How Companies Are Actually Spending Money on AI Now

In theory, all of this AI spending has to deliver some kind of return. Companies (or other end users) will have to get tangible value from its outputs in order to justify the billions spent on research, chips, energy, and more. So what's actually happening at the corporate level? On this episode, we speak with Eric Glyman, who is the co-founder and CEO of Ramp, which helps corporations manage their expenses. As such, he has front row visibility in terms of what's actually being spent and who is actually getting the money. We talk about trends he's seeing in terms of spending going toward companies like OpenAI and Anthropic, as well as how AI tech is affecting the operations of his own business. Read More:SoftBank Shares Soar as Masayoshi Son’s AI Vision CoalescesTrump Pushes to Make US an AI Superpower, With Fewer Guardrails Only Bloomberg - Business News, Stock Markets, Finance, Breaking & World News subscribers can get the Odd Lots newsletter in their inbox each week, plus unlimited access to the site and app. Subscribe at bloomberg.com/subscriptions/oddlotsSee omnystudio.com/listener for privacy information.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Finance is increasingly being built on digital rails. But as the adoption of virtual assets, tokenization and stable coins grows, can regulation not only keep pace but help accelerate it? Find out more later in the podcast.

• 0:15Speaker 01

  Underdog already has team picks, player props, money lines, spreads, you name it, they got it. But apparently somebody over there said, you know what's missing? Screaming like a maniac about a rookie card on your couch. And we love them for that because Underdog now has rips. You get a digital pack starting at just five bucks. You check the live odds, see which grails you could pull, then rip it open to reveal a real, physical card. And the grails are no joke. They have cards up to $125,000 in value and plenty to pull. From football and basketball to Pokemon, with new limited edition packs dropping regularly. Even cooler, you can decide to keep the card and Underdog will ship it straight to your home or sell it back instantly at 100% of its fair market value. And the cash goes right in your account.

• 1:00Speaker 01

  No listing it online, no stranger low balling you in the messages. Just rip, reveal and cash out instantly. Or make some room on your shelf. Download Underdog and get in on the chase. Get five free rips packs when you deposit $5. Real rips, real cards, real cash. Only on Underdog. Unleash your dog. 18 plus, terms apply. Top grails are extremely rare.

• 1:26Speaker 02

  Bloomberg Audio Studios. Podcasts, radio, news.

• 1:42Joe Weisenthal?

  Hello and welcome to another episode of the Odd Lots Podcast. I'm Jill Weisenthal.

• 1:47Tracy Alloway?

  And I'm Tracy Alloway.

• 1:48Joe Weisenthal?

  Tracy, we're recording this January 8th. We ran a good piece in the newsletter this week from Skanda about AI spend is like a sort of like meaningful macro driver or getting close to where it like starts to move the dial.

• 2:01Tracy Alloway?

  Yeah, that was a really good piece. And I have to say some of it is slightly worrying. But I think. One of the big things that's happening now is, okay, AI has become such a big pillar of the market.

• 2:15Joe Weisenthal?

  Yes.

• 2:16Tracy Alloway?

  Right? Like the entire S and P 500, basically.

• 2:18Joe Weisenthal?

  It's like an AI play.

• 2:19Tracy Alloway?

  Yeah, it's an AI play. And so at some point, the hype has to be matched by reality, i.e. all that investment has to be matched by some sort of revenue, right? You have to get money out of making this investment.

• 2:32Joe Weisenthal?

  I think there's three ways things could go. I've been thinking about this. There's three ways this could play out. One is companies don't get a lot of productivity gains from these tools. They cut back spending. There's a bunch of other projects get delayed, market goes down. Another possibility is there's this great productivity breakthrough. Companies are more efficient than ever, incredible boom, that's great. And then the other possibility is that none of this matters and they build God at one of these labs. And then everything we know about economics doesn't even make any sense. So to even talk about productivity or the S and P 500 or earnings in that regime is just like, it's like a secondary concern to like the way the world has changed when they achieve AGI.

• 3:15Tracy Alloway?

  Joe, are you okay?

• 3:16Joe Weisenthal?

  No, I think those are, those are like the three.

• 3:19Tracy Alloway?

  Yeah, you might as well go big with your scenario analysis. Yeah, but you're right. It could go either way. And obviously, there's a lot of talk and nervousness about a bubble in AI at the moment. So I think it would be a good idea to maybe try to get an understanding of how much companies are actually spending and benefiting from AI.

• 3:38Joe Weisenthal?

  That's right. And we have the perfect guest because he has a great view into what companies are spending money on, including AI. And also, as the CEO of a company himself in the tech space, but not directly in the AI space, is sort of user of AI tools. And maybe could talk about what's being used, what's not being used, where productivity gains are being had. We're going to talk about all of this stuff. We're going to be speaking with Eric Glyman. He is the founder and CEO of Ramp, which is a New York City based company. It's a spending management platform for companies, help them deal with expenses. We might also talk a little bit about expense management platforms because I have.

• 4:16Tracy Alloway?

  Do you have complaints?

• 4:17Joe Weisenthal?

  I have headaches about dealing with expenses.

• 4:19Tracy Alloway?

  I think a lot of people do.

• 4:20Joe Weisenthal?

  Yeah. I guess that's why there's a business, a new business to be made. Eric, thank you so much for coming on Oddloss.

• 4:26Speaker 05

  Joe, Tracy, thanks so much for having me. It's great to be here.

• 4:29Joe Weisenthal?

  Real quickly, why don't you describe a little bit? What's Ramp? What's your story here?

