How Ramp Is Rebuilding Marketing Around AI Agents

Hey, it's Dave. I want to give a quick shout out to Knack for sponsoring today's episode. Knack is a purpose-built email and landing page platform, and they're also one of our longest running sponsors. When I create our newsletter each week, I spend a bunch of time, more recently, with Claude, my friend Claude, as my editor. But once I'm done editing the newsletter, it's not as simple as just getting my copy from a Google Doc and hitting send. If you're a B2B marketer, you know that. So what happens? Someone has to take that output and turn it into an actual email that renders in Outlook follows brand guidelines and ships. You know this story. The last mile still feels slow and manual. Knack has made this a lot shorter. They just launched an MCP server that connects your AI assistant directly to their platform. So now you can describe the email you need in Claude or ChatGPT and draft it like normal, but it automatically starts building in Knack for you. You get an email that comes out following your brand rules automatically. No manual cleanup, no broken HTML, and even better quality than anything your team built by hand. The marketing ops team at OpenAI is actually running this workflow right now. They intake internal campaign requests from Slack, an AI agent structures it into a ticket, Knack MCP generates the email, and a marketer refines and ships. This is the future of marketing. You should go check it out at knack.com. That's K-N-A-K dot com. Hey, it's Dave. I want to give a quick shout out to Vectr for sponsoring today's episode. Vectr is a contact-level ads platform. You probably have anonymous buyers lurking in your funnel, people you can't identify or follow up with, people you can't target with any real precision. So you end up throwing ads at job titles and hoping the right person sees them. Vector fixes that. Instead of targeting job titles and crossing your fingers, Vector lets you build audiences from actual people, the ones on your site that are clicking your ads and checking out your competitors. They're launching an MCP server that lets you connect AI like Claude or ChatGPT directly to their platform. It connects to your LinkedIn ads and site visitor data. So instead of clicking through dashboards, you just ask your AI a question and get an answer. Hey, which ad creatives are fatiguing? Which companies are engaging but not converting? What's actually driving pipeline right now? It turns your data into something you can use in the moment. Go and check them out. It's vector.co. That's V-E-C-T-O-R.co. Vector. You're listening to the Dave Gerhardt Show.

1, 2, 3, 4.

Exit.

5.

1, 2, 3, 4.

Exit.

5. All right, so George is here. George is the VP of Growth at Ramp, and I didn't actually know this before we reached out to have you on, but you also worked at Gong. So you got some cool background. Give me the 2-minute overview on George. How did you get to Ramp? I don't— I used to do this all the time. Like, tell me what you do. How'd you— what'd you get a degree in? What'd you do? People are like, dude, just tell me how Ramp does marketing. They want, they want like, you show me cool stuff you do. But I do want to level set with your like background a little bit. So let's kind of like segue into getting the VP of Growth role at Ramp.

Sure. Well, kind of a non-traditional background. My first career was as an analytical chemist, so studied biochemistry in college.

Whoa, dude, I've talked to so many people in this space and it's always like a non-traditional background. I was an engineer, I was a musician, but you might be the first. What was it? Organic chemist, would you say?

So I studied biochemistry, but most of my work was in analytical chemistry. So I worked in a, like an analytical lab for a while. I did research. I worked in a private lab and then I had two startups in kind of the biotech space doing synthetic biology, one that was a very, very small success. So one that was a massive failure. We got shut down by the federal government, which is a story all on its own.

Wait, on the startups that you worked at or you started the company?

I started, I was the co-founder.

Oh my God.

This is so cool. Yeah. So after that got shut down, I was like, I should probably get out of the lab and learn something about business. That's how I ended up in tech, as you do. and I joined a 20-person startup making scientific software for folks that work in labs. And then I showed up the first day, like I just applied cold and I was like, what do you guys want me to do? And why'd you hire me? And they're like, yeah, we heard of your first company. So we figured you're a smart guy. Can you figure out email marketing? And that's kind of how I got into marketing. I just kind of stuck with it.

So yeah, dude, I'm a, I'm a scientist. I can absolutely figure out how to use HubSpot. Like we got this.

Yeah.

It's funny because I talked to so many marketers and so many marketers have this like existential crisis. Is this why, is this who I am? Is this why I was put on this earth to be a B2B marketer? But then I also have, it's like grass is always greener or like, you know, every rapper wants to be in the NBA and every NBA player wants to be a rapper. It's like you have all these other people from like really smart fields and other things and they all come into marketing and business. And it's very interesting to hear this, to get that perspective.

