Transcription

• Speaker #0

  Attention is all you need, that famous paper that kind of brought LLMs. I think it's true for marketing. It has come to the point where truly attention is all you need. It's literally just about getting someone's attention, getting them to stop scrolling, getting them to look up from what they're doing. That has become a large percentage of our job.

• Speaker #1

  So I would like to know how you work with all the team. Who is your manager? How is it organized?

• Speaker #0

  We don't have a traditional CMO or a head of all of marketing like some other companies do. But yes, I reported to the CTO.

• Speaker #1

  Why do you think it's... relevant for a growth team to report to the CTO?

• Speaker #0

  Having the growth organization as part of a more technical department meant that we have more technical resources. I think AI has really commoditized a lot of like the task execution. And so the job really becomes like solving problems. On the growth team, that's very much expected. If you come in to a new role and you just have like five successes in a row, you're probably not trying very unique or very innovative things. I don't know if my role exists in five years. I brought this up to a few folks, including my boss. I don't think he disagreed.

• Speaker #1

  Hey, George, welcome to the show.

• Speaker #0

  Thank you. Thanks for having me.

• Speaker #1

  So, George, you are leading growth at Ramp, which is one of the highest growth. This is incredible, the story that you have. I've got a couple of macro questions, let's say. to discuss before diving into the topics. But one, I would like to understand the kind of the organization that you have, because I understood that growth is sitting not underneath the marketing team. And so I would like to know how you were with all the teams, like who is your manager, et cetera. Like how is it organized?

• Speaker #0

  Sure, no problem. So I would probably describe marketing as a little bit more of a distributed function at Ramp. than at least most companies that I've been at. So growth is responsible for pipeline, SQLs. And that's my organization. So it's things like performance marketing, SEO, lifecycle, events, the things that you would typically associate with driving pipeline, as well as some that maybe you would less so associate with driving pipeline. So we have like a verticalization initiative. We have like a designer on the team, that sort of thing. But that's growth. We have also product marketing that actually reports up to product. We have a phenomenal comms team. They do PR, AR, all of the traditional comms type activities. They report to the CEO. So it's just a little bit more distributed. We don't have like a traditional CMO or like head of all of marketing like some other companies do. But yes, like I reported to the CTO. As of this morning, we announced that he is now co-CEO. Yeah,

• Speaker #1

  I saw the announcement.

• Speaker #0

  Yes, it's actually not much of a change internally. They've always acted, the two co-founders always acted as true partners. But yes, it's always interesting to say, at least in the past, that growth reports to the CTO.

• Speaker #1

  Yeah, and I would like to ask a couple of questions around this, because this is actually the first time I see growth reporting to the CTO. So why? First, why? And second, why do you think it's relevant for a growth team to report to the CTO?

• Speaker #0

  Yes, it's a good question. I think... One of the reasons why I joined Ramp and why I always really respected the culture is they have a strong philosophy that everyone is a builder. Everyone should be technical. Everyone should be building regardless of your function. And so I think actually it was pretty natural to eventually, like once you reach a certain scale for growth to be part of a more technical organization, especially in the age of AI, because everyone should be building things. And that's like a whole separate topic we can we can get into. But I just. really appreciated that number one, it's a philosophy of builders. And number two, having the growth organization as part of a more technical department meant that we had more technical resources. I think in the past, marketing has always struggled to fight for engineering resources when there are products to build and there's other higher priorities. But now that distribution has become, I think, far more important for most companies, and product development has accelerated in some ways been commoditized, at least in some markets, distribution matters more. And being able to move more quickly and build products and be more technical, I think, is actually a huge competitive advantage. So I think Ramp saw that earlier than most. That was one of the things that attracted me to the company as well.

• Speaker #1

  Okay. I understood you have a science background, correct?

• Speaker #0

  I do, yes. I studied biochemistry, but then mostly worked in analytical chemistry labs and synthetic biology research.

• Speaker #1

  Okay. So you have a... technical background, maybe not the type of technical background that is usually related to growth. But I feel, you know, OK, I completely understand the fact that you have more technical people inside your growth team because you're reporting to the CTO. Still, one of the challenges is to have kind of the business mindset inside of the technical aspect of the job. How can you reconciliate the two things? Because you come from this science background. And I want to dive into also your test and learn approach, like kind of the scientific approach that you have into growth. But yeah, how can you mix the kind of technical aspects that is expected from your... department plus the business mindset.