• 4:33Speaker 05

  Well, you can think about Ramp as a financial operations platform. It's a single place where companies can issue cards, make payments of all kinds, and even automate both expense management, which we'll get to, and accounting. But the ethos of the company and why we exist is actually to help companies save time and money. We're the only company in our space that actually reports back to our customers, how much money and time did we save you? Over the past four years, we've saved our customers over $2 billion, 20 million hours, and half of that has actually come in the past year. And so we serve 30,000-plus companies from small, medium to publicly traded.

• 5:09Joe Weisenthal?

  Got it.

• 5:09Tracy Alloway?

  What was the gap in the market that you saw? Because on the one hand, as Joe just laid out, a lot of people hate expenses and they're clunky and very bureaucratic and they take forever to do. But on the other hand, this is a space that is dominated by some very powerful legacy players, right? I'm thinking about American Express, for instance, and you're basically going up against them.

• 5:34Speaker 05

  These are companies that are great in their own right, but I think we're built for a bit of a different era. Even the company you mentioned, the founders quite literally wore top hats, you know, and are, you know, thinking not so much about, I think, really the needs of the 2020s, where I think luxury in the 2020s is actually having an hour to yourself at the end of a long work week versus you have expense reports to do at the end. And so we thought the gap was a few folds. First, could you actually infuse technology not to make an easier-to-use expense report that only took an hour instead of two hours, but actually an expense report that does itself, books that keep themselves? And so the difference was, we saw an opportunity to create a card where you can tap it, make a purchase, we pull the receipt from the merchant or your email automatically so your

• 6:17Speaker 05

  expense report is done for you, your books and records are done for you. And more for business owners, we found this strange distortion where they were trying to market products like spend more money, earn more points. But every business owner I ever met actually wanted to spend less and be more profitable. So we just tried to keep it simple and build a company just on those principles.

• 6:37Joe Weisenthal?

  I mentioned in the intro, because you have this expense management platform, you have some insight into what companies are spending money on these days on AI. Like, what can you see? You know, what are you able to see about AI spend within large corporations?

• 6:53Speaker 05

  I mean, it's real. It is dramatically increasing, but in actually interesting ways. And to give you a sense of the panel data that we're looking at to get these insights, we see over 50 billion a year in spend by companies. Some of these are publicly traded. Most of these are private. And often, these tend to be on the bleeding edge. So these can be from AI research labs themselves to farms, to nonprofits, mom and pop shops. And this is across both credit card data as well as bill payment data. So it's a pretty good subset of it. And what we've seen is maybe twofold. First, just in terms of raw and aggregate numbers, an average customer on ramp from the start of 2023 to the end was spending about four times the number of raw dollars on AI-based products.

• 7:38Speaker 05

  And so there's real budget that's starting to go to this in increased ways. And next, the products themselves are starting to actually go from experimental to operational.

• 7:49Joe Weisenthal?

  How can you see that? How do you, what do you see, what do you see in your data that backs that up?

• 7:53Speaker 05

  So the best way we think to know this is that if you looked at an average AI purchase, maybe you purchased some software seat in 2022. There was a 50% chance that within the next month, a customer that bought it would no longer be a customer.

• 8:06Joe Weisenthal?

  Oh, okay.

• 8:07Speaker 05

  They were experimenting with it. In 2023, that had jumped to a 70% chance. In 2024, it's continuing to go higher and we'll release data.

• 8:17Joe Weisenthal?

  Wait, sorry, 50% chance you continued at the next.

• 8:19Speaker 05

  That's right.

• 8:20Joe Weisenthal?

  Okay, got it, got it.

• 8:21Speaker 05

  And so there was a radically higher chance that you were keeping this around. And so it went from tinkering to this is starting to become a real part of engineering processes, sales tools that teams were using to be more productive, to even back office tools to manage accounting, manage expenses. And so I think we're still on the trajectory, best in class of products are going to be in the 90s of percent, but the jump was dramatic in 2023 and '24.

• 8:47Tracy Alloway?

  How granular does your data go? Like, can you see people spending on, I don't know, a basic LLM subscription versus something else?

• 8:57Speaker 05

  Very much so. So the interesting part about what we do is because we automate the expense report process, we can see not just that a company spent on OpenAI, but specifically, was it an API call? Was it a ChatGPT license? And so even among products, you're seeing itemized and SKU level data. And so you can start to get really interesting insights of even in terms of submarkets. One of the emergent themes that people are talking about now was in 2023, there was only one name in AI that mattered, and that was OpenAI. In 2024, suddenly 20% of developer market share was going to Anthropic, which was, I think, at 3% in data in 2023. And so you can start to get very granular of how is this even being used

• 9:42Speaker 05

  across which models are being called. And so it's actually this interesting level of insight that hasn't quite been seen in these markets.

• 9:49Tracy Alloway?

  And then, sorry to focus so much on the data, but how do you actually classify an AI use versus something else? Because I imagine there's a lot of software, for instance, out there now that incorporates some sort of AI component, right? It feels like the Venn diagram of AI and basic tech spending is kind of starting to come together.

• 10:10Speaker 05

  I think you're totally right. And there's a variety of, I would say it was an easier question in 2023.

• 10:17Tracy Alloway?

  Yeah.

• 10:17Speaker 05

  There was only a few strange companies calling themselves AI. Now you see kind of AI washing of companies that are not, would love their stock to pop. But I would say that we tend to classify these based off of kind of self-identification of the companies. These tend to be large language models, labs. These tend to be companies that are pure-play AI products, maybe in 11 labs. If you want to generate an AI digital voice, a cognition or Devon, you can hire an AI developer and these type of tools. I think you're right, though. My sense, and if you talk to too many people in the valley, they'll tell you there will be no company that sells technology in five years that isn't an AI company. And we'll see how the jury goes there, but it's this basis.

• 10:54Joe Weisenthal?

