[00:00:00]
Eric: How profoundly disappointing would it be though, if you truly made the best version that could ever be done in the history of the world of any particular thing, and there was no surpassing it. And that was it. Welcome to the Logan Bartlett Show. On this episode, what you're going to hear is a conversation I had with co founder and CEO of RAMP, Eric Glyman.
Eric and I touch on a number of topics, including artificial intelligence and how RAMP is using that both internally for their own employees, as well as externally for their customers. No person has the time to listen to 50, 000 sales calls. I'll A large language model does. And so it was a Slack bot we built in a weekend where you can ask Kobe why the customers picked RAMP over competitor REX.
We also had a long discussion about motivations. I feel lucky to be able to have the transparent relationship with him that I do and for him to open up and share about what drives him and his motivations at RAMP. A lot of times organizations leave out and frankly I learned everyone is the hero of their own story.
Understanding people's motivations, where they want to be in years and periodically referencing back. You'll hear that conversation here now.
Logan: Eric, thanks for doing this.
Eric: Good to see
Logan: Good. Uh, good to see you.
Logan: I guess it's, it's probably been two, two years since we did the first episode of this. I don't know. We can probably look it
Eric: mean, you
Logan: like an easy thing to find out. Probably something I should have known. Yeah. I went back and listened to it. And I, uh, I think when we were, when we were talking, like the market had started to turn and, uh, there were some signs of like softness in there.
I remember. At the, at the time we invested in ramp, uh, we did a big round in December of 21. It was sort of right. Right. That was, uh, the round was announced maybe a March or April ish.
Eric: for far before, but I think first invested, I think, in January of that year.
Logan: 21. Yeah, yeah. But I remember of all the companies when we were sitting down and we didn't know exactly how the public markets [00:02:00] were going to play out, how the private markets are going to play out, how spend was going to play out within your ramps business. I, I, uh, I remember we were sitting there and it was, um, you had built so much so quickly at that point in time, but we were also And listening back and sort of preparing for this, it was definitely more uncertain of how the market was going to play out and you know, what was going to be beneficial to ramp versus not.
But it's amazing how you guys have continued to execute over the last couple of years
Eric: It's, um, look, the support has been amazing. Um, yeah, I mean, look at the, the change in valuations for companies was dramatic and the rise of interest rates is very costly for businesses and, um, no doubt like years ago when we were starting out. Um, the fact that companies could just raise money easily and spend helped, but it turned out to be a huge accelerant for our business because our core value proposition of helping people spend less and, you know, for the vast majority of people, like the free product that pays you to use it, uh, was really in vogue and wasn't a line item to be eliminated like so many SAS, um, pieces of software.
And. You know, [00:03:00] we've been able to just keep up, um, you know, triple digit growth, uh, each year and, and just grow at times when, when most companies, I think we're, uh, really struggling. And so, you know, we're working through, but it's, it's been a lot of fun, frankly, to have a product that went from a nice to have the
Logan: product that went from a nice to have this thing going. Pullback, uh, was actually an interesting thing to say. I don't know if I knew that in my heart of heart in June of 22 or whatever. I'm not sure I, uh, was fully appreciating the puts and takes of that. Uh, you guys, I think at this point are our biggest investment in the history of red point.
And so I, uh, You know, hopefully you, uh, you didn't feel any of the stress from me, but maybe in the, the doldrums of, uh, 22, I, uh, I was a little anxious about which way that was going to play out, but it's really impressive to see.
Eric: It's uh, look, I think a little bit of stress is exactly what you want.
Logan: that's right. I don't want to put it on you though. It's the problem. Yeah, yeah.
Eric: put it, put it on [00:04:00] us makes us better.
Logan: Yeah. Yeah.
Logan: Well, um, so, so a, uh, an invoke thing to, to start with that, um, I know you've spent a lot of time thinking about is artificial intelligence in general and how to operationalize that, uh, and both, I guess there's, there's stuff internally around it and how you can be more efficient, uh, with your employees and some of the stuff that you can do and then, um, externally and what you can provide to your, customers.
At this moment in time, as we sit here, June, July of, of 24. Um, how do you think about AI and the value prop for a business like ramp accessible through an API
Eric: a couple of things. I mean, first of all, I'll say what everyone is. I think it's heard and same, but I think it's worth repeating is it's so profound and true. I think that AI is In what's happened over the past year represents the biggest shift of productivity certainly in my my lifetime. Um, uh, and I think, um, and when you have functionally human level, uh, and in certain cases superhuman level reasoning that is accessible through an API call that any [00:05:00] business can go and access, I think you need to think about, um, how is this going to change the world and what does it mean for my business?
And I think there's a lot of, um, truisms people could rely on for the past 20, 30 years in businesses that are really going to shift. You know, if you were a software business. It was about how many features did you have, you know, how, how much are, uh, your customers locked into your data model in a world where, um, you know, give a five, 10, maybe two years, depending on, on, on who's right.
But I don't like it's, it's, it's outside of that. Um, you have software engineering functionally where, you know, I think, uh, uh, what used to take months, uh, will take days, if, if not hours, uh, it means you're, you're not defended based on the features that you have. Uh, vendor lock in is not going to be there and, uh, and so a lot of the typical moats that software businesses develop are going to erode, um, and I also think the model of, of selling people seats, um, uh, as a way to monetize, uh, isn't going to be perfectly sloped to a reality when you actually don't need people to do, um, a lot of, of jobs and, and so you actually don't want to sell people more seats, you want to drive more outcomes. [00:06:00] Unpacking it, um, A little bit. First, we'll start with like ramps general premise, which is, uh, we think we should get paid based on the outcomes we deliver to companies. Um, our founding value proposition has been, we are going to help you spend less money in time. Uh, and we are going to measure how much money in time we've saved your company.
Uh, when we launched in February of 2020, uh, we helped the average company save about 2 percent on their expenses. And we thought that was great. That was more than what, uh, cashback, um, or rewards, um, really could be for, for most businesses. Today we help the average business save 5%. You know, between hard dollar cost savings and labor costs that companies don't need to spend anymore.
And we think this is only going to go up and up and up. And so I think in a world where AI is not just assisting but is actually driving outcomes for companies, I think companies want to position themselves to be selling an outcome, selling a service like that. That's one.
Logan: Did you, um, did you think about selling seats ever in the early days? That was the, the model of Concur, right? With seat [00:07:00] pace pricing, I assume Expensify, I actually don't know. I know they've kind of moved since then.
So was that ever a consideration? Was the status quo in the industry for some of the other people that you were looking at?
Eric: It's um, That's one. So first, I still think there's a place for selling seats, but the big thing when you think about ramp, well, very early on, um, I think in our very first board deck, we, we tried to really, we had no revenue, uh, and we wanted to, uh, on a first principles basis reason through what is our business model.
Logan: sick,
Eric: Telling us how does it actually work? Um, and if you look at most car businesses and FinTech businesses, they're at their core consumption based businesses. Uh, the more you use, um, uh, the more ultimately, um, you know, we are able to, to monetize as a company. Uh, and we, if we do our job rates, the more value we're going to create for, for our customers.
Um, and so the, I'll, I'll, I'll spare you all the details of the business equation, but at the core of it, it was. If you want to make revenue, there's a few core variables. How much purchase volume are customers putting with you? What's [00:08:00] your take rate on that volume? Uh, And what is the cost of funding it? Um, and, uh, it turned out if you studied every single variable, there's only one thing that mattered, which is purchase volume.
It turns out if you make more purchase volume, you make more revenue. If you make more purchase volume, you keep more interchange. If you make more purchase volume, you have a lower cost of funding at the bank. You're better at preventing fraud and losses. So there's one variable that drove the whole thing.
Um, which is how do you get people to want to use your product, um, and spend with you? Um, and so our whole view was, look, we want, you know, um, this is a business that fundamentally at scale gets much better and better. Um, how do you get to scale as quickly as possible? And our hypothesis was, well, if you just make spending with us just simply better than spending anywhere else.
Rationally, people should want to spend more with you. Um, I think how this gets expressed in, in real life is, you know, if you spend on ramp today, um, you have very simple cash back and we show you ways to cut spend constantly. And so more spend that's put on ramp, you're able to better benefit. You, [00:09:00] you find that you actually are spending less over, over time.
More, um, spend you put the ramp. There's more expense report. You're able to automate more accounting. We're able to automate more. We're able to, you know, to show you ways to better benchmark and run your business better. Uh, and so all we simply try to do is a great, we have a simple model. Um, let's get more purchase volume and the way we're going to do this is create more and more customer value.
So internally, all our product is think our product team is thinking about is, um, you know, how much money and time are we saving customers and how do we radically increase that? Uh, and if we do that, it should.
Logan: uh,
Eric: Fairly directly drive the output variable we're trying to drive, which is purchase volume. And then over time, we should be able to expand margin, which is what's played out.
It's been a big part of why is as quickly as we've grown. Our bottom line has grown dramatically faster over the past few years.
Logan: So, so you were set up well from the start, uh, that you were already focused on outcomes and you didn't have sort of an orthogonal pricing mechanism that was, um, at odds or, or potentially conflicting with the, the value prop.
So, um, I, [00:10:00] I guess as you thought about ai, uh, it was a first principles thing within the business, but we've, we've reached this new vector of, uh, step function change, I think, in some of the reasoning.
Logan: And so once chat GPT came out and that moment was kind of unlocked with LLMs and all that. What did you do?
How did you operationalize it within the business?
Eric: So, um, a couple of sets of things mean first, um, I think with anything you want to develop taste about how, um, you know, a breakthrough ultimately works before you go and productize it. Uh, I think that the first wave of products that came out after the chat GPT moment were Frankly, in hindsight, very bizarre, right?
Like, I've never met a single customer who said, you know, my, I just would love to chat with my bank account. Um, you know, I want to chat with my, my HR records. I want to chat with, like, it just was sort of crazy in hindsight. People did this, but, you know, people were trying to, you know, slap on, you know, my company is the AI company of X and just kind of went with it.
Uh, and so at the beginning, rather [00:11:00] than going and trying to go and say, here's, uh, ramp AI and all the ways that we could do this, we said, let's actually use this internally. Um, um, let's understand the ways that we can, uh, um, reach more companies more efficiently, um, win more deals, learn faster, uh, from customers, uh, in improved processes.