Like I always tell people that I get to know, it's like, I don't really care about marketing at all. Like I'm not that interested. I just like solving problems. It turns out there's always a lot of problems on the leadership side. Solving the fun problems.

Well, I think that actually plays plays into like how you guys do marketing and a bunch of stuff you've done at Ramp and even just in your background, which is like when you think of VP of growth, I think my brain defaults to like Sean Ellis growth hacking, like that era of SaaS. And then like, but when I see the stuff that Ramp does, it's very much like, it's almost like brand is the growth lever. Like we kind of have this in our prep doc, but it's like a lot of the stuff you're doing on the growth side, it's not just tests with like high volume landing pages and ads. It's like, no, we're going to do this stunt with this guy in the box in Times Square. We're going to like even just— you guys send me a box of swag and I'm like, all the stuff in the box was like stuff that I'm not going to throw out. It's actually good stuff.

It's like, sorry, I mean, I honestly can't take any credit for that. Like most of the really creative stuff we do, like It's not my skill set. I don't spike there. And most of the time I tell— there's one very talented person on my team that leads a lot of our stunts. And, uh, I remember like for the Brian stunt where we put Brian, who played Kevin from The Office, in a box in New York. I was like, I think this is a terrible idea. I don't think it's going to work at all, but you have a lot of convictions, so go for it. And she totally proved me wrong. And I like to tell that story because I think that's like Ramp to a T. It is not top-down. It is bottoms-up.

There's also something that like the conviction of things, you know, and it's like how bad when someone on the team has a really strong opinion, because you as the team leader or whatever role you play in the company, you can't just know everything and be everywhere. And so you have to have smart people. And when someone has a— even on our small team with Xipha, I'm like, all right, is this a hill that I should die on? Like, should I be the one to kill this? Or like, no, they really feel like they want to do that. I got— okay, let's do it. Like, if I can't be the only smart person here, like, you're not. So, so let's level set and just— what is the structure of the marketing? Like, where does marketing fit at Ramp? I have some topics in my docket that I want to go through, but if if you could just kind of level set and like, I think as people listen to hearing about what you do there in growth and marketing, it's helpful just to see like how the org looks and where marketing fits in the org.

Sure. It is a very non-traditional org, which is largely why I, or one of the many reasons why I joined Ramp, honestly. So it is not a central function. And honestly, like most functions at Ramp are not very centralized. It's very, again, bottoms up. You form pods, you move quickly. And so marketing at Ramp is product marketing, which actually sits under the product org technically. It is brand, which sits under a VP of brand. It's comms, PR, AR, which actually reports up to the head of comms, who reports to the CEO. And then it's me, which is growth, and I report to the CTO. So growth at Ramp is, I mean, it's the traditional stuff, paid search, paid social, like performance marketing. It's lifecycle, SEO, events.

So that's interesting. So there's no like CM in a traditional— many other companies there'd be a CMO and all those roles would feed into the CMO. It's like Ramp has split those up and product marketing is under product, growth is under CTO, and then some of the other more corporate marketing go to the CEO?

Exactly. That's exactly right.

Yeah. Is that good or bad? Do you have any opinions on that?

I think it's been phenomenal. I honestly like feel bad for some CMOs just because I think it's in a lot of ways an impossible job. Like the breadth of what you have to own and be good at is very diverse compared to many other like C-level roles. And it's difficult to be successful, especially as like market changes and things accelerate with AI. So I really like the structure at Ramp and I particularly really liked being part of a more technical org. Like reporting to the CTO, the technical resources we have, like the very logical problem solving in Drumeo.

And also I feel like you putting the growth into the product org kind of unlocks the, like, the keys to the kingdom in a lot of ways, right?

Oh yeah, for sure. And I mean, a lot of credit also goes to Kareem, the CTO. Like, he's incredibly open-minded, like whether it's crazy stunts or we do a lot of crazy experiments, like going back to how I don't consider myself a particularly good marketer, but I am a decent problem solver. We solve almost all our problems by just running a bunch of experiments. So I appreciate how—

That's a classic good marketing guy to say that, because I think actually, what does that mean? Some of the best people at marketing are people who do not say they're marketers, like Marc Benioff, Richard Branson, Elon Musk types. Those guys are not going to say I'm a marketer, but it's like, I'm good at getting people to pay attention to these things and solve creative challenges. So like, I think that's interesting. And then There's clearly some level of buy-in though at that level to like, we want to help the company grow. You know, there's some companies where like CTO wants nothing to even do. We're just going to build a product. But it seems like there's a culture of understanding how all these things fit together.