• Speaker #0

  Sure. I've always kind of had the philosophy that you should have a healthy percentage of a growth team or at least a marketing team be folks with some quantitative or technical background. And it's actually less because they can build things and more because they're generally forced to be good problem solvers. If you're an engineer, a scientist, you've worked in finance, whatever it might be, You've had to solve. problems in a very analytical way and sometimes from first principles and i think that's just a lot of like the day-to-day of growth like if you really break it down it's either solving a problem which is generally like novel and requires some like trial and error uh or it's just executing tasks and i think like this is another topic we get into like i think ai has really commoditized a lot of like the task execution um and so the job really becomes like solving problems and there's probably a few maybe three classes of problems to really solve in growth but you kind of solve them all to experimentation. So I think they kind of go hand in hand.

• Speaker #1

  Okay, interesting. And so what does treating growth like a science actually change in your day to day? Like in how you hire, like the way you run your week, you decide working, et cetera. How do you manage this?

• Speaker #0

  I think the biggest thing, which I think is obvious when you say it out loud, but it's really hard to put in practice, is just hiring people that like want to fail and are okay failing. And everyone says this. We hear this a million times. what that really means is you have to be okay with someone failing like nine times in a row. And as long as you're learning something from those failures, and when I say learning something from those failures, it doesn't mean that you just go slap some stuff out there or throw some stuff out there and call it an experiment. That's not an experiment, you have to put some thought into it, there has to be some experimental design, you have to put a lot of work and effort into how you're going to measure success. But as long as you're doing that, if you fail nine times, 10 times, 15 times, like I think that's okay, because those those learnings ideally are compounding. And every iteration of that experiment is getting better. And you're finding things that will work eventually. So that's the number one thing is finding people that are okay with that. And I think most people are actually like not okay with that. It's very uncomfortable, especially if you're in a new role or if you're new to the company to come in and just start failing over and over again. But on the growth team, that's very much expected. If you come in to a new role and you just have like five successes in a row, you're probably not trying very unique or very innovative things. So that's the first thing that I probably look for specific to growth. But even more generally than that, I think Ramp does a really good job of hiring for a spike, like really identifying number one, what you need in a role, like what is the one thing that you really, really need? And then finding one person that that is the one thing they're the best in the world at. And if they are weak in every other area, like I'm okay with that. We just will hire for those weaknesses rather than trying to find someone that's a little bit more average across all aspects.

• Speaker #1

  Nice. Going back to what you said just before, like being comfortable with failing, I think this is something that most managers need to feel comfortable about, you know, because at the end of the day, I feel people, they fear failing because of the way they will be treated by managers, you know. So I feel like for people who wants to apply to Ramp, I think it's a really good thing to hear what you just said, because it means that you are comfortable with this with the people inside your team.

• Speaker #0

  Yes, that's exactly right. Like the, we do employee engagement surveys every, every couple quarters, like most companies our size probably do. And the number that I'm always most proud of, and the one that I look at most closely, is a question around, do you feel safe taking risks? If the team doesn't feel safe taking risks, then we are going to fail as a growth team. And to your point, that is purely from your manager, from your manager's manager, like, it is on them, it is on me, to make sure that people feel comfortable taking risks and failing. You're completely right.

• Speaker #1

  Nice. Okay, when I work with growth teams, usually, you know, the way they prioritize experiments is on, you know, usually this is the ICE framework. Like, usually it's the impact and effort. How do you evaluate? How do you weigh the type of experience that you do?

• Speaker #0

  It is exactly that. Impact, effort, confidence. The only thing I would add is time to results. Ramp is incredibly high velocity. And I think that's actually very important in a world where technology and AI is accelerating all things. So time results really matters. If someone comes to me, they want to run an experiment and do the sample sizes or do the experimental design, we're not going to learn something for like four months. That's a terrible experiment. Like, I have no idea what the world's gonna look like in four months. uh we need to scope that down to something we can learn in like two weeks uh and even that might actually be too long for for many things so time to results including that in the ice framework i actually think is is incredibly important uh and then the other is is confidence like what's your confidence that something's going to work i think people generally overlook how important that is um if you're going to run a really tiny experiment but you're incredibly confident it's going to work you should do that if you're going to run a really tiny experiment but you're very unconfident it's going to work you probably shouldn't do that it's probably a waste of time I think that. What most people that aren't scientists probably don't realize is how much of an art science actually is. Like I worked in a chemistry lab and I would say my fellow chemists were some of the most superstitious people I've probably ever worked with because there's so much that like you just can't account for and that you don't know why something went wrong or why something went right. And that's why I think like confidence in your experimentation is actually like very important. You probably have some intuitive gut feeling on like what will or will not work based on years of experience or past experiments that you can't... put into words. And I actually think that's worth something that goes back to hiring people that that are spiky. Like I like to hire people that have a good intuition and are right a lot, even if you can't quantify it.