  Two things. First is a statement. I've never coded in my life, so I've made a goal for 2025 to like, use AI to like build an app. And I actually built a really rudimentary app, but it wasn't really doing what I wanted it to do. I'm not going to talk about what it is. It wasn't really doing what I want to do. And then I like fixed the code and I tried to re-upload it and I broke it. So I had an app for about five minutes and then it died. But I'm gonna, this is like my like goal. I really want to like learn to use technology. I want to talk more though about, you know, me when I use, quote, use AI. It's just me like going to chat dot openai.com, just like the most rudimentary user interface.

• 11:31Joe Weisenthal?

  Talk to us more about what you can see on the gap between just someone subscribing to the website versus someone paying for API calls, which I imagine is sort of a deeper level of sophistication, building these models into a workflow in somewhere.

• 11:45Speaker 05

  Oh, for sure. And I actually think this is the most interesting development that if this happens successfully, it unlocks what many of the valley are talking about of historically, technology was software as a service.

• 11:59Joe Weisenthal?

  Yeah.

• 11:59Speaker 05

  You know, you could sell seats to people and would do it. And there's this growing idea of service as software, where suddenly there are workflows where if AI is not just a window you chat into, get a response, but actually at every step of it becomes very, very interesting, where you have kind of end to end products, videos being created, books being done automatically for finance teams, even art kind of getting created. And so I, what I would say is the, the way you can see it mechanically is often the type of license. Usually these are consumer licenses. If you're buying like a chat GPT pro subscription versus there are, let's say enterprise or even kind of developer plans, where at the end of the month, you get an invoice from one of these vendors and you see, okay, there were this many calls to this endpoint.

• 12:43Speaker 05

  These many tokens were ultimately used. And while the specific use, we aggregate and anonymize and don't report down to that level, you can start to see suddenly this is much more similar to how maybe a company would call Amazon Web Services or Microsoft Azure. And this is core compute for some service that is reliant. And the different things about. These graphs is done right. And what all of these companies are betting on is that they're going to grow exponentially and that it's going to be deeply embedded. There are some distinctions. One thing that I think makes this market very different and in some sense more vicious than anything I've ever seen is usually there's this idea of lock-in. You host all your cloud services on one provider and you can't change from one cloud to another

• 13:28Speaker 05

  only for small use cases. But in AI, there's this practice of kind of multiplexing. And so what developers are often doing and sort of why Anthropic came out of nowhere seemingly in a way that wouldn't be possible was people would try some knowledge work or some response on multiple libraries, open source, open AI, see which one's the best. And the winning call starts getting more calls. And so things are just, the markets are moving faster.

• 14:07Speaker 00

  What's next for virtual assets? I'm Jacob Grease, host of Economy Scale, a series exploring growth in key global sectors. And in this episode, we're diving into the digital forces shaping finance. Sean McHugh from Dubai-based Virtual Assets Regulatory Authority has this to say.

• 14:22Speaker 06

  We're seeing a lot of institutional players come in, digital asset, dedicated hedge funds, high-frequency traders and that sort of thing, and folks who service institutional clients.

• 14:31Speaker 00

  As more capital enters the market, adaptive regulation is likely to become increasingly key. Reese Merica-Ripple says.

• 14:39Speaker 07

  A lot of regulators globally would like to keep the industry at arm's length. That's the opposite here. I think also what you'll find is that regulation and frameworks can come out, but some of them are maybe a slight iteration on some 80-year-old policy and not actually in line with what the technology can do and offer. And again, that's very different in Dubai.

• 14:58Speaker 00

  Listen to the full episode wherever you get your podcasts. This series is brought to you by Bloomberg Media Studios and the Dubai Department of Economy and Tourism.

• 15:07Speaker 01

  Underdog already has team picks, player props, money lines, spreads, you name it, they got it. But apparently, somebody over there said, you know what's missing? Screaming like a maniac about a rookie card on your couch. And we love them for that. Because Underdog now has rips. You get a digital pack, starting at just five bucks. You check the live odds, see which grails you could pull, then rip it open to reveal a real, physical card. And the grails are no joke. They have cards up to $125,000 in value, and plenty to pull. From football and basketball to Pokemon, with new limited edition packs dropping regularly. Even cooler, you can decide to keep the card, and Underdog will ship it straight to your home. Or sell it back instantly at 100% of its fair market value.

• 15:51Speaker 01

  And the cash goes right in your account. No listing it online, no stranger lowballing you in the messages. Just rip, reveal, and cash out instantly. Or make some room on your shelf. Download Underdog and get in on the chase. Get five free rips packs when you deposit $5. Real rips, real cards, real cash. Only on Underdog. Unleash your dog. 18+, terms apply. Top grails are extremely rare.

• 16:15Tracy Alloway?

  So the specific companies are anonymized, but can you see stuff on a sectoral level?

• 16:21Joe Weisenthal?

  We both went to the same, but that's what I was thinking too.

• 16:23Tracy Alloway?

  Like, for instance, can you see which industries seem to be ahead in AI? And which ones are perhaps lacking behind?

• 16:30Speaker 05

  We can. And I think that there's a few things that were obvious, and others that have started to jump out in interesting and unexpected ways. The obvious ones of certainly the earliest adopters are technologists like technology. These are engineering offices, startups. A lot of folks doing training are trying and adopting very, very quickly and are nimble to use this. The interesting thing there is that have actually most surprised us, maybe correlate to what you might see and experience. A lot of newsrooms are actually using kind of recording tools. So if you're having a call, notes are taken for you. A lot of sales teams are using this tools that listen to sales calls, take down the notes, suggest the next steps, and start to go into this.