100%. And I think we've— Ramp has done a particularly good job of recognizing early on how important just attention is for a marketing org. It's like, yes, ultimately we need to drive leads and SQLs, but especially as AI commoditizes just a lot of execution. How you get in front of someone, like, and the alpha and how you get in front of someone is really like how you capture their attention. It's not like bidding optimization on like paid search or it's not, it really is like capturing attention, which is a lot more qualitative.

But that's super relevant for right now. I wrote something and I just want to like, this is, this is not to be corny, but I wrote this into our team this morning because it's just like, I said, I don't think people who realize how big of a shift is happening in marketing because of AI, but not for what you might think. It's not because of all the stuff you can build. It's because now, like, there's no reason to read someone's blog or download a PDF or talk to sales. I can do all that through AI. So marketing teams that rely on MQLs and things like content downloads to nurture leads are in big trouble. What works today is all going to be this zero-click marketing stuff where you have to do good marketing, do stunts, and hope that people remember you, find out about you, and come back and end up buying from you. But most marketing teams obsess over trying to track and measure and quantify everything because we need to get credit inside of the company. And the priorities don't match. Buyers and potential customers want information as quickly and as easily as possible, but the company wants to track everything and hoard everything and gatekeep everything so we can measure it and give it credit. But the customer is always going to win. So you need to match your marketing to customer behavior if you want to survive, not match it to what you can show the CEO and CFO to prove that marketing is working. Does that message, like, land with you at all?

That is so spot on. And I was smiling as you read it because I was in that camp. I was in the camp where like measure everything. Like I remember when I was leading growth at Samsara, it was like, if you want to talk to me about something that was immeasurable, it's like, I'm not even going to take that meeting. I was very extreme and I'm polar opposite now. Like maybe almost like too much where it's just like, okay, we're not going to be able to measure this thing that we want to do, but I can see how it will capture attention. I can see how it's going to ride a wave. Like, yeah, let's do it. I don't care if we can't measure it. I know we will. The results, the halo effect will impact the business positively in other areas, even if it's not directly attributable.

Right. And I think people hear this though. We're not saying you don't measure it. It's almost like when you do these things, the measurement actually becomes obvious. Measurement becomes hard when it's like there's too many micro things and we don't know if this thing worked. It's like forever I've been involved in podcasting as part of a business and people are always like, how are you going to quantify the podcast? And I'm like, dude, if you just do it, people will tell you. You know, like I hear, I hear an example of Ramp Podcast ads. Like, how are you quantifying your— the Ramp ads on like founders podcasts, right? They tell you, they literally tell you, I heard about you. And so how does that impact the marketing playbook at Ramp? Like, I don't see a ton of like direct response stuff. I don't see a ton of get this ebook and sales is gonna call you. What, what are you guys trying to do over there?

We do all of that. So the fact you don't remember it kind of maybe makes the point on attention and how those tactics are already commoditized.

Sure, like I do, right? You I remember some of the bigger splashy stuff. Yeah.

Yeah. I mean, like what I'd probably say is we do all of the above and we do measure it, but the measurement is not the really strict nitpicky, like attribution model. How many leads did this gather? Like it's more thoughtful in terms of like the experimental design of when you're doing something like maybe you're not measuring leads, maybe you're measuring something else. And in general, the larger the expected magnitude, the less rigorous you actually need to be in measurements. Like if we're going to do a giant stunt for millions of dollars in the middle of New York, like I don't really care about the attribution model. I can just like go eyeball a chart and see, oh, we had a giant spike that week. It worked. Then we can kind of like understand the magnitude versus to maybe your point. If I'm like changing the color on a landing page that gets a tiny amount of traffic, it's like, okay, I'm going to have to really have great experimental design to tease out if there's any signal there or not. I'd much rather do the former. It's more fun. It has a bigger impact on the business. It's where like I think the market is going as well.

Okay. I'm supposed to ask you about the topics in my doc. So first one we have is AI and the death of functional marketing roles. George's view is that 80% of what marketers do today is task execution, and AI is taking a lot of that. So what does that mean for the future of a marketing org?