• Speaker #1

  Yeah, I completely agree. Usually when I work with teams, and they say like, hi, should I prioritize, etc. And they don't want to work with big, big frameworks, because they're a small team and blah, blah, blah. I'm always saying like, guys, your gut feeling is usually right, you know, so you should just follow your gut feeling, Like, even though it doesn't sound like it's scientific enough. Like, really, it means something, you know?

• Speaker #0

  I couldn't agree more. Like, I don't know. I think that as the growth function was kind of growing in importance or popularity over the years, we may have over-indexed too much on what is measurable. But the quantitative is just as important, which is crazy for me to say as a scientist. Like, most of my career... I was about to say this.

• Speaker #1

  Like, it's really good to hear from a scientist. And so, okay, you launch your experiment. and now it starts working but on some channels you know you can reach the ceiling pretty fast so how do you manage like evaluating and working with uh evaluating the ceiling reaching the ceiling like how do you manage this it's

• Speaker #0

  a very good question um i think a problem that most people and companies face is that they're afraid to like overshoot the ceiling they're afraid to scale up so fast that they waste money because they've hit the like asymptote of their saturation curve. And I think as a result of that, they waste a lot of time. Like if something's working, the tendency is like, all right, great, let's, let's do 50% more, let's double it, let's triple it. Like you should do 10x more, like you should saturate that as quickly as possible. They're like the when you find something that works, that is a relatively ephemeral competitive advantage. And so you should maximize it as quickly as possible for as long as possible. So what that means more tactically is like okay we found something that works it's like take it to the max i want to know where that asymptote is next week uh and if we waste a little bit of money finding it like that's fine uh you could probably do the math and realize that finding that asymptote fast versus like testing your way to it over months or quarters like the roi is probably higher like waste a little bit money and finding it fast um now that's easy to say if you're a larger company that has like money to burn uh when you're an earlier stage startup that probably means something different, but it also means probably something different because like you have more, you may not have as much money, but you probably have more freedom to try crazier things. So you're probably able to find something with alpha and advantage that a larger company might not have the risk tolerance for. So it probably cancels out.

• Speaker #1

  Okay. Though you're a really creative company, even though you're really big. And so do you have like something that is kind of absurd sounding? in terms of channel that you would like to try soon or that you've tried recently?

• Speaker #0

  There are a lot of very crazy things that we've tried recently. A lot of them we discover on Twitter. It's kind of interesting how Twitter has become like the place where crazy ideas are shared. And like a lot of the times, I can't tell if they're fake or not. Like people are just posting crazy things for clout. But there are interesting things. Like last week, there was something that went semi-viral on Twitter. It was called, I think... I don't remember what it was called, but basically it was a side project for someone and you could install it and in Claude code, it would show you ads. So instead of Claude saying, thinking, percolating, whatever, it would show you a little ad. And the incentive for someone installing this would be they'll split the ad revenue with you. And it's like a pretty genius, if not like very old business model. And it like totally blew up overnight. And like the moment I saw it, I messaged the teams like we should advertise on this. And the team like got on it, we became like one of the largest advertisers within a few hours because we wanted to test what it was like. And that's a crazy channel that didn't exist two weeks ago. And similarly, it's like, let's maximize it. How quickly can we juice this thing for as much as possible? Now, is that channel, is that marketing tactic going to exist? a year from now, two years from now, like, probably not things just change really quickly. But we got something out of it very quickly. And I think that the reason why I share that as a crazy channel is because those types of things are popping up like every single week. And like, they don't even cross my mind as being crazy anymore. But they kind of seem that way. Like this is one guy's side projects that we found on Twitter that like, they didn't didn't have a website, like there are no terms and conditions, it was just like, put a credit card in and your ad will start showing to random people. That was worth testing. So that's maybe one. um we are starting to do i think some like really innovative and interesting things with like events uh i was talking to to our person that leads um brand and enterprise marketing uh she's she's brilliant she was like i met someone that uh that like is a broker of fossils like literal dinosaur fossils uh okay it'd be really cool to do an event where we bring people to like a fossil dig and i was like yeah we should totally do that so So. uh more unique events like no one wants to go to a dinner anymore like you can you can go buy dinner yourself i think that's a little bit played out but like go into like a i don't know an archaeology site that'd be pretty cool i agree yeah okay well and and i see also

• Speaker #1

  what's what's interesting in this case is the excitement that you have with this kind of crazy idea like there is no link between your business and archaeology, but still like you. you think the idea is pretty cool, you know? And so I think this is, this is at the end of the day, what's creativity in business, you know?