So really, for the first year, a lot of what we were investing in was this applied AI, uh, Uh, team, which is a horizontal team, um, that is distributed out some of the most common use cases today that we use, uh, AI, um, and I think have been just transformative for us is, um, a very heavy, you know, a heavy portion of how we grow, um, is functionally through, um, Um, an AI based sales process where functionally, um, you know, the lead scoring, the outreach, the iteration on copy, um, you know, is fairly programmatically done.
Um, uh, and we focus, um, really the high value work, um, people's time on responding to customers who are interested. Um, we spend, um, uh, you know, every sales call, [00:12:00] um, between us and a, and a customer we record initially was for quality, um, you know, assurance and training and helping our reps do better. You know, but we look back and realize we had tens of thousands of sales calls that, um, you know, we're digitally recorded.
We had transcripts. And so, uh, internally, uh, you know, about a year ago, um, we, we, we started building a, um, there's an ad bot called Toby, which. Toby, you know, no one on the team, uh, no person has a time to listen to 50, 000 sales calls, but a large language model does. Um, and so it was a slack bot we built in a weekend where you can ask, you know, Toby, why does, um, you know, to customers pick ramp over competitor X, right?
And SDR can ask that kind of a question. Or, um, what do customers think about X new feature or, you know, reporting or some need or something that we're researching and want to do? Um, and Toby is able to go, um, frankly. Within seconds, um, process all this information, query and pull out the right transcripts and show you, um, what a cross.
Um, you know, hundreds [00:13:00] to thousands in certain cases, roll transcripts, the summarized insights. And if you want to go deeper, you can go deep into this. And these are the types of things that are, um, you know, have very little, you know, you could be selling widgets. It doesn't have to be, um, spend management that helps companies become more efficient.
It could be any nature of business. And we said, let's actually apply this. And you start to develop a taste of, of where large language models are well suited, uh, to problems. I think they're much better at, you know, incredible at summarization, I think at times in generation. Um, uh, and as we developed taste for that, we started developing these hypotheses of where it would be deeply powerful in the RAMP product.
And, um, happy to talk, talk
Logan: Yeah, no, I, I definitely, I definitely want to there, um, on the SDR point and then, uh, on the Toby point as well, I guess the, the SDR, I assume a lot of the, the tooling that you use is off the shelf, uh, stuff, be it, uh, outreach or gong or, or whatever it is, uh, but then I assume you've built a lot of things on top of that.
Can you maybe talk about? [00:14:00] Uh, who was responsible for these things? How much, there's all these AI SDR companies that are coming out there. Do you think these are going to be commercialized products? And you guys are just at the early edges of this market right now. And so you home grew a lot of this stuff, but ultimately it's an opportunity for another company to do it.
Eric: think there are opportunities to build for sure, um, I think a lot of the, the state of the art comes with the ability to customize, uh, to your own business, um, uh, in, in, in ensure that you're, you're, you're learning from lots of different signals and feedback loops that come into it. And so what I would say is.
I think like a relatively narrow, uh, AI SDR is, is going to be a tough sell, right? Sure. You can send a lot more emails and if you're not careful, create a lot of spam. Um, and, and, and test, but the most, the more interesting insights are actually based off of not just the customers that you win, the customers you lose, what, what happens after customers have been with you for six, nine months.
And so if your data model only extends to, you know, here's what, you know, I'm seeing in my [00:15:00] outbound data and response data, it's relatively narrow compared to what's happening in it. And so I, I think that companies that are deeply integrated down the stack should be able to do a lot. You get a backup when we think about the problem generally, it's not just a how do you outbound and respond to people quickly, but it's a question of do you have a great customer data platform to the great CDP to really understand, um, you know, from a graphic data, um, you know, what's the nature of these companies, what are intense signals, um, you know, that might be predictive in some sense of, of this is a company that may be in the market now or soon for the nature of the product that Um, we're selling.
We don't want to sell somebody a product that isn't relevant to their needs. Um, um, there might be, uh, details about what's happening in that company. Maybe they've raised funds. Maybe they have personnel changes. They've hired someone into a position. Uh, maybe, um, uh, someone has gone over who used in love ramp before and we can see this, and so you need to think about structurally, uh, not just what are you sending, but what's all the data that upstream that's informing that how do [00:16:00] you get information and signals in constantly to make sure it's new and fresh, you don't want to.
You know, act on the list that are many years old. Um, and, um, then once you're sending out what's happening, not just in the back and forth, what drives immediate conversion, but downstream, uh, what do you learn from that? And then the interesting thing that, that, that I would argue, uh, ultimately makes these.
Um, it was really, really profound, um, is when you start to see not just, okay, a customer has taken, you know, a prospect has taken a meeting, they're in our sales funnel, but what happens, um, um, you know, over the course of, of a year, um, you know, how much value are able to bring these customers, um, what are the ways that it's been for the business and do you start to bring that back?
And so what I would say about, um, maybe like a larger abstracted points about AI, you know, um, AI in general. Um, I, I think like one loose analogy to, to think about implementing AI in your business is, you know, um, you, you might think about there, there are people with profound [00:17:00] levels of intelligence, but maybe have lost, um, uh, you know, interaction with their, you know, spinal cord or ability to use certain kind of limbs.
Um, and you can have people, you know, like Stephen Hawking, like amazing intelligence. Um, and can act as he's, you know, was very respected. People wanted to, you know, listen to and learn from, but there's a lot of limitations on his output, um, based on, you know, his, his, his, his, his own body, um, you know, and, uh, inability to act what I'd say is like. you have great intelligence, you want to think about what's what, you know, do you have enough signals of input speeding back into it? And so it can process and act on a wide variety of things. But then do you have the ability to go and act on and get feedback on, on, on things? And so what I'd say about, you know, a lot of AI based tools, if they're very, very narrow and don't have the ability to start to build into learn from, um, benefit and improve, Um, lots of different, um, areas, um, of how a business ultimately operates, the utility is far more limited.
Um, so I think point based tools [00:18:00] in certain clever areas can work, but I think it's especially important and I think part of why, um, I, I would say even people in, in your line of work and in investing aren't so sure, um, that, um, AI will. Um, absolutely, uh, benefit, you know, startups. I think there's a lot of views that, you know, this might actually beneficiary to the very, very large companies, um, that have a huge amount of data.
A lot of it isn't well integrated. We can talk about how, um, you know, companies are approaching that to try to make sure, you know, data is, is, is, uh, stitched together, uh, in a way that you can act on it. Um, but yeah, I, I think it's, it's intelligence in action. I, I think are, are the interesting paradigm to think through.
Logan: Before we hop off of AI internally, uh, you referenced Toby, can you describe, um, what that is at maybe a finer level and, uh, How it came to be and how it gets used internally.
Eric: Yeah, for sure. So there was a few [00:19:00] points of major releases where we wanted to try to get. Exposure to how the latest APIs worked and how the new large models. And so I think it was created, I think, in the weeks following the release of GPT 4. And functionally, maybe to back up, you know, for many, many years, because we're a hybrid company with salespeople everywhere, you know, product folks that we wanted to have learned from customer calls, you know, every call.
Um, Polk's always, you know, can send out, um, um, has a gong, uh, recorder that, that joins at ramp. And so, uh, effectively we were, we were using that for sales training. We can see what percentage of the time is the ramp rep versus the customer speaking. It could give you a sense of, Hey, we're talking too much.
We're not listening. What's helped this salesperson improve and be more customer centric, um, uh, and feedback. But over time, we built this very large library of transcripts, um, um, uh, and if you add in the video and audio data, terabytes of data, [00:20:00] uh, uh, Rahul, um, who's one of the leaders of the Applied AI, uh, team, basically realized this could be a treasure trove of at scale, what is every customer basically telling you about why they're going to buy, why they're not, the problems that they have.
The reactions over the course of the sales process, and this might be an incredible way at scale to understand what our customers actually thinking and feeling about ramp. And so effectively what he did, I think this was over a 12 hour sprint was the first version. He built this out, um, uh, transformed all the audio data into transcripts, uh, dumped it, I think it was at a size, um, uh, I think we kept running to both into a redshift database.
Uh, then built effectively a model to efficiently query. This is, I think, when calls at GPT 4 were still an order of magnitude more expensive. Uh, and so rather than, than having, uh, this large language model simultaneously act over, you know, uh, call it a hundred thousand transcripts at once. Um, you know, let's say you'd ask, why do [00:21:00] people, uh, you know, prefer Ramp over Expensify?
Where do people prefer Expensify over Ramp? Um, you could do a first round of querying to understand where is Expensify showing up in, in, in Ramp. Um, you know, you, you take certain search terms and that sort of limits down the database and, and the set of, uh, objects you're gonna query on to let's say, um, who knows, a thousand, right?
Something like that instead of a hundred thousand. Okay. Uh, from there, um, uh, you know, you, you, um, uh, you know, feed in, um, you know, um, you target this as the output data, you, you, you, you overlay the, the question that you want to ask to it. And, um, you know, it's, it's, it's out, it's, it's effectively a summary, um, answer to your question.
Logan: And is that going into a large language model at that point?
Like it's sitting in Redshift, but how, when you get down to a thousand transcripts, is it then going into, are you feeding that in to get the answer back?
Eric: Yeah. I mean, effectively it's, it's a pointer where, where these are, you know, that's the, um, you know, part of the, the database that, you know, our input is, is going to be, um, running over. Yeah. And then we [00:22:00] take the output of the models, um, as well as some of the alternative outputs that were produced, um, into our own database.
Um, and then part of that we put into a Slack bot. So the way anyone at RAMP can interact with it is you can say, you know, hey Toby, um, go ahead and ask your question. Uh, Toby queries it a minute later, there is this output, um, response and if you want to go deeper, um, Toby can even say, you know, in customer call X, Y, and Z, they, they said this.
You know, there were 500 relevant transcripts, um, and if you want to dig deeper and actually go and understand or later on reach back out to understand what was this customer saying, um, it's really, really, it's, it's, it's, it's easy. Um, but the interesting part about it is, is it actually makes like a very complex set of operations.