• 17:13Speaker 05

  But you also see, I would say, historically, If companies wanted to cut costs and sort of focus on efficiency, their choice was higher, lower cost labor, often in spaces where some of this can be, if you can get the rules right, things like healthcare, which tends to be late adopters, the rates of use and increase is actually starting to be much faster because when costs and margins are very low, if you can start doing network calls to do more work when there's few two people, these are some interesting emerging industries, but it's still quite early. I would say the the big question that people have out there about the valuations of these companies is still very present. The relative increase is dramatic. For X year over year in raw dollar spend per company is very, very large, but to

• 17:57Speaker 05

  start to pay back some of these large capital cycles, it's going to be early. And so the bet that needs to happen from where we sit is AI can't just be a product SKU that people buy, but I think to pay this back, you probably start to need everyone to use it.

• 18:10Tracy Alloway?

  Actually, this was going to be my other question. So we're talking about AI spend. Can you see the other side of it in the form of, I guess, savings, either by being more efficient or perhaps cutting jobs?

• 18:25Speaker 05

  I would say it's very interesting questions. And I would say generally, as AI has taken place, unemployment has started to come down. And so I think these are very real questions for the long term. And in fact, I actually think one of the biggest misses in 2024 for those promising AI was it was going to be the year of the AI agent was whatever you pitch. You'd have the AI CFO, the AI engineer, the full kind of jobs. And I think that what you start to see is you see slices of things happening. For example, I hope it's not anyone's job in 2024 to do just expense reports, but actually an AI can do your expense reports. It can kind of look at your invoices. It can do kind of the lowest value tasks where that's starting

• 19:10Speaker 05

  to become a present thing in these tools. And so I think in the short run, I actually find generally, we hear from customers, work is getting more interesting in some sense when AI is being adopted. I think long term, there are real questions of might it actually be able to take a workflow end to end. I think practically AI often has very limited context. It gets a question, it can prompt out a response, but doesn't see the rest of the knowledge work. But as it starts being everywhere, it might be possible.

• 19:37Joe Weisenthal?

  I need that end-to-end agentic AI for this app that I'm building because like what's really annoying is it'll be because like I'll like write some code and like Google collab and then it'll be like I'll try to push it to GitHub and then I have to like go find my token and I don't want to like do and then I'm like, wait, where do I find the token in GitHub? So that I can put it in here and that's the stuff that I don't want to, I mean, I guess I have to do it, but I need the AI to just like go find.

• 20:02Tracy Alloway?

  Does that count as coding if the AI is doing the end-to-end?

• 20:05Joe Weisenthal?

  No, like I don't care.

• 20:06Tracy Alloway?

  If you're literally just typing in, design an app.

• 20:08Joe Weisenthal?

  Taking the wheel. I have like, you know, I have some ideas. I'm like trying, but it's like, it's still a little bit annoying, like all these different windows and everything. Okay. But speaking of all that, let's talk about your, what you're seeing in your company. I imagine every engineer that I talk to is like, yes, in 2024 or 2025, they have a window open with their AI and they have their coding window and it's improved their productivity. So like, I assume that's happening in your company, that a lot of code is being written either directly or indirectly or with assistance from AI. Where else? Besides engineering, what actually are you spending money on in terms of AI resources?

• 20:48Speaker 05

  So the three big places, and you nailed it. Number one is in engineering. It is one of the most digital jobs. All code is digital. And so in a strange way, that is actually arguably maybe the first industry that

• 21:00Joe Weisenthal?

  Yeah.

• 21:00Speaker 05

  is closest to AI. Second is sales and growth. Ramp beyond having a large amount of spend data is one of the fastest growing companies or startups in history. And part of what's allowed us to do this is the average salesperson at Ramp is about four times as productive as an ex closest competitor. A lot of that is a heavy use of AI to automate aspects and low.

• 21:20Joe Weisenthal?

  Specifically, let's talk specifically. What do they do? Yeah. Someone who makes a sales call. So what are they using AI for?

• 21:26Speaker 05

  So one of the most important functions and the first role for anyone going into sales is this role called a sales development rep. And the job of that person is book meetings. Find people out there who. Maybe are doing expense reports the old way and let's bring them in. And if you kind of decompose what that task actually is, it's first asking, who are these businesses? What is a relevant moment? Have they just raised funds? Have they just hired someone on their finance team? Maybe they're posting an open role in some needs. There are these signals out there in the world. Then you notice, okay, you've got your lead list assembled. You have to go write them. Maybe you have a junior person just out of college going and trying to guess. What's this person's email? What's this person's phone number? How do I get a mailer in front or how do I knock on their door?

• 22:11Speaker 05

  Then they write kind of the message, what's going to be compelling? There's all these little steps. What makes a human seller great? And what makes sales interesting is the genuine human connection. Someone who can go deep, understand all the context and actually close that great sale. And yet, if you looked at the task in 2022, how most people were spending their time was things that algorithms are great at. Finding people's email address, testing which copy will ultimately work better, detecting across vast swaths of data, what's the signal in the noise? And so effectively, part of what's powering this level of growth is a broad set of AI tools which do exactly that.

• 22:51Speaker 05

  Where AI is finding the person's email, AI is detecting these signals for intent. Sending the message in the job of an entry level salesperson now is the majority to respond to interest, to close people on the call. And so that's one example. We can go through through several kind of throughout the sales cycle, but it's changed the role to be much more interesting.

• 23:14Tracy Alloway?

  I'm getting flashbacks to Glengarry Glen Ross. Didn't companies used to buy lead lists as well?

• 23:21Joe Weisenthal?

  They still sell them.

• 23:21Speaker 05

  They still do.

• 23:22Tracy Alloway?

  Yeah, right. So I imagine if you can basically build your own lead generator, you would save some money as well.