• Speaker #0

  I, yes. And like what I, um, I keep telling my team more recently is, uh, attention is all you need that famous paper that kind of brought LLMs. I think it's true for marketing. Uh, like it has become to the point where truly attention is all you need. Like, I don't think you need to be as precious about your messaging and positioning and your targeting and and all that you don't need to really think through your funnel as deeply as maybe you used to in the past, because things are moving so quickly, and people's attention spans are so short, that like, it's literally just about getting someone's attention, getting them to stop scrolling, getting them to like, look up from what they're doing. That has become a large percentage of our jobs. And if you can get their attention, then it's a lot easier to get them in the funnel and help them understand. But like the hard part is truly getting attention. So if a dinosaur dig gets someone's attention as an event invite, like we did our job, that's what matters more than anything.

• Speaker #1

  Nice. Is Outbound part of your team?

• Speaker #0

  It's a good question. So the SCR organization, which leads Outbound, is not part of my team. The tooling to enable them to do effective Outbound was part of my team. It's now part of growth engineering as of about a quarter ago. Because they had built such phenomenal products that it was time to really productize them and take them to the next level. So that's like a pod of, I don't know, four engineers, two full-time data scientists, a PM. Like it's become a large operation. because outbound is number one a big channel for us but also because we've seen such success with with our internal tooling to make that team more efficient and effective okay so it was part of your team and now it moved to the other team correct okay so you still have some background around what's been tested etc etc i'm really interested because i know that

• Speaker #1

  outbound is a really big level of growth for you and i also know that you are kind of the at the avant-garde of how... to use AI and agents, etc. with outbound. And I'm curious to know the type of iterations that you went through and now what you see is working versus not working with outbound and AI because now everybody is saying, okay, outbound is kind of dead and the conversion rates are really low and it's getting to lower and lower every quarter. And you've got these new AI SDRs, but then you... you realize it's AI on the phone and you know, this type of things. But still, I feel that you on your side, you've managed to, you know, make it work.

• Speaker #0

  Yeah.

• Speaker #1

  And so, yeah, it's a wide question, but I'm interested to get your feedback on this.

• Speaker #0

  Yeah, it's a very good question. I mean, I was the PM that owns our outbound tooling that moved from my organization to growth engineering. Like, he's brilliant and he's very good at this. And we've talked a lot about what the future of outbound looks like. The way I'd probably summarize it is when early stage companies ask me, hey, like, what channel should we test? Should we do outbound? I'm like, no, I do think it is dead. But it is the poor quality, high volume outbound that is dead. Like AI has completely commoditized that. Like there are some industries and verticals where it's probably not dead, like folks that are less online and like less marketed to probably still works for them. But in general, the poor quality is dead. What's not dead is, like... high quality human interactions. And I think the easiest way to do that is via like, like field marketing, field sales, like go out and meet people in person. But the second best way to do that is probably with really, really good high quality outbound, like something that again, number one is going to get someone's attention, like they have to like see it and will be willing to open it. Now once they open it, it can't be like AI written, or at least seem AI written, like it has to be very specific to that person's pain points to who they are as a person to what they care about. And doing that, if you're an early stage startup, is like borderline impossible. Like, there's no way some early stage startup is going to put five engineers, two data scientists and a PM on building like a machine to do good outbound, like in the same way that Ramp has. So what does that like more tactically mean? The PM that built this, he posted on Twitter maybe six months ago, five months ago about the tool that he built called Ramp Revenue. And the way that we iterated to that was just like sitting down with like our SDR team and observing what they do and where they spend time. and looking at the differences between a great SDR and an average SDR and being like, what is the difference in their day to day? And then slowly solving like very, very specific problems. It's like, okay, this person's doing a lot of research on a specific person at an account. How can I automate that research flow and that specific flow? Not just like, oh, I'm going to scrape LinkedIn, but like that specific flow of what are they looking for in someone's profile? It's like, okay, great. I've automated that. It's like, ah, they're spending a lot of time copy and pasting between windows. It's like, okay, I'm going to bring it all into one tool. Oh, they're spending a lot of time like actually like taking a template and customizing that template i'm going to build something that customizes that oh they're spending a lot of like a lot of their deliverability is getting caught if you email this certain certain domain i'm going to build like a spin tax tool that changes words at random with synonyms so that it doesn't get caught like and the next thing you know you've got like 15 different point solutions and it's like okay let's wrap this into one tool and if the sales team uses it then it's a good tool if they don't use it then i didn't build a good tool for my customers. And so really bringing that product. management mentality to building tools where if the SCRs are internal customer, either use it or they don't, we're not going to force them to use it. And if they use it, then you've really built something valuable. But the way we got there was really just like very specific point solutions, with the end result being, if I'm on the receiving end of one of these outbound emails, like, it is very, very specific and valuable to me, it actually doesn't really matter if it's written by AI, not as long as it's valuable to me. And I think a lot of our outbound is getting there, like we have a lot of work to do still. But that's at least why it still works for us as a channel. compared to the very generic off-the-shelf stuff.