Uh, in questions, uh, that really only highly technical engineers could ask before, um, you know, democratize. Uh, and so an SDR who has come up against, you know, saying, Hey, but we have, you know, Expensify. I can go and ask this a minute later, come back with all the reasons of why people prefer us or the competitor, uh, and write better copy.[00:23:00]
Um, a product marketer. Um, Who has to think about a competitive page can go and take that and sort of think through what are people actually saying. So it's not just conjecture or what do people say in the product marketing cave, you know, but what are customers actually saying? Uh, And so, you know, this is a, you know, in some sense, it was actually a relatively simple thing to build because we had large data sets and it was about building an analytics tool on top of it.
Um, but you can start to see and think about where this is going over time. Um, um, if you have a great data model that's running, let's say, let's extend the SDR analogy, um, how we're reaching out and communicating with people at scale, um, you know, uh, there's one world in which, you know, SDR is writing a better response.
There's another in which what we're learning over the course of customers is actually piping back into, um, uh, what we're ultimately sending and we're having kind of self learning. There's Uh, pages that can be really improving based, um, you know, not just when, you know, someone in product [00:24:00] marketing thinks to ask, ask the question or someone running a conversion rate improvement is asking the question, but we're taking that and informing new experiments that we're able to drive, um, to even to, as we sort of think about, um, ultimately building.
Um, part of how we build, uh, new products at ramp is, is reasonably algorithmic. What's the, the spend that we could win or lose? How many more customers could we serve if we have it and what's the most efficient way to build it? Um, a lot of, um, uh, prioritizing comes down to, um, do you have a good sense of how many customers want it?
How much they want it? Um, you know, and, uh, is this the, the only thing they're asking for a broad set? And actually you can start to, uh, prioritize. These features are coming up a lot in this segment that we're trying to win. We're hearing it now. Let's prioritize for the roadmap. Um, there's research that can get done in order to kind of build that.
And if you have great data models that stitch together, you can actually build more customer centric products much faster. I believe.
Logan: It's such an interesting example of, of AI. And it's kind of unique in that, um, you had a [00:25:00] large dataset that was, uh, aggregated in some way. Uh, and so, and then you could layer on top reasoning and summarization, Seems to be the best AI use case today, at least within some confines. Uh, the generative stuff is, is cool, but there's, you know, abstractions
Eric: a lot of abstraction, there's
Logan: or, or different ambiguities that come with generative hallucinations or whatever it is.
And so it's such a, um, it's such a perfect, uh, I don't know, distillation of the possible today and hearing it in practice, I guess I'm curious. The democratization of this access and information. You mentioned a few different manifestations of it internally. Is there something that's counterintuitive that's maybe come out of people having access to this in a way that, uh, wasn't possible in the past?
Eric: There's things we've observed because there's both what we're building and also like, we're very interested. We, we love partnering and using lots of tools and so, uh, we tend to get more pitches than most from like, you know, the new avant garde startups, what's the
Logan: You're a good logo to [00:26:00] have. Uh, yeah.
Eric: It's, it's, uh, we, I love working with people who are super high energy in building new things and it's funny, I think right now, um, there's a lot of noise being made in customer support, you know, as areas of this, is this going to be one of the areas that's, that's odd, I mean, I still believe people who really listen are taking to these journals and ultimately, you know, create a sense of trust and be there for customers is important.
We'll talk about how many people need to be into it, but I actually think relationships get more important. I think there's certain parts of, of, of job functions that people thought, um, uh, it would be like the last to touch. And what's funny is actually one of the maybe hottest areas now, um, which I wouldn't have expected a, a, you know, a year or two ago, um, you know, of, of AI startups is sort of AI engineers.
Um, and, uh, when you think about it, I mean, it's such a very challenging, it's a really challenging, uh, job. Uh, in order to be able to perform, but yet when you, when you sort of flip and you think about what actually drives great AI products. It's about great input and output data. How much of this is digitized and can you sort of start to see the [00:27:00] effect?
And when you sort of think about a lot of software engineering, um, you know, is fundamentally digit. It is fundamentally
Logan: digital, it's deterministic. The inputs and outputs are measurable, right?
Eric: you can start to see, um, yeah. You know, standardization and how things are, are, are, are built. Um, and, uh, I actually think that's been a part of the market where there's the most fascinating tooling. And so I, I also believe for, for, for what it's worth, like truly extraordinary, you know, engineers, I think become more and more valuable, um, uh, and can act with even more leverage than they could on their own.
And, and so, uh, I, I'm not of the view, um, uh, I think jobs are going to change. I, I, I don't, you know, I'm not of the view that like this stuff is, is, is existentially scary and that humans fundamentally adapt. Um, but I think there's been counterintuitive areas in, uh, where I think the meta and the most interesting tools have shown up.
And I think that the other, which I think as a practitioner is obvious, but I think is worth, I'm stating for for people who are building businesses. Um, it's in my view, like it's not just about, [00:28:00] um, what is the functional use case? Um, what is the AI tool tool do is being like the most important tool selection.
Actually, one of the most important things is that, you know, again, if you have, if you're able to bring in digital intelligence into some aspect of your business, um, having the ability. Yeah. Uh, to have well structured, organized data that is connected and interoperable enhances the ability for AI to add value to our business.
And so, uh, I think that, um, you know, building applied AI engineering teams is super important. But I think building extraordinary data teams, being thoughtful about data models, uh, and then thinking about, um, really tuning your business to, uh, not just get insights, but to act on them becomes more important.
And so I think it might make organizational design, uh, data structure design, uh, more important. Uh, and so I, I, I, I think it, it, it forces people to think about different areas in their business than maybe they may have assumed.[00:29:00]
Logan: changed your hiring at all on the engineering side with the, uh, Some of this tooling that's, that started to come out, or is it still too early, uh, to, to see the impact
Eric: Um, I mean, I, I, I would say it's.
Logan: think we
Eric: think informed probably every area of hiring it, at least in, in, in some side, right? Like I am, you know, engineering is, is, is one, but, um, you know, look, I'll, I'll, I'll talk about a few others. Like I, um, uh, when I think about, you know, design, um, I think one of the things that, um, people inside the industry, we've got a lot of notes and excitement about is I think we were, uh, Yeah. I think we were the first company to run a scaled out of home advertising campaign that was primarily powered by mid journey. Um, and you think about some of those images, um, that were out, um, you know, in, in, in the course of campaign, if you wanted to produce some of those single shots, um, um, to get the effect, the lighting, the field, the props, [00:30:00] and, and, and kind of the emotion of it.
I think some of those shots would have cost into the tens, a thousands of dollars, um, you know, to, that would have been a multi hundred thousand dollar campaign to produce.
Logan: thousand dollars,
Eric: It was done over the course of a day, um, with as much as it cost to create a seat for some of that. Uh, and I think that, um, I think it's more important to find, you know, people with extraordinary tastes, um, what, you know, can give you a sense of what is a striking image?
What is going to connect with people emotionally? What should I produce? Um, that is interesting work, but the cost of production, I think of artistic and great work is, Uh, probably as low as it ever been in, in human history. Um, and so I, I, what I'd say has become more important in every hiring, uh, or hiring for any kind of job profile is, is this person fundamentally curious, um, uh, are they interested in how the world is changing, um, um, or, you know, in, in, in, in, in how it's going to change your job and how they can do more and, and do their, their job differently.
Uh, I think it's more important to find crafts people. Um, um, I, I think that as companies get bigger, I think something that [00:31:00] everyone feels who's been in part of these organizations is. You know, the distance between craftsperson and craftsperson, people making things, gets really large, um, um, and there's, you know, a lot of, of editing and things get diluted down to the point where it's, it's very few people actually working on the actual product.
Eric: There's lots of people just making the organization, uh, and the bureaucracy work. And I think that companies are going to get a lot flatter. And so finding makers, people who want to learn, who are curious, who are interested in how does something fundamentally work, and how will it change your job, I think is more and more important.
And so I don't think that's a, um, you
Logan: Unique engineering necessarily, and I guess one of the points that we were talking about before we went on that I think is an interesting one as we, um, segue from the, the, the, the AI internal topic is the, the role of bookkeepers over the last 10 years or so, uh, maybe can you speak to that point?
And, uh, how. Bookkeeping has changed, uh, as an industry and the number of jobs.
Eric: Oh for sure. For sure. I mean, [00:32:00] it's. I mean, first, just stepping back and thinking about it, um, I mean, I think it was, I think double entry accounting, um, was invented, uh, I think in the lead up, lead up to the Renaissance, um, uh, in, in Italy and in, in some people say actually the, um, you know, that was one of the precursors that, that generated a lot of wealth, um, um, which could give rive to one of the greatest periods of creativity in human history.
Um, but. Like I, I think the reason it exists is that humans make mistakes. Um, you know, you have one set of how, you know, your finances are. Um, but when you're interacting with lots of different parties, you know, it's hard to do it without. Um, you know, it checks and balances and how you ultimately keep and do do keep your own books.
Um, and that was that has been the avant garde of how record keeping has been done. There's a lot of great things about it for, you know, Uh, 700 years, um, some, something like that, um, and it became very, very important, um, you know, call it a hundred years ago, uh, as companies [00:33:00] started getting a lot bigger. Um, uh, it wasn't just regional banks.
Um, there was the creation of branch banking, uh, and you had national banks that could operate in far flung places. Um, you had, um, you know, large industrials, um, the rise of standard oil, um, of General Motors, um, of, uh, Standard Chemical, these, these large companies, uh, and suddenly people needed to be able to understand what was happening in these local branches, you know, how do we have multiple sets of records between what they're reporting, um, you know, audits against that, uh, and so you needed to build.
Um, not just a, you know, great systems.
Eric: You could scale like double entry accounting, um, but swaths of people, uh, in order to be able to measure audit reports, uh, uh, and then act, uh, on this where I think it gets to your question is very interesting. I, I, I think it was an article on the wall street journal.
We'll try to dig it up, but I think it was something like, um, you know, it said the stat that looked, um, wild on the face of it is said, you [00:34:00] know, there, um, a decade ago, there was 3 million. bookkeepers, you know, professionally, um, or 2 million bookkeepers, especially in America. Now it's, you know, it's a million less.
Um, um, and you know, it looks like this, this, this crisis until you realize that the number of, you know, um, financial advisors and analysts has gone way off, you know, if I, you know, call it a million people. Um, and I think is an indicative of what happens in progress. Um, you know, when you have, um, better tools and systems.