• 23:28Speaker 05

  That's exactly right. But the interesting thing is just the speed moves faster. There's suddenly signals of intent of maybe there's an IP that you can back into. This is a company has gone to your site five times today. Maybe that's a low value list. It's the C list. It's not, it's not the A grade list. But with kind of a modern stack that's sifting across these signals, you can get more interesting. And so I think there's things like that. The other big thing that has just transformed the job is there's just so much more noise than there is signal out there in the world. So if you're for manager and you're trying to help a, you know, 22 year old new in their career, get better at sales. It's just too many hours of calls to listen to across your whole team to do that.

• 24:11Speaker 05

  The large language model has no problem listening to a thousand years of calls in a single day, more than any human can. And so suddenly, I actually think one of the more interesting stories and lessons I learned about this actually came from a hundred plus year old credit card company where I was first skeptical when I heard the story. The executive was explaining how they stopped checking net promoter scores. And I'm like, wow, they must have gotten that bad. But the truth was something more interesting.

• 24:36Joe Weisenthal?

  What's a net promoter?

• 24:37Speaker 05

  So a net promoter score is this process of often you sample. So let's say there's 10,000 people who call in with customer support and you ask them at the end of the call from zero to 10.

• 24:47Joe Weisenthal?

  Okay, yeah.

• 24:48Speaker 05

  How happy are you with the service? And often in these surveys, you ask a small sample and you can see in aggregate, how good are your agents? The response was much more interesting. What they started doing was applying a large language model to listen to all calls simultaneously. And they could apply sentiment analysis from the tone of a voice, how happy were customers or not? And what the algorithm started doing was routing calls from customers increasingly to people who did a great job that made customers happy.

• 25:18Tracy Alloway?

  This is always the case. If you're good at your job, you get more.

• 25:20Speaker 05

  That's the reward.

• 25:23Speaker 05

  But it's a real way. Suddenly, like you can actually tell from every customer, how are they doing and actually get more output per unit of input? It's a real use case. I don't know how you'd do a few years ago.

• 25:34Tracy Alloway?

  And I'm sorry, you were talking about where you're deploying AI and you had engineering and sales and we didn't get to the third one.

• 25:42Speaker 05

  Oh, sure. Yeah, we'll see if it works. Maybe most avant garde would actually be kind of in growth and marketing. Of course, there's customer support and other areas that the people have talked about. But I actually think there's a real case that marketing and growth is becoming a technical function. People thought that art would be one of the last use cases and suddenly you can create images, videos, interesting things. And you know, now in a world where you can go and start to do this, you can see based on intent, on conversion, can you start to combine the mathematical function of what really works with people to creating beautiful, striking, interesting images on demand. And so the jury is still most out with that one. But it's been real. I think early tests are promising. And I think that the early example of this that to me, in some ways, kind of inspired

• 26:25Speaker 05

  this was maybe twofold. One, Amazon, they had kind of the editorial and the personalization team. It used to be that folks at Amazon actually wrote and recommended you bought, you know, here's a newsletter of all the things that we have. And it was like the Sears Robux catalog. Eventually it became you bought this, you might be interested in that. And eventually things went out. And so you may be able to do that, not just with clustering of items, but actually understanding people and making beautiful things. And I think of inspirations in New York, like Andy Warhol, his factory where every day they'd make something new. Things were incredibly striking and had this method of of new levels of art, thinking about being incredibly generative. And I think that every industry in some sense can be improved and augmented and made more interesting, actually through kind of the

• 27:08Speaker 05

  just the leverage that these tools can bring.

• 27:11Joe Weisenthal?

  I'm going to be cynical for a second, and I'm going to press you a little bit further on the question of AI and sales, because I certainly take your point about searching for signal. What companies might be in a position suddenly where they're looking to upgrade their expense management platform? Or what might be the person who best to call? Some of that, I imagine, Stuff that someone was selling via Salesforce for a while and calling it five years ago, machine learning or 10 years ago, machine learning. Obviously, what is technically AI is one of these things that people argue over to get a better multiple, et cetera. But for the purposes of what many people in the stock market are excited about, a lot

• 27:52Joe Weisenthal?

  of it's the sort of, you know, the post 2022 generative AI that somehow comes in the lab and was inferred on an Nvidia chip or something like that. Say more about sales and what are, what other things that you can do in sales today in 2025 that you could not have done in 2021?

• 28:11Speaker 05

  I think that probably the most exciting area has to do with reasoning. When I think about a lot of classic machine learning, which is still deeply important, it's often around correlation of certain variables and prediction in a very narrow sense. And so maybe the first phase is a machine learning is prediction of what comes next. You know, my frame around a lot of this is now there's generation based on what come next, what do you create? And I think the most interesting, when you think about kind of the 01 models, 03 and the new reasoning models where it's thinking and can think multiple steps ahead, some of the techniques that made AlphaGo so good at what it's doing. It's those types of work. And so you're exactly right. Some of this is actually just good data infrastructure and prediction.

• 28:55Speaker 05

  But when you start to fuse that with based off of these signals, what maybe should we write based off of the context of the calls and the usage of this data? How do we follow up and orchestrate it across multiple steps? That's where I think generative AI starts to get really interesting in these boring use cases like expense management and saving people time and money where it starts to get very useful.

• 29:32Speaker 01

  Games on a break? Don't doom scroll. Try rips on Underdog. Buy a digital pack, rip it open, and reveal a real, physical card. Football, basketball, even Pokémon. Packs start at just $5. Ship it straight to your door or sell it back instantly at full market value. Max value up to $125,000. Download Underdog and get in on the chase. Deposit $5, get 5 free packs. Real rips, real cards, real cash. Only on Underdog. Unleash your dog. 18+, terms apply. Top grails are extremely rare.