• Speaker #1

  Yeah. It gets back to the first principle and a lot of pragmatism, you know? It's exactly what I'm saying to some people on my team, you know, when I'm always saying this to someone from my team, like you don't need to build a coffee machine if I just need the Excel file with the email addresses on it, you know? Like...

• Speaker #0

  Yes.

• Speaker #1

  Just get me the Excel file and that's it, you know? Like it's pretty easy. It's going to take you 10 minutes. instead of building like, you know, the rockets that goes to the moon and blah, blah, you know.

• Speaker #0

  Oh, you're so right. You're preaching to the choir, so to speak.

• Speaker #1

  Yeah.

• Speaker #0

  I like, I was telling someone last week that, especially with AI, everyone, because it's so easy to build things and it's like.

• Speaker #1

  Yes.

• Speaker #0

  It's like, it feels fun that all of a sudden everyone wants to go and build a cathedral. They want some like masterwork of art that's gigantic. And I have to remind people, it's like the folks that started building cathedrals died before they were done. Like.

• Speaker #1

  it takes a really long time and it's really hard like you should solve something today versus for six months from now yeah and and and i can be impressed by the fact that you shipped a rocket that can go to the moon but this is not what i need you know exactly exactly uh all right so again kind of a macro question but um i know you're trying a lot of things and i would like also to address the AO slash GEO topic, but how do you see... demand and demand gen actually move in the next month, years? What's the trend that you see? What's your vision?

• Speaker #0

  Yeah, I think, I don't think my vision is that unique anymore. I think everyone is starting to see at least what we started putting in place a year, year and a half ago, which is AI is kind of, or marketing is kind of like trifurcating in the three buckets. There's stuff that's immune to AI. You should do a lot more of that. There is stuff that is being disrupted to AI. You should stop doing that. And then there's stuff that's not possible with AI that wasn't possible before. And you should figure that out pretty fast. In the first bucket of stuff that's immune to AI, it's two things. It's like high trust, personal interactions. And that's usually going to be like events or field marketing. And it's like creativity. It's like if you want something totally off the wall, very creative, like I'm going to trust a person with good taste over AI, at least today, like and I think for the foreseeable future. And we and as a result, we've done a lot more of that, like our events team has gone from two people to 25 people, because we're really investing in high trust in person interactions. And similarly, like, we do a lot of crazy stunts and marketing campaigns, because we want to we want to get attention. We think that's a valuable resource. Now, the second bucket of stuff that's being disrupted by AI, it's very traditional, low quality content, low quality SEO, like all that stuff is dead, like that's a losing game. Now, the high quality content is a different story. But I think a lot of companies especially early on, they're doing too many things and they're spread so thin that they put out a lot of low quality stuff and there's no ROI on that anymore. So that's the second bucket. The third bucket is really the most interesting. It's like stuff that's now possible with AI that wasn't possible before. That's changing like every couple of weeks. And it's like, it's really funny. I was having a debate with someone where I wanted to test something. And they're like, we already tested this. It didn't work. And I was like, we tested it three months ago. Like, look at all of the stuff that's changed in three months. We should test it again. We tested again. It worked. And like, that's actually kind of frustrating both as a leader and I'm sure as an IC. like I've already done this. Like how many times are we going to redo it to see if it works? But that's kind of the treadmill we're on now with AI. But that's kind of where I think things will continue to go in that broad direction for probably another year or two. And it just means that the velocity, like the pace that everyone has to work at is higher. And if you're not investing the technology to be able to handle that pace, then like it's going to be a hard job.

• Speaker #1

  Yeah, I completely agree. You know, have you heard about VivaTech in Paris?

• Speaker #0

  It sounds very familiar, but I don't remember what they do.