Um, it becomes a, you know, uh, a system people move to, uh, more strategic or a high level of abstraction, um, is, is, is one meta point. I think the second meta point that's a little bit interesting is computers, you know, uh, don't really make mistakes. Uh, so there's people that program them that make mistakes and you can, you can model things incorrectly but, uh, fundamentally excellent, uh, at keeping of records, uh, and I think in a world in which more, um, uh, you know, transactions are, are [00:35:00] digitized, um, it's easier to know with accuracy what ultimately happened.
Um, what I would say today for the reality of, of someone using, uh, I think for the average business, the regular experience is. You know, to buy one thing, um, you need two tools. You need like a credit card, and you need an expense management software, and you take a digital transaction, you get an analog image of like a receipt or a printout, you then take a picture, you re digitize it, you put it on some other app, that then needs to determine does this transaction match this transaction, and you port it, and it's just very Kafka esque.
It's very clunky.
Eric: Um, you know, today, I think one of the original drivers of Ramp's product market fit was, You don't need two tools to buy one thing. You know, you put, you tap your card, we check your expense policy in real time, um, you know, you get your receipt, we pull it from the merchant automatically, or email, or we text you when the receipt's in your hand, you snap a photo, and you're done.
And so it sort of takes what used to take a month for you to get your receipt in, on average, on, you know, on these old tools to, 30 seconds is the average on RAMP. Um, so it's an easier process, but as [00:36:00] you look at how this is extended, you know, today, as you know, we, um, we issue corporate cards, bill payments, reimbursements, um, you know, we manage tens and tens of billions of dollars of purchases, and we stitch data from what's happening in our HRAS through what's happening in our HRIP.
Uh, and, and where I think this all comes together, uh, is that for the average business using RAMP, um, they're dramatically more efficient. You don't need such large finance teams, uh, in order to be able to keep books and records, which is a good thing because it's not high value work. It's actually, I think, very painful parts of people's jobs and instead they're spending time, uh, on the interesting questions.
Uh, which is like, where does return on investment come from? Where should companies be deploying more capital? Not just, you know, uh, doing lots of work in order to be able to measure what's happening. You know, they're not spending time on downloading a bunch of spreadsheets and running large pivot tables to figure out, okay, this transaction is UBR star two, three, seven UBR star one, two, three, all these say Uber.
And so after an hour, you can finally say, this is what we spent on Uber. That's done for you. Um, you know, how your books and records are kept are [00:37:00] done for you.
Eric: Uh, and I, I think that, uh, over time there's been this like white whale, uh, in, um, I think investing of this is this idea about a decade ago. People talked to this, this notion of one day there will be self driving money.
Uh, and, and companies and people will have their assets in the highest yielding, uh, security or asset at all times. Uh, they're not going to waste a dollar on things that they don't need. Um, and they're going to have better insights on how to live a better financial life. Uh, and it didn't happen. Um, but I think that as companies become increasingly digitized, they're connected together and you have a command and control system like RAMP.
Uh, I think self driving money is actually kind of a real thing and in the same way, You know, a decade ago, people were excited about self driving cars and, you know, it's finally here and people are just getting used to it. Uh, I think this, I think that's happening, uh, in the world of, of, of finance and, um, whether it's us or, or someone, uh, it's going to happen and, and, and we want to be the company that does that and brings it to people.
Logan: So so as you guys have dog fooded.
Logan: We spent a lot of time talking about the, uh, the internal processes that [00:38:00] you've made better for for AI. And then we also talked about the on prem. The, uh, outcome based pricing that you've been able to align with your customers. Um, how has that now manifested in artificial intelligence for the end customer of ramp and how did you, I realized it was a first principles thing from the start of the business, but, uh, maybe, maybe take me through as chat GPT happened and there was this proliferation of LLMs or it went mainstream.
Um, how you upleveled or thought about upleveling the, um, the end customer experience with
Eric: sure. Um, so I mean, a couple things. We're always interested in the field of it, um, simply because there have been aspects of machine learning, um, that have dictated how risk and underwriting have worked for decades.
Like, it's been in production, um, for quite a while, uh, and so very early on, um, some of our first hires came from Facebook AI Research, Google Brain, like, first [00:39:00] Five people at the company kind of a thing, um, and so even back in 2019, our receipt matching was finally, you know, um, you know, driven by machine learning.
By 2020, a lot of our underwriting was ML based, um, you can have human underwriting, but actually you could use data signals in order to inform. Um, how a company's finances may perform and how much limit we can extend. And so there's been aspects of it for a long time.
Eric: Uh, what I would say is, you know, you know, it's certainly not today, but I don't think it has been for years possible to use ramp without using AI, uh, in some way.
Um, the difference is, um, historically, we haven't talked about it a whole lot, you know, you look at,
Logan: called it ML and not, uh, AI probably in the early days.
Eric: yeah, yeah, and well, well, even this, like, you know, we have lots of, um, you know, most of our businesses are traditional, like they're, they're farms, um, you know, restaurants, hospitals, uh, manufacturers, um, and if you look at reviews of Ramp, the most common thing you will see is it's really easy, um, you know, it's really, really simple.
[00:40:00] And, uh, if you look deeply, um, it's because there's AI driven in almost every process. Like, you snap a photo of your receipt and it's zero touch AI, um, it just matches to the right thing. Um, you go and, you know, you buy something at Costco and you get a text from Ramp saying, you know, we have this suggested memo one, two, and three, click one which feels right or fill in your own.
Um, you know, people pick one, two, and three almost every time because we're better able to predict, um, or better able to, um, you know, model, um, and, um, and tag transactions certainly more quickly, but also for the vast majority of customers more accurately. Um, and so it's stitched in, in lots of different places.
And so there's a lot of work around. You know, this category of products that we think internally about zero touch AI, where you don't need to call it that, um, but it just feels smoother, more intuitive. There's less friction in any part of the workflow than normally you think about. That's one. There's another class of more external pacing products that we're really interested in, [00:41:00] um, that, Um, you know, we think about is agentic, uh, a I, um, and when you think about that, um, I think probably a use case that blew up on, on, on Twitter, um, about a month ago was it was, you know, similarly development of Toby, um, uh, that followed the release of GPT four, um, uh, open AI released it's a GPT four, Oh, um, um, multimodal model, um, that you could take.
Um, you know, images, video, audio, um, you know, text, um, and get the output of. Of, um, you know, these large models, um, and the prompt was, you know, one of the top, uh, uh, sets of questions that, that people will ask about ramp is, hey, they, they know what they want to do. They know ramp can do it, but they're not sure how to do the thing.
Um, so an example could be, hey, um, I want to issue a card, um, that only works at coffee shops, you know, and, um, It's, uh, give it a limit of 50. Um, but I couldn't figure out how to do that. Um, [00:42:00] of course, you know, number one feedback, we want to make a product more intuitive, make it more easy for people to figure that out.
But we said this actually might be a great application, um, of the GPT 4. 0 model. Uh, and so, you know, after, um, you know, a little bit of work, uh, I think about a week of work from, you know, Alex and others on the team did an extraordinary job. Um, uh, yeah. What you can, what you can do is you can, you can say, Hey, ramp.
I'd like to issue a card with a limit of 50. Um, uh, just for coffee. Um, help me do this. What happens next is, is very interesting. What you see on the screen is ultimately fed into, um, this model. And so it takes this, this, this text prompt, uh, inquiry, um, and, uh, audits what's happening on the screen as well as other, other screens that are possible to click through Uh, and then shows you how to do it.
Uh, and, you know, it says you can click down here. Um, you know, into the cards model. Hit new [00:43:00] card. Um, we're gonna hit controls. You're gonna say allowed merchants. Um, you know, coffee shops. It'll pick the, you know, it'll write this. And as it's doing it, it's showing you what it's doing. It's saying, I'm gonna click over here.
We're gonna write this text in. And then it's going to click issue. And, um, I think this is really interesting because it's, it's, it's one of the first cases where it's, as an agent, it's not just explaining how something is done, it's actually acting. It is doing things as an as an agent of yours, uh, doing things in this case with you monitoring, uh, the outcomes, um, but it's very different than how companies, uh, products historically have been done, um, um, where they're taking some action in the real world, um, uh, and it does it.
Now, of course, you know, I think the easy question is that sounds great. Why show me the steps? Why don't you just do this? Um, I think some of the answer is these are, these are, um, you know, alpha products. We're testing it. Um, um, and we want people to kind of monitor and sit shotgun and see, um, in the same way for Waymo's before, um, you know, they [00:44:00] were driving on their own.
There was someone kind of there with a steering wheel just to see what's going on. Um, we think that's a kind of a good paradigm to, to see how these go. Um, but over time, uh, you can kind of imagine, uh, you know, um, you can learn all the way a SAS app works and all the tools and knobs and ways to configure it.
Or you can say, here's what I want, um, loosely, can you go, tool, configure yourself. Um, and can you go do this thing on my behalf, um, and, um, you know, tools should be more intuitive, um, to be able to use directly, but also, uh, I think tools are going to act on your behalf, not just in the tool itself, but outside of it, and generally on the internet, um, and I think that Paribus is an interesting example where, we It was an app that wrote emails for you and chatted with people on at other companies on your behalf.
Um, and I think those kind of use cases are really prominent. Um, how do you get better deals, um, on the software and sets of things that you know you're going to be buying? Um, can you perform certain analyses for, for me? Uh, and so when I, when I, [00:45:00] I would say over time is the, um, reasoning capabilities of these models increases dramatically.
I think that just Even the nature of what, uh, software can do for people is going to expand dramatically and I think, um, having a sense and developing taste of, um, of, of, of where, uh, things are great and working as expected really quickly and where they're running into issues, uh, I think is a really worthwhile thing.
Logan: As you think about that multimodal, uh, for, uh, use case, it sounds super interesting and cool. Um, as these things get done internally and you empower people to do them, you're thinking about the. The incentives or like what success looks like. And I was just taking down as you were talking, like thinking of what success looks like.
And one is that people actually want operate that way. I think that would be the most idealistic one. Uh, people actually like, it's a good functional way of, of doing this. And that's the most idealistic. Then there's two other ones that I've sorted or three other ones. I guess one is employee motivation in some ways that [00:46:00] like you're empowering people to do cool things.
