE and no longer be an investor. I strongly recommend you pick option one or two." So this is them essentially saying to a company, you, that's from Gary? Is that confirmed? No, this is from someone with the initials MS. I know Michael, maybe. I wasn't going to say it out loud. I also know that could be Michael Seabold, but Michael's fantastic. Gary's fantastic. But the point here is that YC will often, well, not often, but sometimes play a bit rough. And so if they're allowed to play a bit rough, then why can't founders do the same? Jason.

Okay. Um, all right. Listen, it's business. Things can get a little chippy sometimes. It can be very annoying if a founding group like this, you fund them and then they start creating chaos because what you don't want to do is have a distraction in the incubator for the other companies. That's not fair to the other companies. So, just like if you got accepted to Columbia or NYU and you know, you start causing drama everywhere and we've got you on a scholarship and you're going to be on the basketball team or the hockey team and you're just running amok and it's like, well, maybe this isn't the right opportunity for you. We take our scholarship back. In this case, it's $125k. In this case, if I was YC with billions of dollars under management and, you know, dozens of unicorns, um, including their most successful ever, I think, is Airbnb, which is worth, I don't know, close to a hundred billion dollars. Um, my two unicorns are worth more than that, but I know it's not. Uh, $74.74 billion. Airbnb is an incredible company. I think it's the largest one to ever go through there. Robinhood's worth $125 billion and Uber is worth $197 billion. There you go. So my two are bigger than their biggest, but it's not a competition. Alex, I see it's not, there's not a scoreboard here. Oh, it's everybody just tries to help the ecosystem. I'm kind of making a joke here because people do get competitive. And I can understand if it's Michael Seabold or if it's Mary Susan, whoever. Um, you want to have a clean separation if there's going to be drama. I don't think that email's too aggressive except for maybe the last line, like, "Pick one or two." But sometimes you've got to be firm with a group of people who are causing chaos and be like, "Listen, one, two, three." And I, I have had similar situations happen, um, where something comes out during due diligence. We haven't signed the deal yet, but maybe we're between handshake and due diligence. And we, you can say a thousand times to a founder, "Pending due diligence," and they will not hear that. That is like a, a frequency that like they're not capable of hearing. But if you do due diligence and it turns out like your customers don't match or whatever, or, you know, it doesn't feel like particularly defensible technology, you have the right to back out. Here, I think, I don't know. I'm, I might be on Michael's side here that he gave them some great options. He said they could keep the money and they would, they would get off the cap table. I mean, that's that's the opposite of a lawsuit. Jason, I think I'm going to give Michael the win here that I think actually he gave them a firm, crisp set of decisions. You might say like, "Pick one or two" is my best suggestion. Might be a little aggro, but I don't think it's overly aggro at all. I think he's giving them good founder advice, which is, "Hey, listen, if you guys want to have chaos, that's fine. That's not what YC is about. We need harmony because we've got two, like I mentioned earlier, like there's 200 other, there's 199 or 299 other people in your cohort. Like, please. But stay in your lane and just be productive for 12 weeks. The end." So anyway, long story short, YC is amazing. Gary's great. Michael's great. And this kid's great who's being a little punk rock. You can't optimize for punk rock and then be upset if somebody's punk rock with you. The end. Full stop, everybody. And for the YC people, did they delete their tweets when they were dunking on the kid? I saw there's a bunch of deleted tweets in that thread. I don't know whose tweets got deleted or if we know. I took those screenshots of the, the Pete and Tyler tweets myself and then I grabbed that thread, Jason, just to highlight how many things have been taken down because Daniel had removed some of his initial tweets that kicked off the controversy. So, we had to find them via business. So, Daniel did, uh, deleted his tweets. Anyway, I'd like to have Daniel on. Maybe this is a good investment for me. And if you get rejected from Y Combinator, don't wait six months. Email your boy JCal. Okay? It's very simple. JasonCalanis.com for life. Or you can email me if you love the All-In program, AllIn.com. Or if you want to get a meeting with the 11 people on our investment team, forward your YC application that got rejected to YCLaunch.co. YCLaunch.co. And you will get a meeting with my team within 24 or 48 hours, including a little bit of weekend time because my team works a couple hours on the weekend to meet with founders. We will meet with you quickly. We'll do a 20-minute first, uh, call with you. You pitch us your product or service, 10, 15 minutes. We ask you one or two questions. You ask us one or two questions. Then at the end of 20 minutes, you know, we, uh, end the call and then we will talk to you and have a follow-up and see if it makes sense for us to go to a second call. We do this because it's founder-friendly. Like it's just good to, you know, save you time. And if you don't get into YC, I don't think you should just apply to YC. I think you should apply to our program, Founder University, if you're pre-revenue, you know, like year zero, or apply to the Launch Accelerator. We have a common app, launch.co/apply. I also think you should apply to Andreessen Horowitz's Speedrun. Antler, Pair does a Pair VC. Mah does a great program over there with Pezman. These are great investors, great founder-friendly folks. Ruof and the team, Stephanie and everybody, they do the ARC program. And we'll put those links in the show notes today. I am not a zero-sum person. You should, I think the Y Combinator folks have a little bit of like, circle the wagons. They're not, I, I think they're just a little too cutthroat. I'll be honest. It's a bad look because it, they don't need to be. When you're winning, you should be magnanimous. I've had to learn this in my life. Uh, all of us have to learn this. Chamath's talked about learning this. Uh, when you win, especially in, you know, when you get to the top of the, you know, um, the, the top rungs of the ladder, uh, where I've been lucky enough after a 30-year brutally hard career, fought my way in here, I get it. I had to go punk rock. I launched a Zen. It's as punk rock as it gets. Like, I couldn't get published, so I started my own magazine and photocopied it. You know, people didn't respect me. I started my own tech conference because I couldn't get into other ones. Period. Full stop, right? I started my own podcast. Um, it's okay to be a punk rock, but then when you do win, you got to flip, Alex. And this takes personal development work. And it, it starts from the top and the leadership. The leadership has to say, "Hey, we've won. We're going to be relentlessly magnanimous." If I could put a post-it, you know, here on my teleprompter, it would be, "Be a mensch." Like my guy Dave Goldberg. Rest in peace. He was the menchiest guy ever. He, I modeled my career after Dave Goldberg, Goldie, rest in peace. He ran SurveyMonkey. He ran Launch.com in a way. I did Launch.co as a tribute to him because I always loved the brand Launch.com, which was his music, uh, startup. He gave me time when I was coming up in my career that he didn't need to give me. And if you ask anybody who met Goldie, he gave everybody an hour or two. He didn't need to. He was rich already. He was living the life. He could get any meeting. You could hang out with any powerful person you wanted. Died too young. But when he was alive, what did he do? He was, he was, if you met 10 menchs and there was a mench lunch, they'd say, "Where's Goldie?" Because we want to have a mench at this lunch. Literally, he's a mench's mench. That's what I aspire to be in my life. I aspire to be Goldie and be a mench. Another amazing episode of This Week in Startups. I'm going to get emotional, so I'm going to leave it there. All right. Unless you have anything else we need to add or any housekeeping we need to do here?