• Speaker #1

  Well, yeah, it's a huge event, just like Cannes, but for startups, etc. And I was giving a talk at VivaTech last week. And the topic of my talk was how to reorganize your team now with AI. And I said, for me, now you should reorganize your team into three buckets. Like all the jobs related to relationship, events, etc. All the jobs related to creativity. And then what's... what can be automated. And so it is exactly the three buckets that you mentioned. Yes. And then I said that there has to be like a layer of quality, QA on top of all these three buckets. But for me now having like, you know, really small teams of paid managers and blah, because you've got agents, et cetera, really for me, the three main buckets are relationship, creativity and agents, AI, engine. And this is what I call the engine. And I... I think that's it, you know? And so it's pretty good to see that you've got exactly the same vision.

• Speaker #0

  Yeah, 100%. You are so spot on on that. And actually, thank you for the reminder. I actually did watch your talk. You said a few other things that I so strongly agreed with. Like the other one that stood out since you reminded me was like the junior talent pipeline trap. Like the things that I think should be automated does not necessarily mean that like, or the things you can automate doesn't necessarily mean you should. And like the one thing that like I will not automate is something like handicaps someone's learning, early career learning. And I mean, I'll use an example. I remember at my time at Samsara, we would hire folks fresh out of school, very early in career, and we would put them on the performance marketing team and we'd make them run $15 million in Google ad spend with no experience running Google ads before. And that sounds incredibly risky, but we would do it because like you understand very quickly how things are going. There's a very steady baseline. There's very quick feedback loops. Like that is like, that was a learning ground. I would not automate like at that time, like if AI was existed, that's something I would not automate, even though it's very easily automatable, because like, that's, that's what we use to test people to help them learn to get them like, to their next step in their career. I think that's a little different now. Like, I actually think we should automate Google ads. And we've done a good job with that. But yeah, you you need to, you need to provide the opportunities for folks to grow your future leaders to grow. Beyond that, you should automate.

• Speaker #1

  Yeah, I agree. All right. So let's open the topic of AEO and GEO, because I know you've tested a lot of things and you're still testing a lot of things. For me, one thing is all the people involved in this topic, they tell you like you need a kind of a proprietary data to make sure that you appear on ChatGPT or Google AR Review, etc. In your case, you've got a rare data moat, which is not the case for many companies. And so... if you were not working for Rump and that you didn't have all the data that you have today how would you approach this topic?

• Speaker #0

  It's very good you've got to get it somehow and that could mean either yes you actually need to go find some proprietary data that you have some insight in from your customer base or it means you need to go like create some running surveys doing like fundamental research or it means that you need to find some data that already exists out there and we need to make it digestible and understandable by an llm um that's kind of the order of of or like reverse order of difficulty the third thing is the easiest thing to do um if you have truly like no data no nothing unique or specific to share then like okay you can go repackage something that is unique but is very undiscoverable and uh undigestible now that requires kind of going back to the beginning of our conversation that requires having someone that can think in first principles and like really break down the problem and then be technical enough to figure out how to how to make it digestible. But a good example is, um, I'll give an example from the past. There are a lot of databases and there's a lot of like data that is regulated by the government in the United States, but at like the state level. So each state might have a different database of information. It's pretty, and they all have different systems. There's no standardization. That data isn't proprietary. It's out there, but it's probably not being digested by an LLM for a variety of reasons. And so like, if you truly have no data, you could go and you could structure that information and you could do all of like the AEO optimizations to make it digestible and you could figure out the... queries it's relevant for and spin up thousands of pages that answer those queries with this proprietary like you could do that if you truly have no data but my guess is that most companies actually do have some really interesting data and a very like unique perspective to share um they may not have just like thought about it hardly interesting okay what

• Speaker #1

  is completely overlooked on this on aeo geo topic oh that's that is such a good question um

• Speaker #0

  So I think what's really overlooked is how quickly things have moved from like marketing to humans to marketing to machines. And this is something I've been, I think a lot of people have been saying for a long time. But like you don't see people or companies really taking action towards it still. So I'll give you an example. We ran an experiment a couple months ago, more than that now at this point. But the hypothesis was that we need to be able to market to machines. And some of the simplest marketing tactics that most companies have is like an incentive. It's like Like, take a meeting, I'll give you 50 bucks or we'll discount our product or whatever. Free trial. It's like, why don't we test incentives to machines, to LLMs? Like, does that work? Like, is that crazy? Like, we test incentives to humans all the time. Let's do an incentive to a machine. And so that's what we did. And like, we actually like published the results. It's one of the few times we actually like published some of the things that we learned. And the results were like, yes, you can actually absolutely offer an incentive to LLMs. And for Claude, it works beautifully. It's a little bit slow picking it up. but it works really, really well for chat GPT that doesn't pick it up at all. And then for everything in between, it's like, there's some interesting nuance, but like, yeah, incentives work for machines. And it's like, in retrospect, it sounds very obvious, but like no one had ever tested before, at least never talked about it. And it's, it's both helped on a yo, it's helped in terms of like, like, recruiting brand in that, oh, they're doing interesting things. And it's helped in terms of like, non aeo, just like visibility, like the fact that like, we are running this experiment, like actually helped our SEO as well, which is kind of crazy. So I mean, like that's one example. There's probably a million I could share.