And like whoever the engineer was that did that, I'm sure had a really fun time figuring this
Eric: Yeah, yeah.
Logan: Um, then there's sort of this, like, I don't know the right terminology for it, but like some, um, like obligation in some ways that like you are a company at the forefront that has the ability to do these things.
And so there's some like obligation that maybe you guys feel, I don't know, uh, to continue to push the frontier of these. And then the other one is maybe the most cynical one is like PR and it shows really well, and it looks really cool, uh, in doing this as you think about like those different components.
Eric: are,
Logan: Are all of them some factors or does someone you don't even think through these things and someone comes to you and you're like, Hey, yeah, this seems like a good idea and we should be at the forefront of continuing to push this stuff forward. And so let's do it.
Eric: Does
Logan: Or does it not even bubble up to you and someone just goes and does it and it comes back and you're like, that's cool.
There's a
Eric: Someone
Logan: Yeah. Yeah. Yeah. [00:47:00] One
Eric: does it, and it comes back, uh, and you're like, that's cool.
We make Tide Pods and we want to sell more Tide Pods. We're a credit card company, we want to sell more credit cards. We want to have more computers on more desks, that kind of a thing. Um, Ramp started with a very different guiding mission. The way that we all measure ourselves is how much money and how much time have we saved our customers.
It's not how many cards that we sell. It's not how many bill payments that we process. It's any of that stuff. It is what is the outcome that we're trying to drive. And when we do product mode maps, I think we're relatively hazy like on it. We say like, this is the future state that we want to drive. If we do this, you know, finances can no longer be tedious, monotonous, but strategic and insightful.
But how we get there, we actually give, live, leave like a lot of discretion, uh, to teams. [00:48:00] Um, uh, and teams themselves, uh, are quite distributed. Um, some of the largest teams, like the spend management team, uh, at RAMP, um, you know, is on the order of a dozen people. Um, which I think is very surprising for, for, for people when they, they know that RAMP is a 800, um, person organization.
Uh, and so when I think you have a clear, we're trying to, to drive this, you're responsible for this area of the business and improving metrics on it, but how you get there is up to you. I think you, you, you inherently foster creativity and sometimes people look to and have the permission to take new paradigm shifts.
And so it's not really me mandating, Hey, we want to do cool. You can always feel it when there's someone with like too much centralized control saying they're going to do this stuff. Like, Oh, we're this. AI this or we're pivoting to big data or any of these large stuff like I, you know, we, we hardly talk about this stuff.
We really talk about like this tourism. We exist to save you time and money. Here's the products, here's how they work. Um, and I think it leaves the place for incredibly [00:49:00] bright and curious and just like excited people to say like, I'm going to do this, but I'm in a radically improve the amount of time that we save, you know, and I'm going to implement AI as a tool, but it's not about like, what's the shift.
It's more of what's the problems that people ultimately have and how do we go build, build against that? Um,
Logan: of the things that you did touch on there that I think is, um, Is an interesting one is is org design and you touch on this a little bit earlier as well, but there's this balance that exists between autonomy. An empowerment of the individual and, um, consistency and centralization of the experience. And those two things are kind of at odds with one another, uh, at times.
And, you know, if you, if you run a dictatorship, uh, the design, the output that someone sees, uh, from a, you know, company product or design or aesthetic. Probably a lot easier to keep consistent the font, the fonts and the colors and all that stuff than a decentralized organization that people are making a lot of independent [00:50:00] decisions.
And there's this tension between the two that exist. And so how do you think about, like, what are the shared services, uh, that the organization has? Hey, we don't need, AWS shop and not a GCP shop. And that's not something we're going to. Let individual people make their, their own decisions on versus, Hey, we really do want you to solve the business outcome and how you get there is, you know, on, on you and where the tension or the, the balance exists between the two.
Eric: It's, I think, an amazing set of questions and I'll maybe start, I mean, you highlighted, you know, AWS, Amazon Web Services versus, uh, GCP, Google, uh, Cloud Platform.
I, we, you know,
Logan: Yeah. Or whatever it is.
Eric: yeah, I remember this debate when RAMP was on the order of 40 to 50 people, um, pretty early days. And we were try, we were trying to reason through. Would we be a highly centralized, really deliberate because we do have a point of view. Um, we are opinionated on [00:51:00] products should work a certain way.
They exist to save you time and money. Um, if you tightly couple certain products, they work better, uh, for certain people. And you're a small company, you have very limited resources. Versus, you know, do you want to give lots of flexibility and sometimes you'll have, you know, um, some wasted effort, but you don't slow people down.
People have a lot of discretion. And I, I would say probably the most extreme examples of this are, are probably the organizations you highlighted. Amazon on one end, And Google on the other were Amazon. If you use an explorer, um, all the tools and services available on AWS, it's like a zoo, like there's, you know, multiple versions of tools that do the same thing.
Um, uh, competing teams, um, where certain teams win and lose out. Um, but you know, for a long period, like really radical, lots of good stuff and a super thriving ecosystem. Google, uh, I think is very famously the, kind of the opposite way. Um, where, uh, it's [00:52:00] highly centralized decision making. Um, you know, as products come out, they tend to use the new norms.
They sort of, you know, uh, turn off products all the time. Uh, and there's these like mega organizations all working towards these one super products. And, you know, that that's an extreme example of reasoning through which one do you want to be? Do you want to be Google? You know, with like one or two super large products, you're going to be Amazon where it's like, You know, it's more of a rainforest with, like, crazy stuff going on in there and,
Logan: the measure twice, cut once kind of dynamics.
Eric: yeah, and, um, I'll tell you, like, after putting it in, in those terms and really thinking through, he said, you know, like, I think Amazon, um, has got to be the way, um, because it when I, I, I think that one of the only, um, when you're a startup, like having really focused effort, high velocity, moving quickly, getting things out and learning often is the way because We're wrong a lot all the time.
Um, but if you can, um, understand how customers react to it, you can [00:53:00] improve and more accurately point the direction and vector of your product. Um, you get a lot further in the right direction, a lot faster if you take these super large bets released in multi month to multi year cycles. Uh, and so we said, you know what?
We are, we are at times going to have many different ways of solving the problems. It's going to create sometimes some clutter, um, and we think that's a good thing. And we need antibodies and internal mechanisms, um, that are, um, you know, doing occasional QA in order to be able to not stop the production and the creation of new things, but to be cleaning, uh, in sort of sweeping the streets so often.
And so when I think about what that means, Is like you actually start to build teams that are both you have, you know, single threaded teams, um, where you have people from the beginning and inception and customer research, you know, um, interviews of the products to building alphas to betas to shipping these productions where it's the same teams, um, really small groups that may be a PM, you know, uh, you know, three to seven engineers, depending on the nature of maybe a [00:54:00] designer, um, and then as it gets ready, then you later start to bring other people on and we have lots and lots of small teams.
Yeah. And so most of the organization is that. Then you also have certain horizontal orgs that you really build. It's how do you make sure that for any developer, I don't care what you're working on, it's easy to develop. You have standard sets of infrastructure. And if you need new tools, we're going to make sure your experience of procuring, uh, working with and using that is really fast.
Um, uh, and SLAs are incredibly important. Uh, and you even review people differently. Like when you review, you know, products at how well do they do? You ask customers, you look at speed and velocity of productions, you look at outputs. When we are reviewing folks in more infrastructure type organizations and more cross functional organizations, I actually care very little about what the manager thinks about a particular employee.
I'm really interested in how the cross functional partners that speed, they impact this ability to do their work to impact, um, that's probably the most important signal that you get. Uh, and so I'd even say it even goes down into like review design. Um, and thinking about [00:55:00] performance where, um, I, I would say I wouldn't just pick one.
You, you need a nature of both, but even, you know, assessing individual performance, thinking about what are the SLAs, what are you measuring and try to optimize for becomes very important. And I, I, and what I would say is that where organizations get this wrong is, is they oversimplify and they say, you know, everyone needs to get reviewed by the managers and all their downward reports.
Um, and that works perfectly fine for some, but it's horrible for others and it becomes very bureaucratic. Um, or you have certain practices that are really well set up for horizontal functions that don't, that don't deal with, you know, uh, single threaded teams. I think that's why, like, you know, I had lots of friends who work at Google and they're like, I, I leave because I work on something that maybe if I'm lucky, like a year later ships.
Uh, and that's fine for the horizontal people, but really tough for the vertical people who want to get things out. And so I would say no system is perfect. Um, you know, I've heard of, you know, uh, some organizations that reorg every 18 months just so. Yeah. You know, you have some period of time when you get the benefit of changing it.
Um, you know, I would rather not do that and be all one way or another. But sort of [00:56:00] thinking about every individual function and how do you structure, uh, what it does, how it can operate, the goals you set against, and even later on a, uh, performance basis, how you measure those people
Logan: So when you set it up in that way, like maybe a new product initiative or something, uh, you're going to roll out, um, you guys recently announced travel. And, uh, so does that get set up as its own independent entity and with a PM and a handful of engineers and designers? And then ultimately, does it roll back into some more centralized product organization when it reaches maturity?
Is that the typical life cycle of it or does it stay independent and autonomous or does it depend?
Eric: It's, um, totally right. Travel is an awesome example, um, uh, of this and bill payments and, and others. But, um, so first, um, what you said is exactly right.
We started with a couple sets of people, which is, hey, um, uh, I think a few years ago, maybe 10%, um, uh, of all car transactions on ramp were, uh, bookings of flights, [00:57:00] hotels, autos, travel, entertainment, that, that kind of a thing. Yeah. And we could see very quickly. This is going up and up and up. Um, you know, today, about 20 percent of all transactions on ramp are booking a flights and hotels and all those.
And, um, there was this customer need. People were saying, look, um, I would like to have more control, um, over, um, when, you know, an employee is booking a hotel and flight. Um, I would love if your, if your great card controls could actually, um, Uh, dictate maybe a booking portal. So if people ultimately are booking, they see hotels that are in policy, they see flights that are in policy, um, when the receipt, you know, is, is, the transaction is, is affected, the receipt is automatically pulled back, right?
And so there was, there was a clear customer need, it was building on top of additional product, all this kind of stuff. Now, we didn't go say, spend management team, go build travel. What we said is, we, we, we, we sort of took off a, a small team, we, Functionally locked him in a room for like a year and we told people don't talk to them.