• 30:03Speaker 02

  This is the Bloomberg Tech Minute, brought to you by ChatGPT. I'm Carol Muster. Millions of Americans are taking peptides, short strings of amino acids that can act as biological signals for cells. Bloomberg's Anika Inamputi reports peptides may be exploring their next frontier: the brain. People are taking peptides for mental clarity and cognitive benefits and to gain a competitive edge or stave off the effects of aging. Both insulin and semaglutide are approved by the U.S. Food and Drug Administration. Because importing or selling unapproved peptides for human use is not legal in the U.S., many people turn to online forums for guidance and overseas vendors with questionable credentials to obtain them. Unapproved peptides, which are marketed as research chemicals, have been circulating

• 30:47Speaker 02

  in what's become a billion-dollar gray market amid the rise of do-it-yourself medicine as people increasingly turn to social media or artificial intelligence for medical help. That's the Bloomberg Tech Minute, brought to you by ChatGPT. Get started at ChatGPT.com.

• 31:03Tracy Alloway?

  Can you talk a little bit more about what data exactly you're scraping to infer those specific signals? Like, what do you have access to and what do you find most useful?

• 31:12Speaker 05

  Yeah. So first, some of this is just, let's take a use case of, let's say, an account manager. They might be overseeing hundreds of individual accounts at Ramp, and their goal is to get back on the phone and make sure people are getting value out of it. We want to save your business time and money. We want to automate accounting, but is it set up that way? Are you seeing these terms of use cases? And so some of this is going to be internal use cases based off of spend that you want. Wanted to bring over versus which you actually spent. How is that going? Call and log data, large language models can remember all calls, all notes, what people kind of committed to you by what dates and do that. And so when an account manager gets on the phone, they can go and have all the right contacts in front of them that they didn't need to go and spend hours the night before kind of creating, but it's pulled up and pulled in terms of what's most useful.

• 31:57Speaker 05

  You next to me have interactive data can pull through based on the website itself. Where their flows were appeared, people just got stuck and confused. Maybe they wanted to go close the books and it seems like too many steps and they kind of paused. We can kind of.

• 32:10Tracy Alloway?

  So you're tracking like actual physical movements on the website through beacons, I assume.

• 32:15Speaker 05

  Yeah, yeah. So you can do things like that, coupled with even external data as well. Sometimes a company could be doing really well. Announce the fundraise, you know, need to expand and make sure, hey, we've seen this good news. We want to make sure that we're expanding with you. And so I think that any one individual point a person can do, but often the fullness of what does it take to really understand and be a partner with how are one to thousands of employees at a single individual customer? How are they actually doing? So we can be a more useful partner is often where this comes together. Does that make sense?

• 32:48Tracy Alloway?

  It does, although I have, maybe this is a weird question, but like, how does the system actually generate its suggestions? So if I'm a salesperson and something like the system spots a lead of some sort and it thinks you should get in touch with this person because of whatever reason, like, how does that message actually get to me?

• 33:09Speaker 05

  Yeah, and I think that's exactly, in some sense, probably the million dollar question of which SaaS companies are going to do really well. And I think even the story of technology now is you have different people with different aspects. Some are in the browser, some have this sales tool, some have this data tool, some are a data warehouse, some are a training tool. Where does it show up? And I can tell you, we leave it ultimately in the hands of our sales team and the growth team building for them of whatever works. You are free to pick, given the period of change. But for them, often, like, I'll tell you, one of the more useful tools for that use case for account managers is a company called Rocks. I think it's like. ROX.com. They're less than a year old company, but effectively are pulling data from Salesforce,

• 33:51Speaker 05

  usage data, internal level data analytics, and append notes for an account manager's calendar prior to meeting. Here's the core things to know. Here's links. And if you want to poll, so you can get things at a glance in your calendar, and you can actually poll from the website itself. Here's more data to go see it. So it's, they're trying to become a bit of a mission command for sales. Of course, Salesforce is trying to do these things, too. To, in engineering, there's tools like Cursor and Devin have different bets. Cursor is like GitHub Copilot, sort of won the love of many developers, where as you're coding, it's like a better auto complete that you just say, I want to build this app, and it can go ahead, it can audit lines of

• 34:32Speaker 05

  codes and knows your repo. There's stranger bets like Devin and cognition, where the form factor of that is, it is meant to be a digital AI engineer. And you can tell Devin, I want to build this app, go research these websites, make it look like this, come back to me when you have questions.

• 34:48Joe Weisenthal?

  Should I do that for my app?

• 34:51Speaker 05

  Try it.

• 34:52Joe Weisenthal?

  Because like, I was talking to someone, I forget who it was, maybe it was even someone from one of these companies, and they were talking about like, different approaches to some of themselves, including ones where like, the AI basically controlled the mouse, controlled the cursor, and clicked on websites and read websites. So we think of API calls, but there's a different model where it's just like, you're scanning the website like a human is, right?

• 35:14Speaker 05

  You are exactly right. And this is one of the strangest and most interesting things about the time that we live in, and that computers can kind of think, they can kind of see, they can kind of hear and process different levels of data. And I think that the general story of computing is increasing levels of abstraction. Back 80 years ago, people were writing machine codes. It was one, zero, binaries. And over time, it went to C, to Python, which were increasingly higher order compromises between the language you and I speak.

• 35:44Joe Weisenthal?

  Separate it from binary, like did layers and layers of a binary. Yeah.

• 35:48Speaker 05

  And there are many people who would make the argument, I think convincingly so, that the next programming language, in fact, Is English.

• 35:55Joe Weisenthal?

  Language, yeah.

• 35:56Speaker 05

  And what you see. And so I even think of now what many people talk about, one of these coming battlegrounds is actually, what are you seeing on your computer? And the best interface is not chat, but actually it's just a large language model that sees everything you're doing on your computer and it can predict what's next. Who knows?

• 36:15Joe Weisenthal?

  It's nice. Well, when I keep getting error messages in my Google collab, I just take screenshots and then I upload them to chat GPT. And I say, what's this error message mean?