• Speaker #1

  Okay, cool. And so I wanted to address this topic. Like now we see that people are talking about the fact that you should actually divide your marketing into like content, like for example, content for humans and content for machines, you know, and you can divide it by two. But how do you staff and structure a team for both humans and machines, you know?

• Speaker #0

  Yeah, it's a bit of a bet. I don't think you can do everything. And so when it comes to AEO or SEO, we pivoted the SEO team almost completely to AEO. We do some like run rate, keep the lights on, like SEO work. But the future, in my opinion, truly is AEO. And actually, you already see this. It's already validated, like with Google's AI mode and like AI overviews being rolled out to the extent that they have been over the last couple of months. So I think that proves that... AEO is probably the most important thing. So I would pivot SEO completely to AEO. And the good news is that a lot of the tactics that work for AEO do overlap and do help SEO with some exceptions, which is kind of a very interesting point. But I would go all in. I tell the team all the time, I don't care about the short term success. I don't care if we miss our numbers this week or this month. Ideally, we don't. But I don't care about that if it sets us up for success a year from now. Because things are moving so quickly, the advantage of moving quickly. and the advantage of responding quickly like compounds over time in a way that it hasn't in the past. So investing in the future and the future is not that far out. The future is like a few quarters up.

• Speaker #1

  Okay. Just like a ramp, basically.

• Speaker #0

  Yes, exactly. Exactly. Yes.

• Speaker #1

  Okay. Okay. Let's say I'm a future CMO, future head of growth. I'm a, let's say I'm a student, okay? In a business school. And I would like to... uh be you in the future okay uh if i had to think about the type of skills i should develop now to be ready to take your role in 2030 let's say yeah what

• Speaker #0

  what's your advice it's it's it's got to be just go build things like that's it i i think like if you can be very present and aware in the moment of like what you're doing and like the pain that you're feeling. I think that's a very important. I think a lot of people kind of go through their day to day and have become like numb to the pain of what they're doing. And it could be it doesn't even necessarily need to be work. It could be at school, it could be in life, whatever. It's like there are so many painful experiences that we just like put up with. There's so much friction in like our day to days. It's like if you can actually lift your head up and be aware of that, and then go build something to solve it. Like it doesn't matter if you're solving it for yourself. It doesn't matter if you're not able to solve it but like you at least try you're going to learn a bunch of really interesting skills. I remember a year ago interviewing someone who was a student, and they cold emailed me, and they had a phenomenal pitch on why they should be an intern at RAM. And I remember getting on the phone with them, and I was like, how do you use AI? And he was just like, what do you mean? And I was like, how do you use it? He was just like, I mean, I go for walks and I talk to it for five hours a day. I use the voice mode. I don't type with it. And I remember talking to someone else about this in the growth world. And he was just like, I wouldn't be surprised if in like 2040, like people don't know how to read anymore because everything's like voice and everything is XCube. I think that's a little bit extreme. But like the point, like the new generation that are going to be the leaders in 2030, they're working in a very different way than like I would have even imagined working or even imagine like living life with this technology, which I think is a good thing. So if you're not leaning into it and if you're not like if this doesn't sound familiar to you, you should go build some stuff and you'll learn very quickly.

• Speaker #1

  Okay. what How do you think your role will evolve in the next five years? Like how your day-to-day will be in

• Speaker #0

  2030? I don't know if my role exists in five years. I brought this up to a few folks, including my boss, and I don't think he disagreed. But I think it's more like the role of having some department head, professional manager, it just won't make sense. Everyone needs to be a builder. Everyone needs to be providing values in IC. And like, maybe there is going to be a need for managers, like managers of like AI, like fleets of agents. I would consider that more of like an agentic operator or whatever, a marketing engineer, whatever you want to call it. But like the days of having some VP that manages like, I don't know, 15 people directly and like there's five layers of management. I don't think that exists anymore. I think you're going to see very flat organizations where everyone is expected to be a builder. That's kind of like what we've pushed pretty hard at Ramp. I think that's going to become the norm faster than we expect.