They're just going to build travel interviewing customer because we're researching what's the [00:58:00] right form factor to build this and it was exactly as you described. Um, it was, I think, initially, um, predominantly engineers. I think, um, some from product joined reasonably early on and it was decisions in trying to.
Do research, what were the customer problems, uh, reasoning through the right way to build, and then working with a, you know, increasing large set of first design partners, alpha customers and, and all that. Eventually, we start seeing certain metrics get hit. We see, you, you give it to a customer, they like the design, say, great, you like it so much, would you use it?
You build the thing, they have access to it, um, they start booking with it. And they don't just book once, they book twice, they book three times. You know, you start to see retention go through and eventually you can start to get to this early set of product market fit, uh, type metrics where you can see this bucket is holding water.
Um, customers are getting, um, you know, value off of it to the point where they don't want to use other tools. Um, and effectively you, you allow like a young baby child [00:59:00] product to start to grow large enough where it's ready to go off to school, it's ready to start to join all the organizations. And so what I'd say is that, that team was predominantly interested in development around, um, you know, travel related products.
They were absolutely beneficiaries of existing products that, that exist. You know, um, you know, DevOps making standard tooling that makes it easy to create, uh, new tools on, on, on, on top of ramp, a BD team that can go and negotiate deals on their behalf, um, as we're working and evaluating different infrastructure partners, uh, spend management, which can go and surface this.
Um, you know, uh, car data to know what's travel, what's not, um, a receipt automation team, an accounting automation team, and so there's, you know, uh, they're effectively building single threaded, how do I build a great travel booking experience, um, or a great, um, you know, uh, travel insights tool, whatever it is.
Um, and sort of thinking about what's the API interaction between, uh, this cluster of services and the other ecosystem. But, uh, what I think is so [01:00:00] important, um, is if you, if you put, you know, small orgs that are building new products into the large matrix right away, um, you know, it's very hard to get resources.
You know, people would say, you know, um, Uh, I've got this large product, like give me the engineer to go work on this. And I have a large, you know, you know, portion of the business and you know, things die still bursts all the time. Um, and I think even if you look at like early history, um, if you look at some of the breakthrough products that like Apple ever developed.
Um, if, if, you know, uh, under the Scully era, um, it was all kind of large teams and they were mandated to work on some new products and existing products and it was really challenging to do. And, you know, for, I think largely for better, like Steve Jobs would go and put people in a different building and say, go talk to the, don't talk to these people, let them build that.
And we try to model that, uh, on it. Um, and so I think there's a lot around product market fit. I think that once you get to a certain scale. You start, the question is not to get a thousand people to love it, but how do you make sure this is a stable, robust, and the foundation is strong for 10, 000, a hundred thousand, a million people to use and love this.
And so the nature of things [01:01:00] change, the goals are not product market fit. They're more operational SLA based. Um, how well the service interacts with and reliable is it. And those can change when it's ready for it.
Logan: does that change often the people that are leading it? Like you start with pirates and move to Navy over time or something like the people that like getting it from zero to, you know, whatever, almost one versus the people that like rounding out. Uh, do you find those to be different types of people internally?
Eric: I don't know. Um, or like, I like the one piece fans out there like people want to be king of the pirates like, you know, eventually you go from like this little ship to running large organizations and there are some people who are that way love it understand a problem space so deeply and actually make that evolution and there's other people who like, I just want to build the new cool stuff and, and, you know, work on that and so what I'd say is like, actually, like, um, I think it's probably like a good meta point even just about like talent, like, sometimes I think [01:02:00] interviews over optimize on um, Uh, is someone on, on, on skills, uh, and how capable is someone at building something or doing some job?
And like, I will never fight with anybody about it. You should understand how, how well is able someone to do the job. You should assess, are they a good engineer? Are they a good designer? Um, what does it take? Do they have functional expertise? But I think a lot of times what organizations leave out, uh, and frankly I learned, I think it was one of the positive things that I learned when I was at Capital One is.
You know, um, everyone is the hero of their own story, um, understanding people's motivations, um, what they want to do, who they see themselves at, where they want to be in, in years, um, and, um, not forgetting that and periodically referencing back, uh, and checking in and saying, you know, Hey, you, you wanted to, you wanted to start a company one day, um, you know, the zero to one problem is really, really important.
I want to put you on zero to one problems and build that and then check in once they get to one. Do they want to go from one to ten and how are they [01:03:00] feeling about it? And um, it's not just, uh, I think in company and organization and building companies for the long term, um, you know, if it's true, all a company is, is a set of people, um, you know, uh, making sure that people are working on problems and feeling like they're progressing against areas in their own life that they want to go against is, you know, is I think one of the most important things that you can do.
And I think that, uh, I think that, um, uh, so many leaders, uh, across ramp or, or just excellent at this. Like, I think that, uh, Kareem, uh, Jeff, uh, Diego, Colton, there's so many people through the organizations that are really trying to understand not just functionally what they're accountable to, but, but thinking about, okay, like how are people ultimately doing, how are they feeling about it?
And if they want to go deeper, they do it. But if not, like, Putting them on types of problems that, that feels like moves them closer to where they want in their career is, I, I actually think what makes, what gives organizations longevity. Uh, and I encourage people to do.
Logan: want to move to hiring, but um, [01:04:00] and the, some of the things you look at there. But before we hop, I'd be curious, um, is you think about resource allocation within the organization. How much of you were to split it into 100%, how much, uh, effort, uh, or, or head count do you think it's dedicated to? like. The meat and potatoes of ramp as a core product, the spend management, the card, all of that stuff versus, I guess, new product initiatives.
And I don't know if there's a third bucket of moonshots or baby, maybe that's the AI team that you guys have as well. But how do you sort of think about the resource split of what goes where a
Eric: Yeah, we'll, we'll come back to the AI team because actually they work on a lot of optimization of existing processes, um,
Logan: I assume it's cross functional probably at all. Yeah.
Eric: yeah, we'll, we'll go into all that, um, but, um, I think a couple of things, I mean, so first, I think, uh, uh, I mean, he's very in vogue, but he has some crazy kind of management, um, recommendations, but like, uh, I think something [01:05:00] that, Yeah, yeah, yeah.
Eric: I heard, uh, Jensen Huang at NVIDIA say, which I think was really profound is, you know, um, uh, he believes his role as CEO is to create the conditions, uh, by which people can do their life's work. And I think it's amazing, right? Um, when you think about the journey of, of that company, um, uh, you know, he describes it, one is, one of the important things to think through is.
Uh, you know, first, how do you create the conditions? How do you find extraordinary people, create an environment in which people want to work, feel motivated, all this stuff that people, um, You know, obsess on, right? Maybe this is that, that whole, you know, do you have the right talents? Are they in seed? Do you have the right culture, all that?
There's a lot that goes into that. Um, and, uh, you know, by which people can do there's live work, I think is an important, you know, um, addition that I think is easy to really glaze over. Um, you know, that company for a long time was [01:06:00] making chips for video games. Um, and for crypto mining and multiple time, he sort of gambled a company and he said, look, we could go from being like a.
You know, a 5 20. If we just go deeper in the video game industry, but he said, you know, I've hired people of such caliber, um, that actually, uh, I need to be thinking about which work do I select and which do I de select from the company. And I believe that the opportunity to build, uh, GPUs for AI or for biological research, um, and genomic research is much more important.
Um, and so one of the important things that he assesses his own, uh, Um, effectiveness as a CEO is, um, are you making sure that people aren't just doing work? Um, but are they working on the right sets of things? Um, and I think that, you know, that company in how deliberately they've, they've taken large bets, gambled and, and I think time and time again done right, I think is a testament to his ability to do that.
Do you select the right work? Uh, and so I think that that's a big part of it, of a CEO's job. I think that the other [01:07:00] is, you know, um, I don't say I think we're one thousand nine hundred and twenty nine days old today. Um, um, you know, it's less obvious now, but like I remember day zero when we had three people and I was thirty three percent of the head count.
Um, and when it's, when there's just three people, it's obvious that, you know, Um, all the company is, is a collection of people, uh, and so, you know, if you want to have any shot at building something great, you better find really great people, uh, get them into the company, and then hopefully you can create the conditions for them to do great work.
And so, I, I, I think to actually back into the question you, you, you really asked, like, I, I think, um, You still need to spend an enormous amount of time doing it. I think I do spend about roughly a third of my time, uh, whether it's, uh, hiring, um, and, um, probably a little more. If you look at just trying to inspect what's happening in the business, you know, these really smart people.
Are they able to produce great things? Are people getting in each other's way? Um, and, um, are organizations working really efficiently? I think is a very large part of it. Um, uh, and I think it's important [01:08:00] because one of my, my, I have the power to convene meetings. I have the power to change structures in organizations.
And, uh, I need to use that, um, at times. And so I, I, I think there's a lot. Um, uh, and I have the ability to, you know, now more people are interested in responding to cold email from me than maybe a couple of years ago. And, um, yeah, I, I can meet from and learn from a lot of people. And so I, I think like, you know, some of my job too, is thinking about what are the things that only I can do, um, or maybe do more easily and make sure I'm, I'm putting my time on that.
And so maybe it's, it's finding extraordinary people, um, assessing structures, changing things out. Uh, and then. Um, you know, um, maybe outside of this third, am I selecting some of the right work?
Eric: so, so that's you at a first, the, the, I guess existing products, or how do
Logan: the, the, I guess existing products versus new products. How do you think about the, that split or that ratio?
Eric: I, I think it, it differs based off of job function, right? In, in some sense, if you're, you know, um, you know, a sales or account management or customer success, [01:09:00] um, you are working on existing products. Um, Uh, and that will, you know, um, I think understanding customers and bringing in to maybe think about where the future is going is an important part, but you are definitely living in the today.
Um, uh, I think for things like engineering and design, uh, uh, product, um, um, even aspects of like product marketing. Brand you, you want to be thinking about, um, you know, you know, structures that exist over many years. Uh, and so there's some element of where's the world going and how do I make sure I position myself as best as I can, um, for where that's headed.
And so I think these organizations tend to have much larger percentage percentages, um, of these orgs. Uh, working on the nature of those types of
Logan: The new
Eric: on the new stuff. And so at the beginning, I would say is like we, you know, before a company has product market fit, it's all the new
Logan: It's 100 zero
Eric: It's 100 0.