No, other than saying that we're going to have a, uh, a really fun AI TAM sheet all ready for you on Wednesday. It's going to be great. I noticed you taking those notes in the docket. Well done. Uh, if you, uh, want to tune in live, go to ThisWeekinStartups.com/YouTube and it will automatically send you to YouTube and subscribe you to the show. All you have to do after you subscribe is, it gives you a little pop-up on YouTube, "Hey, would you confirm you like to subscribe?" You confirm you want to subscribe, but you hit the alert and the bell there. And then if you could do JCal and Alex a favor, write us a review on iTunes. I hate to beg for reviews on Apple Podcasts, but it is a big part of the show getting surfaced to new people. If you write a great review, we're going to shout you out at the end of the show, which we're about to do at the end of the show here. We'll read one of the great reviews. And if you email me your review at JasonAllin.com or jason@calacanis.com, all goes to the same place. I'll write you back and say thank you. And you get to say hi to me because I'm a real person trying to be a mench every day of my life. I am a real human on planet Earth. And if you get to a position of power, which Y Combinator is the height of power and you have this impact on people, best advice, do what I did. Do a little personal self-discovery, you know, whatever it takes. Just reflect. You can never go wrong by being helpful and being a mench. And you can go wrong by being critical of people, especially publicly like this, especially for a nascent founder because some people will frame it as bullying. I'm not framing as bullying, but obviously that, that's the reaction online here. So this is the thing about the power imbalance that YC maybe needs to, and this is something Gary has inherently, um, in his DNA. I believe he's a very competitive person. He's a full-contact person, as you can see with his, you know, opinions on, uh, San Francisco. All good character traits, but he should probably build into the YC, um, culture being magnanimous. Be a little magnanimous when you're at the top, right? Because you, you sometimes forget how much power you have. I don't anymore. I, you know, I know that if I mention somebody on the podcast, it's going to carry a little bit of weight. I'm not overindexing on it, but it could negatively impact them. So, I've been more thoughtful in how I'll say things. I used to be a little more Howard Stern, a little more shoot from the hip, which is true. But we all learn as we age, and we all become a little bit more patient, a little bit kinder, and that's when you can give back. But more on Wednesday. We'll talk AI TAM. We'll have more guests. Uh, actually, we have a very fun guest on Wednesday, Jason. So, everyone stay tuned.

Tease it. Even if they're not confirmed, if it's just, you might be the, uh, first initial S, second initial J, perhaps. Oh, Steve J. is coming. He might be. Hi, Steve. So, Wednesday. I love, I love Steve Jervson. That's my guy. What an investor. Board member of Tesla, SpaceX investor. One of, I mean, you want to talk about a mench and a visionary investor. Steve Jervson, the J in DFJ, and now he's got his own venture firm. We'll see you on Wednesday, everybody. Bye-bye. Bye.