• 36:23Speaker 05

  Yeah.

• 36:23Joe Weisenthal?

  And then it tells me, one thing I've wondered about, and you know, people are very anxious about what are the jobs of the future? And all these things and what professions are going to get disrupted away. And you know, we could, people could speculate on this forever. But one thing I've like wondered about is like, there are certain jobs where to be good at them, you have to have probably like sacrificed years of your life and like not gone to parties and not had friends and you know, not had sleep. Some mean jobs. Because like

• 36:51Speaker 05

  The true artists.

• 36:52Joe Weisenthal?

  Right? No, or the technical skill to require just took years and years and years, et cetera. And so you had to be the type of person that was willing to sacrifice a lot to get good at them or to build up that technical skill. Right? Those hours. And I'm wondering like if AI is going to sort of cut into the jobs where a major part of getting good at it was this sort of being willing to like sacrifice being a normal person because the AI could just do thousand, doesn't need to sacrifice anything. It's just a computer and that it would benefit the people like, you know, the low rounded people who I bring. People that want to have fun. Yeah, you bring some IQ to work, but you also bring some EQ. People that have like emotional IQ. Yeah, yeah, exactly. This is what I'm wondering about.

• 37:35Speaker 05

  I think this is exactly the right line of questions and I think is going to be a real one to confront. And I think in some sense, I think even probably listeners of this podcast fall into this group of like very curious people and people who know how to ask interesting questions, keep going down it and create things, I think will do very well in the future.

• 37:53Joe Weisenthal?

  Thank you.

• 37:54Speaker 05

  Yeah. It's all going to be okay. Don't worry. No, but I think things are going to change a lot. Like even in thinking about coming here today, like I was curious of at the turn of the century in 1900, I think in the United States, 40% of all jobs were in farming. You know, today, less than 1% of all jobs are in farming, and I think things are okay currently. But I think that the nature of jobs is going to change, probably in ways that will be very hard for people, just as in 1900 to now, to predict. Probably the same for us and what it may look like in 50 to 100 years.

• 38:27Tracy Alloway?

  How are you or your clients handling privacy and legal slash copyright concerns with some of these platforms?

• 38:37Speaker 05

  Yeah. Well, first of all, like, we have a bit of a simpler time on at least copyright in that we're not in the business of generation, of how do you go and create new art and images? And I think those are real and present questions, especially for content generation. A lot of what we're doing today is you've made this card transaction. You've texted back 30 seconds later, here's a photo of the receipt. And then we can match, okay, here's what the memo should be. Here's what the accounting category should be, and you're done. And so a lot of this is process automation and workflow automation. And there's an interesting value we can bring. And probably the most interesting part of our business, we need to think a lot about this, is on this area of price intelligence. Now, as a consumer, you can go onto Zillow and know, here's what your home maybe is worth.

• 39:22Speaker 05

  You could go on TrueCar and get a sense of what should you pay for this car based on lots of aggregate anonymized data? And for businesses, it's very useful to know, what is this business paying down the street for this supply? Or for this Salesforce license? What are people paying? And so in that part of our business, we actually do want to be able to go to customers and say, you know, here's where your pricing compares to the rest of the market. And we want to help you negotiate for lower prices, and we think it's really good. The way you get there, and I think probably the core of our strategy around this is aggregated anonymized, is really the part that needs to come. Or if it's individual data, that really sits on ramp and is for the purpose of providing service to you. And if we share things broadly, it's give get. If you want to see pricing data, you need to share in it.

• 40:07Speaker 05

  But we need to have enough data to effectively anonymize, like, what's coming out there. And so we think it's very useful for finance teams, for business owners, to help them pay less. It's part of what helps an average customer cut expenses by about 5% per year. But maybe less good news for people who try to discriminate against our customers.

• 40:26Joe Weisenthal?

  We are in an age of cracking down on spending and cracking down on wasteful spending in DC, for example, and people are talking about Doge, etc. And if we're going to really move the dial on spending, it's not going to be wasteful spending, it's going to be, we're going to have to like actually have different priorities. Nonetheless, cracking down on waste seems good. From the perspective of an expense management platform, what are some fingerprints of wasteful spending that you see? Can you see in the companies and like see the fingerprints of waste? Do you have any advice or things that you look for when you, if you wanted to hunt waste in spending?

• 41:01Speaker 05

  I'm happy that people are thinking about this in a real way because often there's an obsession of how do you spend more? But if you want to make a change, it actually comes.

• 41:11Joe Weisenthal?

  Other than putting the entire government on rent. So other than that, I assume you support that, but what's the next thing you have to do?

• 41:18Speaker 05

  Look, maybe one of the founding fathers, Ben Franklin, I think was known for saying a penny saved is a penny earned. And if you look at reasonably efficient organizations like an average American company, they have a profit margin of about eight and a half percent. So mathematically, a penny saved is actually 12 earned is the same. And I think about organizations like the government, which, look, I have a lot of empathy. They have a lot more complex constituencies and needs than, you know, a profit-making entity. But the exercise of cutting costs has not been taken seriously for a very long time. And it's led to very different behaviors. If you want to sell to the government today, often with a selection criteria is can you

• 41:59Speaker 05

  last through a one to two-year request for proposal, an RFP and process, which is very different than how most companies and people select for things. What has the most value? What's the lowest cost? What can I try and see it is done? Next, the other very counterintuitive thing that's interesting about the government is, you know, you would think that as one of the largest buyers of anything in the world, we would get that discount for all taxpayers. If you're buying a million licenses for a piece of software.

• 42:27Joe Weisenthal?

  Volume license.