Yeah, I still think today, um, if you look at, you know, still the majority of, um, you [01:10:00] know, people related spend is in R. N. D. And in work, and I would say the majority of that, uh, is going into, uh, whether it's new surface areas or sort of taking Um, existing products to the next level. And I think that's great.
And I think that's sort of been what people, um, identify with a ramp. It just constantly improving. And that is one of the value propositions is it's not just, you know, uh, maybe the, the best deal in business, the most high ROI, like it saves you time and money, it pays you to use it and it automates your accounting, but.
I think a feature of using ramp is, you know, uh, every, um, if you're really paying attention, you see it as every day, but, um, you check back every, you know, month or so. And the tool just gets better. Um, uh, and I think that that that is a quality of it. And so, um, I think there is even this essential. Um, um, part of, uh, you know, our, our, our culture and what people identify about our product is like, this is a tool that gets better and self improves.
Um, and so I think that's part of why I think that we, um, you know, put [01:11:00] more effort than most, uh, at that.
Logan: you, you touched on, um, one of the characteristics that you hire for is curiosity. Uh, Is that one of the principal things that you want to walk out of a meeting having assessed, uh, from an individual is that like the canonical trait that you think of a ramp employee having like where does curiosity fit on the Maslow's hierarchy of, you know, whatever characteristics of a ramp employee.
Eric: Oh, I mean, I actually think it's very high and I don't think this is restricted to who's working on new stuff, right? Like I. At the end of the day, like back to the mission, save people money and time. If you want to save them money and time, well, every business is different and these are lots of different people.
You need to understand where are people spending money? Why are they doing that? Um, where are they spending time? And so I would argue if you want to provide, um, you know, build better tools, but frankly, even offer good service, you need to be fundamentally curious, you know, about who is this person at the start of it.
And so what I'd say is like, I [01:12:00] actually don't think there's a single job, uh, at ramp. Uh, in which, um, you know, curiosity isn't a, um, I think a quality that we look for. Um, uh, so I would say, uh, it's about as close to the top of, of, of my Maslow's fear of it, you know, as
Logan: do you, how do you assess that out in an interview or like, what are the, um, characteristics that you think I'm sure everyone describes themselves as curious, but if you're sitting down to a conversation. In a meeting are there certain things that you look for in curious individuals?
Eric: there's different manifestations of it. Maybe like a meta point just for. People building companies thinking about it. Um, an interviewing is, I think sometimes people try to do too much in single interviews where, you know, there's someone who's trying to cover all the bases. And that's like tough, right?
Um, I think thinking about, but. What is the overall, like, how many interviews are you putting someone through? How do you make sure that every interview is very different? Um, and you got a lot of companies, you know, um, people get relatively [01:13:00] similar questions across interviews. People aren't that coordinated.
Um, and, you know, great interview process. People are getting very different experiences across different interviews. Maybe with me, I'm curious. You know, maybe I'm playing like, what's your motivation, you know, um, today and I'm trying to understand kind of desires, your story, how did that meet? That might be one version of interview that I give, uh, maybe you talk with Kareem and he wants to see how you're doing under stress and maybe he'll like disagree with you about anything and see how you handle it to others is more.
Can you do the job right? And so what I'd say is somewhere in there, there needs to be someone thinking about these different traits. I wouldn't try to like combine everything. I would try to have very rangy interviews where everything feels different. One of them should be about curiosity. Not necessarily all would be would be first.
Um, second, when I think about curiosity, you can come in different ways. Um, you know, I think it can Some of this is not just what people say, but what they've done. You know, show me what you've built. Why did you build that? Um, what was the need that you were, you know, trying to solve against, um, who inspired you, uh, in order to do this?
And so people kind of, you know, you know, [01:14:00] demonstrating curiosity in, in their thought process, but even, um, you know, the, um, if they're a builder, do they have the curiosity to use new tools, um, to try things out in, in new ways? Um, for people who, uh, let's say are less in, in, in a builder mode, but maybe organizing resources, you'd say, tell me about a project that, um, you know, uh, you feel was particularly successful.
Like, why were you working on it? Um, what happened? What were you specifically responsible for, for other people? Did it work? How'd you know it worked? You know, and then you ask a follow up question after they kind of go through all that. Most people can't kind of give all the details and maybe focus on one.
The great ones, you know, you know, can, can go through every aspect of that. Um, Um, with a high degree of specificity. Um, but then I think where you can sort of pull out curiosity of that question is great. Um, what would you have done differently? Um, uh, what worked beyond kind of your expectation? Um, and um, you know, what went wrong?
Like, how would you approach it? And if you see people kind of who, they're not just thinking about it for the [01:15:00] first time, but have actively wrestled with this. You know, um, even in the project, they're most proud of like there was something they could have gotten better. Um, I think is what you, when you start to see real curiosity emerge and, and, and what I would say about, um, you know, I think great, um, crafts people, artisans, people doing.
I think a lot of great foundational work. They're, they're never truly satisfied, um, with, um, you know, even on their best day, the, the best finest work that they've produced, there's always something deeper that they could have done something better, some level of improvement. And so it could have been the craft of selling could have been how they organized the product and what they built.
Um, but there's many different aspects, but I think a lot of it often resembles around, um, You know, um, uh, what did they build? Why? What went well? Um, you know, I think another simple way is, you know, uh, uh, what are you reading? You know, tell me your, you know, your three favorite books or the last three books that you read.
Um, I'm not saying books are the only medium in a way to like learn stuff, but like it's, you'd be surprised how many people don't read [01:16:00] anything. Um, um, and, um, you know, I think just sort of asking people just about like actual, not like hobbies, but like, you know, interest in where they've gone deep is, you know,
Logan: The, the, the point on the, um, the everything can be better in some way and the self flagellation or the really thinking about the optimizations, it's, it's such a good point. And I think if you look at, uh, some of the best artists across any medium, there's this high, uh, correlation with depression that exists for a lot of the artists and.
It's this desire. I think a lot of people have to never be satisfied and to keep pushing and it manifests itself in art, uh, and creativity in some ways that can be actually like psychologically problematic for people because they look at what other anyone else would have determined to be a great piece of art or a great album or whatever it is, anyone else would have said it's done and instead they say it's not good enough [01:17:00] and keep pushing or keep iterating or keep working and don't release it.
Pass the point that is normal right in any way, shape or form, and not to say that, uh, you know, the best entrepreneurs need to be depressed in some way, shape or form, but I think there's I mean, if you look at some of the canonical examples of of our time, I think Elon Musk has been pretty public about his battles with with mental health, and there's this desire for perfection or incremental development at all.
Incrementally improving on everything that I think, um, some of the best employees have hopefully not to the extent of mental, uh, health crisises. But, um, yeah, it's an interesting point.
Eric: I mean, I think it's, I think it's real, um, I, I think that if you want to produce great work, create things that are really timeless, like that's a very, um, you know, so few people are really, you know, beyond, you know, sets of friends and family and people around them are really remembered. Um, and I think to, you know, build or [01:18:00] create or work on anything that actually is transcendent moves, you know, the.
You know, um, societies, human race, people forward in some way, I think that, that bar is so high, um, um, and I can understand, you know, You know, for someone who wants to have that kind of an impact, um, and that kind of desire. And it's so hard to know, will this work be timeless? Will it actually move forward?
Like I can see why, um, uh, you know, depression can be a really natural thing. And, um, what I would say is like, I think it's both true that dissatisfaction is, Um, I actually, I like, I think there's something both deeply unhealthy about it as you point out where, you know, um, one way of looking at it is, you know, as good as you've done, you'd always be better.
And, and so, um, you know, ergo, you, you know, you should always feel a little bit bad about yourself. And I think that's real. And I, I think that you're right. I think there's very few entrepreneurs who, you know, I think people usually say like an Eric, you're pretty smiley and chipper guy. And like I am, but like, I, I think like, [01:19:00] uh, yeah, there are moments where like, you You know, I think you can't help but feel this stuff.
Um, I think that the other, um, way to think about it is how profoundly disappointing would it be though if you truly made the best version that could ever be done in the history of the world of any particular thing and there was no surpassing it. And that was it. And there was nothing beyond it. Um, uh, and in some sense, I think that would be more depressing.
Um, and, um, more profoundly sad. And, and like, and I think there's this broader set of points. I mean, maybe even the circle back to how people think about, um, you know, various all, like, I'm not a futurist. I, I don't know. Um, you know, how things necessarily going to play out. And I, you know, I'm more interested in how do you help businesses.
You better today. How do you help people with better lives today? And that excites me. But like, there's, there's one version of the world in which I get so good. Um, there's [01:20:00] no job that humans can do better. Um, Uh, and would people, um, you know, live in some utopia where actually you can just hang out and never work and like, you know, to me that, that feels very profoundly depressing, um, uh, because I actually think, um, you know, for me, part of, um, I think we're, we're, we're, we're joy and, and satisfaction and, and part of what makes life interesting comes from, uh, is the ability to, uh, work, um, and produce something, um, for other people.
It's to improve your skills. It's that constant kind of pursuit. Um, it's, yes, I, I've, I've got, gotten, um, better and maybe produce something at a new height. Uh, there's still further to go, um, and I still want to keep climbing, and I still want to develop better tastes, and I still want to put out better products, and so, um, part of me thinks that even as tools and, and capabilities get better, uh, I think that's part of what actually what makes us deeply human, um, uh, it's the, um, it's the act of creation, it's the, uh, You know, delivery of service, it's [01:21:00] being there and tending to other people.
And so, um, I don't think those ever, ever go away, but, um, you know, I would say for anyone who's like stuck in the rut and is, you know, worried if things are ever better, like I, I think that's, um, that sort of line of thought has certainly been helpful, helpful to me and some others that I know.
Logan: it's such a profound point and I think it's, um.
Logan: I think it's fascinating that there's no hill you'll climb at least I've yet to find it and I've talked I've sat here I think this will be 110 episodes or something and I sat with some of the people that I think have built technology in different ways.
And I think one of the unifying themes throughout is there is no point in the journey that you're aiming towards that you reach some fulfillment that, uh, is like the canonical achievement. Uh, and the second you get to the top of one hill, you realize there's another [01:22:00] hill to climb or the, the beauty and the enjoyment, uh, truly comes from loving the process.