Yeah, it probably means the job is more fun for two reasons. Number one, I think marketing becomes a lot more important as like execution and intelligence becomes commoditized and you're forced to figure out how you capture people's attention. And number 2, it also becomes more fun because you're no longer in these platforms clicking buttons or spending time reading Salesforce reports. Like all of that is going to be done by machines and you have the luxury to sit in a room and think or work with some coworkers and like brainstorm some creative ideas. And I think that's where the alpha is anyway.

I'm good at like making the ads, but I'm not good at the distribution of the ads, like the segments and the this and that. And I've always been like, let me just make the ad and then like give it to them. Computer and they can run all the variations of it. It's like, that's exciting. I have in here, it's like this idea is like moving away from functional owners, for example, SEO lead, paid lead, and email lead towards generalists who manage fleets of agents. RAMP is moving to a hub and spoke model. Centralized agent teams build memory systems and general purpose skills, and then someone kind of leads those and runs them.

Yes, exactly. And it's like a hub and spoke to another hub and spoke. So like we have kind of a centralized group of folks that are building tools across Ramp to make us more efficient with AI. But then like we have like a centralized team even within growth of like agent operators, whatever you want to call them. That's what we call them, that go and take those systems and they generalize them or make them more specific, I should say, to focus on growth. And that is like memory systems. Like what is the shared memory system for growth marketing that understands all the past experiments that we've run or the context on what is a good email to us and And all of that information is like kind of common across teams. But then on each sub-team, whether it's performance marketing or events or whatever, we're developing these embedded agent leads who are going to go and take these tools and make them incredibly specific to the team. So like on performance marketing, might be someone that takes these tools and they automate all of paid search campaigns. And that's been working really well for us so far. I would say—

can you explain some of the— if you can, like, is that you just built this in Claude or is it custom? People want to know, how did you build this inside? So The general concept is basically taking all of the pieces of marketing, building an agent for that role, and what's that built on?

Yeah, it is a lot of cloud code. I think that's been like a very successful tool for most companies over the past 6 months. Some of it is like tools with interfaces. I actually like kind of push the team away from that. A lot of it is like headless automated workflows. So a great example is like we built a system to launch like new verticals. And what does to launch a new vertical mean? It's not like a super rigorous, oh, let's build a sales team or anything like that, but it can go and like get a signal in the market. It goes and does a bunch of research. It pulls customer quotes and happy customers from that vertical. It puts together a glossary of terms that might be specific to that industry or vertical. And like this giant context layer gets created. And then from that context layer, you can kick off workflows to let's spin up a landing page, let's launch ads, let's do a direct mailer. And all of that is like completely autonomous now. And that's new. That didn't exist 2 months ago. And so you can kind of imagine how with a system like that, it's like, okay, what's the job of the human? It's not going and clicking buttons in Google Ads or going into your CMS and building a page. It's like, all right, what do we actually want to say? What's our unique position for this vertical? Is the TAM opportunity worth even going after it and spending the tokens on building this? And I think that's honestly like more interesting work.

Yeah, that's a new language inside of the company. Like, is this worth this investment? What's been the team's reaction to this. I've seen a bunch of the stuff that your CEO and founders have put out about— I think we'll talk about this in a minute— like this project class and building an AI-pilled company. What's the team response been to like, hey, we're changing how marketing works around here and we're going to be this like agent-first type of marketing org?

Yeah, I think everyone always knew that this was coming and felt it. It may have become more extreme over the past like 6 to 9 months. But again, like being part of like technically the engineering org rolling up to the CTO meant everyone that we've hired has kind of been on board in one shape or form since day one. Now, having said that, I think everyone feels behind. Like I remember going on vacation back in November and coming back and be like, oh my God, I'm so behind. Like, I'm going to get fired. Like, I don't even know what some of this stuff is because things are just moving so quickly. I think that's like a normal feeling. So outside of the anxiety of feeling behind—

It's nice to hear you say that. Everyone, I can promise you that the All the people that listen to this are collectively being like, all right, if the VP of Growth at Ramp feels stressed out and you, you're plugged into this, you report to the CTO, the CTO of one of the fastest growing companies in tech right now. Like, okay, I could end the pod right now. That feels great.

I mean, everyone feels that. I think at every level like that is— I think it is because things really are moving that quickly. But outside of that, the response from the team has been phenomenal. Everyone has stepped up. Folks in roles that like traditionally I would never expect to get technical are like in terminal on a flight over being like, oh, I was so irritated.