• 42:28Speaker 05

  But it's not that way. The typical way that government buys is to pay sticker price and to not have discounts. And so you see, not only are different agencies paying very different prices for the same sets of goods, but you don't see normal, common sense things of, hey, we should have a group discount for people in buying. To last, a lot of the tools, because procurement cycles are 15 years. You can't break contracts. You're paying full rate. So these companies are going to sue you if you try to go and do this. You start seeing some crazy things in terms of the actual tools themselves. The spend management architecture that the government is using was primarily selected in the early 2000s.

• 43:09Speaker 05

  And so whereas the private market today can tap a card, their expense report is done for them, their books are kept for them. The result is that you have an incredible amount of waste of people's time, of really hardworking people, in some cases, actually spending most of their time trudging through the bureaucracy of old tools that don't work to each other. To the most shocking, I would say, is like, you don't have to look hard. There's several friends at different agencies who, in trying to learn and understand some of these DOGE efforts, a shocking thing that I learned is that at multiple agencies, four hours per night, email is just shut down. You can't send, you can't receive. It's crazy, right?

• 43:44Joe Weisenthal?

  I've seen that like government websites have like a time.

• 43:47Speaker 05

  That's exactly right, because they're using old private servers, contracts from 30 years ago. And so if you want to talk about, you know, true efficiency, it's nowhere close to the leading edge. And, well, I think you're exactly right. If you want to really make a dent in the budget, you have to talk entitlements, you have to talk debt service and what that's going to look like. But if you want to go to this next level, great tools that prevent wasting of time, that automate auditing of records, so you don't have a Department of Defense that's failing seven audits in a row and can't track where spending is going, but actually it's all digital, it's all tied to it. I think you need to get serious about allowing people to pick best tools for the problems.

• 44:26Joe Weisenthal?

  Eric Glyman, thank you so much for coming on Odd Lots. I'm really glad we made this happen.

• 44:30Speaker 05

  I really appreciate it.

• 44:44Joe Weisenthal?

  Tracy, I really liked that episode. There's all kinds of AI tools I need to now dive into. Because now I'm going to be the person who subscribed to nine different things.

• 44:53Tracy Alloway?

  Well, that's everyone, right?

• 44:54Joe Weisenthal?

  Yeah, I know. What am I subscribed? I'm subscribed to at least like three, probably more right now. Now that I gotta like dive into these specialized coding tools. But no, that was really fun.

• 45:05Tracy Alloway?

  Have I told you the story before about my coding class in high school?

• 45:10Joe Weisenthal?

  Say more.

• 45:11Tracy Alloway?

  So actually, it was just a basic IT class, but as part of the class, we had to program our own little application. And this was like in the early 2000s. So it was all very rudimentary. But our teacher commissioned us to do this and I built a fortune cookie program where you clicked a button. Well, the button looked like a cookie and it gave you your fortune, blah, blah, blah, blah. And at the end of the assignment, everyone turns in their program to the teacher and he made everyone sign a contract, giving away the rights, the licensing rights to him. And he said he did it to teach us all a lesson about copyright and how your work is rarely your own

• 45:55Tracy Alloway?

  if you're working for a big corporation, which is why I asked that copyright question. Because I still think about that to this day. So it was a good lesson.

• 46:03Joe Weisenthal?

  That's extremely funny. But no, there was a bunch in there that I thought was interesting. So one, you know, the idea that you can just like track like what percentage of people like pay for something one month, then also the next month. Also this idea, and like we gotta keep like coming back to it, the lack of lock-in for some of these models, right? We're really used to there just being one winner in search, one winner in social networking, one winner in e-commerce, so forth, one winner in photo sharing. So it's really interesting to think about like all this money going to models like A, where it's like really easy to move from one to the other, a simple project. B, where there's open source competitors that maybe are just as good. That seems like a pretty big deal right there.

• 46:42Tracy Alloway?

  Yeah. And I guess the big question is, will there eventually be some sort of winner that turns? Out to be better at it than everyone else? Or is there going to be space for that sort of specialized? Either, you know, specialized use case or the specialized like actual interface for different jobs? Yeah, that's interesting. We talked a little bit about the importance of interfaces.

• 47:04Joe Weisenthal?

  Totally.

• 47:04Tracy Alloway?

  Anyway, shall we leave it there?

• 47:05Joe Weisenthal?

  Let's leave it there.

• 47:06Tracy Alloway?

  This has been another episode of the Odd Lots Podcast. I'm Tracy Alloway. You can follow me at Tracy Alloway.

• 47:12Joe Weisenthal?

  And I'm Joe Weisenthal. You can follow me at the stalwart. Follow our guest Eric Glyman. He's at E. Glyman. Follow our producers, Carmen Rodriguez at Carmen Arman, Dashell Bennett at Dashbot and Kale Brooks at Kale Brooks. For more Odd Lots content, go to Bloomberg.com/oddlots, where we have transcripts of blog and a newsletter. And you can chat about all of these topics 24/7 in our Discord, discord.gg/oddlots.

• 47:38Tracy Alloway?

  And if you enjoy Odd Lots, if you like it when we talk about how much companies are actually spending on AI, then please leave us a positive review on your favorite podcast platform. And remember, if you are a Bloomberg subscriber, you can listen to all of our episodes absolutely ad free. All you need to do is find the Bloomberg channel on Apple Podcasts and follow the instructions there. Thanks for listening.

• 48:32Speaker 01

  Games on a break? Don't doom scroll. Try rips on underdog. Buy a digital pack, rip it open and reveal a real physical card. Football, basketball, even Pokemon. Packs start at just $5. Ship it straight to your door or sell it back instantly at full market value. Max value up to $125,000. Download Underdog and get in on the chase. Deposit $5, get five free packs. Real rips, real cards, real cash, only on Underdog. Unleash your dog. 18+, terms apply. Top grails are extremely rare.