Yes. And. There's always going to be someone that built a bigger company or there's a new room to be a part of and I, I remember I always just. I wanted to be a general partner at a venture firm when I was, I don't know, when I, whenever I figured out what venture was 23, 24 years old.
Eric: Yeah.
Logan: And then I got
Eric: You got there,
Logan: and I looked around and I was like, well, what I'm, I'm here.
Like I didn't, uh, now what do I do? Right. And it's a funny thing that, uh, there's this nostalgia I have, uh, still for like the process of getting to that climb. And now there's new hills that I found to climb, but the achievement in and of itself wasn't. It was actually the journey along the way. That was the true enjoyment.
I'm sure you feel that with your, your company, each incremental fundraise or [01:23:00] an IPO, when that day comes, um, it'll be a nice checkpoint of celebration, but the, the actual inputs in the process is where the true enjoyment comes from. And if you don't like that, then you're not going to find fulfillment. I don't think at some end state.
Eric: I, um, I mean, I think that's really profound, and I think that's, that's really true. It's, uh, uh, I think it's both a, I, I think it's like an amazing, like, divine blessing of, of, like, this hedonic adaptation, like, you know, you finally get what you want, and, you know, there's, in some ways, maybe no better feeling, and then, Well, you know, it's a few days later and you're like, what's next?
Um,
Logan: insatiable.
Eric: um, and I think it's both, um, uh, in some ways it can be like really sad and deeply depressive, but also I think it's like one of the, uh, one of the great things, you know, it's, it's a pursuit
Logan: Well, and I think that's a good, uh, lesson or takeaway for anyone in any job that like, if you're not [01:24:00] enjoying the inputs, and I know you guys focus more on inputs than outputs as a company.
Eric: yeah.
Logan: If you're not enjoying that, then that might mean you're in the long line of work or you have the long, uh, the wrong perspective, uh, of what you're doing in some way, shape, or form.
And that might be a call to action to go do something else because there's not some end state that I found anyone that's particularly motivated gets to that. They're like, okay, now I've made it. I can hang out or I feel totally fulfilled from this journey along the way.
Eric: Yeah, for sure. Um. It's uh, and it's interesting too.
Eric: I mean, I, I think, um, I dunno, I, it's been fun learning from lots of other entrepreneurs, but like, I, I think one of the, um, you know, to me kind of movies and documentaries that I like learned quite a bit from and, and has informed just like even how we build at, at, at RAMP, but, um, you know, I think about like, the way You know [01:25:00] careers and in some ways like I wouldn't go so so deep But like I think it's just worth worth an hour people's time like I've I don't know we were talked to this Have you ever watched hero dreams of sushi?
Jiro J IRL and I think it came out a decade a decade and a half ago, and it's about
Logan: um,
Eric: Um, uh, I think his name is Jiro Ono. I think he, um, was the first, um, sushi chef in the world to get three Michelin stars. And he started this, this restaurant that's in, you know, the, you know, uh, the basement of a, of a subway station.
Um, and, um, Uh, it's a documentary about him, you know, in, in the years of, of what his life was, was about. And, um, it, it's very fascinating. I think at the time he was in his eighties, um, and still working extremely hard.
Logan: What year was this?
Eric: Um, I want to say this is 2010, 2011, somewhere in
Logan: contemporaneous. It wasn't, I mean, it, it looked back, but [01:26:00] it was
Eric: It was about him. Yeah. It was about him and, and, and his process and, you know, running the restaurant and it's not long, it's an hour, 15 minutes.
And, and by the way, he's, he's still working. Um, I think he's in his late nineties now. And he is still, um, I mean, every night or most nights of the week still showing up to work, um, and trying to make things better. And, and there's, it's an amazing documentary, um, it's worth watching the whole thing but there's this kind of amazing scene I, I, I find it's like a lot of lessons for, for people.
I'm building, um, startups, um, and, and frankly working at anything, any, any kind of craft. And the narrator basically asks him this question, something to the effect of, you know, what does it take to make great sushi, you know, and to make great food? And he answers this and. Uh, sort of this, this, this, this odd, but, but actually, um, really obvious ways as well.
Um, you know, to make great food, you need to eat [01:27:00] well. Um, you need to eat and, and, and enjoy great fish. Um, you know, and if you don't develop great taste. And your sense of taste is lower than that of a customer. Um, then how are you going to impress them? Uh, and then he goes on and he talks about like, is the guy he buys rice from and he's like, Oh, here, Amici or whatever the guy's name is.
You know, you know, I, I only buy rice from him. He knows more about rice than anyone on the planet, you know, and he's going on and moves into the scene. And, and, you know, one, I think it's like this question of, of, I think it's great for crafts people. Like if you don't actually like, Use the product, enjoy, understand the problems better.
Like, how are you going to make something that deeply solves something in a new way? And I think that most people who like market products, build products, you know, talk, don't really talk to customers nearly enough and they should do that more. But I think that the other thing, you know, along the way, you know, as they talk about, um, what he's trying to do and he said, and he, he basically says he, he's been making sushi since he was like 11 years old, something like that.
And so. You know [01:28:00] now, um, almost 90 years of everyday showing up to do the same thing and he talks about he's still, uh, trying to climb this mountain and you know, as high as you go, you're still never going to quite reach the, the peak. But, you know, it's the pursuit of getting closer and it's there. I, I think it was one of those things, it was filmed in this interesting way where people were interested, not just in, in the quality of what was coming out, but who are these people that are, that are developing it.
And, um, uh, Um, I think it's, I think it's great, but, uh, I think, you know, sometimes people look at him as he's such a stern, oppressive guy, but like, I, I think he's, um, I think there's a joy to what he's doing, and you can see it clearly too.
Logan: Well, and it's to hear in this sort of ties back to hiring people into an organization and on this curiosity point to hear you talk about the history of double entry accounting or we did a blog post on the history of dive rank and the implications that came for that for the fintech industry. You can find graft or curiosity and literally anything like I double entry accounting, uh, might be one of the most arcane [01:29:00] mundane topics I think out there, but like there's a level of curiosity in, in your craft and what you do that caused you to go back and actually learn about the derivative impacts of all of that.
And I guess, uh, I've. I don't think I grew up. I certainly wasn't like, uh, genetically wired to be fascinated by the venture industry and technology. Like, I don't think I even knew venture was a thing until I was 23 24. But like, I've enjoyed learning about the history and all the different people and the players and, uh, you know how all these things came to be.
And so. It really can be anything it's it's really just finding leaning into and enjoying that and I found that when you interview and I'm sure you've seen the same thing
Eric: can be associated with these
Logan: can just go so deep in these things and talk about them forever are the people that
Eric: can't talk
Logan: persist within highly functioning organizations and they take such a pride in [01:30:00] the part of the iceberg that never actually gets And if you peel it back a little bit, then they can keep going deeper and deeper, but you know, to your earlier point about like the simplicity of ramp, there's a lot of underlying infrastructure that exists to expose something very simple at the, from a functional standpoint for the end customer.
Eric: And it's totally right.
Eric: Like, I, um, I think one articulation of, like, really the fundamental problem that startups are going up against that I love was, I think, Justin Conn, um, one of the co founders of Twitch and, um, partner at Y Combinator years ago. He basically said, you know, um, Startups biggest battle is people not giving a shit.
Um, and he's totally right, right? Like when you think about kind of like your own life, like you get a lot of stresses. It's like you want to generate like returns, you know, um, you have stuff going on in personal life. Um, you're trying to get better. You're trying to find the next deal. There's all kinds of stuff going on in some random product marketing deal is like not high on your [01:31:00] list of priorities, right?
And so you need to be both simple, interesting, direct, and compelling enough that in a period of about five seconds. You know, I can somehow crack into like it is worth some remnant of your time to maybe look deeper into. And so I think it's both deeply important to remember that, um, you know, in the creation and sales marketing of different products, you want to be very simple, very compelling, very clear.
Uh, and I think for most people, uh, ramp is and it feels like, and, and, and, you know, I, I think this is, this is fair to say, like, It's easy, it's intuitive, you know, and it's a, you know, spend management software that helps your business run better. Um, it saves you time and money. And for most people, that's all you, I think you need to think about.
Um, um, you don't need to think about how it works. You should just know that if you put this in your organization, it's true. You can expect that. If you're like the average company, your company will spend 5 percent less. Um, we work hard at that and we're going to go and do that now in order to create this simple product, that's [01:32:00] easy to understand, easy to consume.
You need a deep amount of interest in like. What causes companies to spend time? What causes them to spend money when money moves? How does it move when it's accounted? How is it accounting for? Where did it, you know, what is the history of accounting? Why are things done that way? Um, is it obvious that things should be built this way and should we be building a car, not a faster horse?
And so I think for the practitioner, there's no way if you want to build something better, you have to go deep, um, and you have to do this. And so I think it's both keeping in one's mind, um, you want to be able Um, articulate and, and, and craft and share this very simple version of the story. Um, but, um, both in, in how you build and also how you motivate, um, people who are working to want to go so deep, um, that you can, um, find ways to, you know, um, you know, to use the kind of, I think Elon Musk is, um, right, like, you know, the best part is no part.
Um, instead of building something with 20 parts, can you build it down to two to one [01:33:00] to. You know, that, you know, that, that's it and keep it really simple. And if you don't really understand the function of why certain parts are there or were built in this way, how are you going to reduce it? Um, uh, and so I think that's part of the fun of building companies.
Um, you got to think about both.
Logan: Well, Eric, thanks for doing this. Leave it here. Uh, that was very fun. Uh, we'll, uh, we'll need to do round three at some point, some point soon.
Eric: Thanks a lot, Logan. I, um, appreciate your friendship and investing in supporting companies since the early days. And I hope this was interesting to listen to.
Logan: I, listen, we, I enjoyed it. Uh, you know, I don't know if it's, uh, I think people will like it. So thank you. Thank you for joining this episode of the Logan Bartlett show with co founder and CEO of ramp Eric Kleinman. If you enjoyed this episode, we really appreciate it. If you subscribe now, whatever platform you're listening to us on as well as shared with anyone else you think might find it interesting.
We'll see you back here soon on the next episode of the Logan Bartlett show. Have a good weekend, [01:34:00] everyone.