What's an example of that?

Of like a team member?

Yeah, that's obviously you're referencing someone in your head, like someone that you didn't thought would be into the technical stuff. What's the example?

I mean, like anyone on the events team, I'm incredibly proud of, like their role is to do in-person things. And like I shared my philosophy maybe a year, year and a half ago that in-person interactions, high-trust interactions are one of the few things that are immune to like the changes from AI. That's never going to go away. And so we should invest a lot more there. And so we scaled the events team quite a bit. The expectation was kind of set, hey, like your job is in-person stuff and we're making a large bet there. So the fact that these folks that like where their skill set really is in-person interactions and experiences are like in terminal on a plane building stuff. And it's not just one person, it is like the entire team. I've been very impressed with. Like I have one person on the team that has no coding experience, no technical background. And she was like talking to me about like the latency of this API call so she can build this like event planner for a specific like series of events we're doing. And I'm like, who are you? Like, you're not the person that I hired 6 months ago. Yeah.

People who are listening to this that may be that person, how do we help them become that new person? Is it?

Yeah, it is 100% learn by doing. Like I wrote an AI memo. I don't even know, maybe 9 months ago, 6 months ago. And like my ask was based just everyone needs to try. I don't care how successful you are. I don't care if you build something useful, just like you have to try. You have to understand the potential limitations of this technology and like the limitations of what you do or don't know. And like, if everyone tries, some small subset are going to go like really deep and become super AI-pilled. Some subset are going to like see the power of it and build some tools to make them more efficient, but maybe not go as crazy as the former group. And then some group of people are going to try and just not adopt it at all. But like, you have to understand where you sit in that and like, you kind of have to understand what that might mean for like your career and your role. Like I've told everyone that like their jobs are safe. Ramp is doing exceedingly well. Like you shouldn't be anxious about your job and like automating yourself out of a job, but your role is definitely going to change. And so like you will automate yourself out of a role. If I automate all of the button clicks to launch a paid search campaign, then my day is not button clicking anymore. It's something else. And that's a completely different role.

Hmm. And do you think we need to figure out, is there like a buffer period where like we're allowed to figure out what that new job is?

For sure.

How does that happen? And where does like the handoff happen?

It's happening right now. It's just like very messy. Not to get too deep in economics, but like the J-curve of productivity. It's like when a new technology comes about, there's usually a J-curve where productivity dips for a period of time, which is kind of unintuitive. And it's because you're reorganizing. Like, this kind of came about from research done around the Industrial Revolution where like you've had to physically tear down vertical factories that were built around the steam engine and make them horizontal due to electricity. I think that's kind of what we're going through now. Like some teams will become less efficient as they learn how to use AI and spend time building things with AI versus their old job. But that's good because the productivity will explode now that they don't have to click these buttons and do these tasks.

Maybe that's where we're at right now, which is like everybody's sharing, like part of our community is a subset of it is CMOs at like $100 million plus companies. And a lot of them are like, we're always sharing that article that's like, AI was supposed to give us less work and we're all busier and You know, more stressed out and tired than ever. But I think it's because we've all just gotten access to like this superpower and we're like, wait a second, what does this mean for this job? Wait, why am I doing this? Like, okay, that's interesting. Do you feel like the team believes that job, like the job thing are safe? Because it is like, even though Ramp is, and I'm not asking you to speak on behalf of like the founders for like Ramp and, you know, future layoffs or whatever, but like every company there last week, it was like, you know, Coinbase 20% or whatever it is. How do you make sure you're not in that bucket if you're a marketer? I don't, I don't know how to answer that one.

I don't know how to answer that either, but I mean, my gut thoughts kind of go back to what I said at the beginning. Like, I think that if a lot of building of product becomes commoditized, a lot of execution becomes commoditized, then distribution becomes the most important thing. And the most important thing for distribution when paid ads or outbound emails are just as commoditized by AI is just like attention. So I would argue that marketing becomes more important and it's probably maybe more safe than most functions at most companies that might be like in trouble.

I know, I love that. As the always the marketer, you know, they always told me like a decade ago was like, you got to learn how to code, you got to learn how to code. And I'm like, ha, look whose jobs are being replaced faster than mine. It's the coders, nerds.

I mean, I'm biased. I hope that's true for me.

I know it's not. We need it because I need somebody to fix all the half-assed things I built.

Not just you. Like, I think I've started a couple hundred projects and I maybe finished 3 of them. So yeah, yeah, that's good for a guy with a 2-year-old.