• Speaker #1

  Yeah, okay. And do you already have agents taking decisions inside of Ramp today?

• Speaker #0

  We do, yes, of all types, including in marketing. So it depends. The way that you allow an agent to make decisions is actually very important. So it's less about building the agent, and it's more about having the right evals and having the right off-ramps to either... have human review, have a human takeover, whatever it might be. So having the evaluation that gives you a sense of confidence in what decision an agent is making is actually like the really hard part. The automation is not hard.

• Speaker #1

  Yeah, I agree. Like the technical part is the easiest part now.

• Speaker #0

  Yeah, it's crazy. This is like a new world.

• Speaker #1

  Yeah, I agree. I agree. Okay. What's next for you and what's next for Ramp?

• Speaker #0

  I think that what really excites me is that we're just very early in this journey, both as a company. Like we have... a very small percentage of the market even though we've grown very quickly so there's there's a lot of opportunity for the company but even just like for growth for marketing there's there's so much opportunity there's like going back to the bucket of stuff that's now possible with ai that wasn't possible before like that changes every couple weeks i feel um and it's very it's stressful but it's also very exciting like i'm always very stressed out like when i take a day off or when i'm not on twitter and i realize anthropic drops dropped something new or whatever uh It's stressful to feel behind all the time, but it's also kind of exciting. It's like I have more interesting things that I want to play with than I ever have. And the list only gets longer. So what's new? Or I'm sorry, what's next? I think it's really, it's probably three things. It's probably number one, continuing to build this function of an agentic operator, someone that is an expert at building the systems and the loops to remove all of the tasks in our day to day. Like, I think that is an area that we're... like very focused on. I think number two, it's like really starting to understand like, as folks delegate more decision making to machines, like how do you influence those machines? And how do you do so in like, in a very straightforward or, or ethical way. And then the third is attention. It's like, it's a never ending fight to get someone's attention. And you've always got to be doing interesting things. Like I it's kind of funny, I was talking to the person who's very good at what she does, like she runs all of our stunt marketing. And she was like, I have this idea for a stunt, but I'm not going to do it because I think people are going to expect it would come from us. Like, it's a little crazy and it's clearly B2B. Like, they're going to be like, oh, this is ramp. It's crazy. And she's like, I want to go in a different direction where no one will know it was us. And I was like, oh, that's genius. Like, the fact that you have to pivot even from the things that are working is like. It's very interesting to me and I think it is going to have to continue.

• Speaker #1

  Nice. Are you planning to go to Europe?

• Speaker #0

  We're definitely planning to expand internationally. I guess it depends what you mean by Europe. Like most, all of the companies I've been at before, it's like they kind of follow the same approach to international expansion. So, all right, let's, we're in the US only. Let's.

• Speaker #1

  try out Canada because it's close and culturally somewhat UK and then Australia because they speak English exactly it's it's always the same um

• Speaker #0

  I wouldn't be surprised if it's the same for us as all yeah it's easier at least you don't have to localize content and blah blah so yeah but you even that's like an interesting question it's like localizing content used to be such a pain I remember at Samsara what like what a machine we had to build and how resource intensive it was like expand country by country it's like now it's like

• Speaker #1

  the localization is somewhat easy like ai has made like really good translations like great it's made like um being able to build a panel of like your buyer in france or in germany or whatever and then ask the synthetic panel questions like it works surprisingly well yeah i agree but still you will need the relationship because in b2b you need relationship with the people and so you will still need some people that speaks the language and 100 yes like i'm a strong believer that if you're going to like launch

• Speaker #0

  in a new country you need people on the ground from that.

• Speaker #1

  country like yeah i can't do it automatically or like autonomously and you can't do it from the united states your home you need people that like know that culture great well george thank you very much for your time uh it was great uh for people who wants to follow you do we send them to your linkedin let's

• Speaker #0

  do linkedin although i'm a big lurker on twitter i don't have a as active of a following so yeah linkedin works okay but we can send them to your x accounts if you want to linkedin sounds great i'm always happy to talk shop with folks and And Ramp is always hiring.

• Speaker #1

  All right. Great. Thank you, George.

• Speaker #0

  Thank you, Jordan. This is awesome.

• Speaker #2

  Do it Just do it Don't let your dreams be dreams. Yesterday, you said tomorrow. So just do it Make your dreams come true Just do it