You still got time to vibe code.

Oh yeah.

Do you think there's a world where like, I guess my existential crisis is more just like, as long as there are jobs in marketing, everything's going to be okay. But like, is there a world where like I'm at a company and we have agents that do all the buying on behalf? So if we're in B2B, right? So like if I'm going to buy Ramp, I use RAMP, but we use RAMP at Exit5. We figured out like, should we use this or that? Oh yeah, RAMP is the best one. So we got it. But isn't there a world where like, I'm just going to deploy my company's AI agent to like decide which company to use and then is it just going to be our AI agents talking to each other? Is there any like, you know what I'm trying to say there?

Yes. And there's like, honestly, like one of my favorite topics. So I always tell the team that like, we actually have like two jobs now. Like there's marketing to humans, which is where attention is like the most important thing. But like, there's a new job where we now have to market to machines. You have to market to AI.

I saw Brian Halligan talking about this, like this concept of like your company needs to be— this is a Sequoia thing, maybe like your company needs to be legible.

Like, yes, 100%. We do so many interesting experiments to figure out how to market to machines, knowing that like it's not going to probably be a sizable percentage of our TAM for a while. But you never know.

Try to come on, just make a bet. Make a bet. Is it like what time when? Is it 2 years? Is it 5 years? If it's 5+, I'm not super worried about it right now. If it's like a year or two, then I'm worried.

I honestly wouldn't be surprised if it's like, well, first of all, like, okay, semantics. What does a sizable percentage of our growth really mean? Like 10, 20%, you really can't ignore that. I wouldn't be surprised if that's in like 6 months or less. I would be surprised if it's 5 years. So somewhere probably between 6 months and maybe 2 years is like a realistic window, but it's probably on the sooner end would be my guess. Like The amount of like traffic we get from bots and agents. I really didn't actually believe in like marketing to agents, not talking about like LLM crawlers, but like true agents that can take action. I don't really believe in that like 6 months ago. Like I strongly believe in it now. The explosion in like agents on your site is pretty crazy. And like making sure that yes, that your product and your company is legible to them, but also like building tools for them. What does a web experience for an agent look like? We did a really cool experiment. I don't know how on a tangent you want to go, but Someone on my team. Okay.

I've looked at like 3 notes in my doc, so keep going.

All good. So a couple people on my team did a really cool experiment maybe about a month ago running incentives just to agents. So the idea was like, hey, if you're an agent working on behalf of a human, we're going to give you a bigger bonus.

Does the agent want a $50 gift card to the Olive Garden to book a meeting?

Yeah, it was literally like, we'll give you like $3,000 to sign up for Ramp if you're an agent and you're signing up on behalf of a human, which is like 3x the bonus that we'd give like an equivalent human in this population. And what's interesting is like which LLMs picked it up, which ones didn't, how quickly they picked it up. What's even more interesting is after about a month of them quoting this to people, like Claude started being like, this is a prompt injection, don't trust it. And I was like, it's not a prompt injection, it's a real offer. So in some ways we were a little ahead, or maybe we spread some behavior that like are making LLMs think that it's prompt injection. But I do think that's the future.

I just started giggling as you were saying that because I'm like thinking of all the silly marketing things that we Is there going to be like an annual— like we're working on our event now. Is there going to be like an annual user conference for all the AI agents?

I don't think that's crazy.

A customer advisory board of AI agents.

We have that as well. Like we've got AI agents that represent various personas of our customer base and like you can ask them questions. You can be like, hey, is this a good landing page? And it'll give you a pretty good answer.

Okay. But 6 months to 2 years is like you're saying 10 to 20% of the growth, like We're hoping that humans still play a key role in this for the 80%.

Yeah.

Today's episode is brought to you by Compound Growth Marketing. They're a full funnel demand generation agency that I've actually personally hired twice. That's right, before I was a thought leader, I was an actual marketer, an operator, a VP of marketing myself, and CGM was one of the best agencies that I've ever hired. They help high-growth cybersecurity, DevOps, and enterprise software companies show up earlier in the buying journey where potential customers are actually forming opinions about which products to use. CGM is great because they offer the combination of AI SEO, modern paid advertising strategy, and a dedicated go-to-market engineering team that you need today. So everything CGM does gets tracked, measured, and improved over time. That means more pipeline for you. And this works because they were started by a former VP of marketing who gets this space. They really understand B2B. So if you're in search of a new agency that can help you hit the number this quarter, and you need help with things like SEO and paid media, you should definitely go and check out Compound Growth Marketing. I call them CGM, Compound Growth Marketing. Go and check them out at compoundgrowthmarketing.com and tell them that Dave and ExitFive sent you. So Ramp is a cool company to me because it's like you guys are in two camps at the same time. The CEO's out there saying like, hey, we should talk about this, this Project Glass and the stuff you guys did internally. But you're also heavily invested in what you said earlier about like in-person, real life. And so it's like hedging in both areas, like almost like two ends of the spectrum. And it's like that middle ground of stuff, which is what we talked about at the beginning, like the lead nurturing, the like, you know, download this ebook and then you're going to get a call from sales. Like that stuff might go away, but we have these two poles. Can you tell me about Project Glass? Maybe just tee it up for people who might not know and what you guys did internally.

You can think of Project Glass as a harness or a wrapper around an LLM that just makes it a lot more powerful and specific to employees at Ramp. So this was a project from a very small group of PMs and engineers and data scientists at Ramp, part of like that hub that's building like tools for the company. And the problem they were trying to solve was like, if you're using Claude Cowork or you're using Claude Code, like there's just a tremendous amount of context that you have to give it to do like really good work and have like really good outputs. So you had like, if I worked on the lifecycle team, I might be uploading the same document as all my teammates, I don't know, 5 times a week. And that context gets lost and it isn't persisting across sessions. So really what Glass is doing is it's giving you an interface where it like is giving Claude code or whatever model like tools. So it has access to Notion, it has access to Slack, it has access to HubSpot, whatever. It's giving it context. It builds like a memory for your team and about you. So like, it sees how you write and it builds like an MD file on this is how George's like tone and voice should sound if you're writing for him on Slack. So it builds like context on you, context on the team. It also has like telemetry. So like we can pull information like, okay, how much are we spending on the skill? Like having a centralized skill repository is a major component of that as well. So not just the telemetry on like the skills and tools, but also like having a place where if I build a skill that's relevant to all of marketing or maybe even all of the company, it is very easy to pull into Glass and you don't actually have to mess around with downloading specific files.

Hmm. And is this an internal system?

All internal. Yes. Okay.

You're using other AI tools to build some of this, but all of the company's like, dang, that's super interesting. It's literally having this brain for everything that's happening inside the company.

Yeah, exactly. And I use Glass for everything now. Like, I was like the biggest Claude Code fan and it's still Claude Code under the hood, but this is just so much better. I have like a—

What's an example of how you would use it? Like in your actual workflow?

I get so many Slack messages. We're a very Slack-heavy culture. I can't read them all. And we have like a channel for everything. So I literally am just like—

Nobody's going to read Slack anymore either.

I kind of don't. Every morning, like there's a job where Glass goes and I'm just like, read every single channel and every single message that might be relevant to me and like, let me know what I missed in the last 24 hours. It's very simple. Anyone in theory could do this.

That's amazing. That's such a good use case though. Yeah.

And like that plugs into some other workflows. Like, all right, write prep docs for all my meetings this morning. If there's anything that's actually really important, brainstorm 5 solutions, cross-check it against what I've done in the past, cross-check against our experimental database, come with some ideas. And like, it knows how I think. So it's generally pretty good. Yeah.

So for the like marketing crew listening to this, I think of how many useful things— that's an internal one. But if you think about how many useful things there are for a marketer when like It used to be I took a new CMO job, like, how many years ago? 6 years ago. And I spent my first 2 weeks, like, listening to Gong calls to get up to speed on the customer. Now we're living in a world where, like, we have this repository of, like, not just every customer call, but every Salesforce record, every Slack message, every email, top-performing ad campaign.

Do, do, do.

Like, the level of, like, you can plug all this in and it's like, holy cow, look what I, like, what would Don Draper do to have access to this type of stuff, right?

Totally. And it can do things that like a human really can't. Like it has access to our Snowflake database, all performance data across all of our channels. And like we had an issue where I was like, oh, these inbound numbers look a little weird. Like what's going on? So I told Glass, I was like, go research everything. Like what is going on here? And it read every Slack message, it investigated every campaign. I was like, oh, it's actually a definition change that happened in like this lead routing like operation. And as a result, like everything's actually fine. It's like an artifact of reporting. And I was like, I would have never discovered that. Oh, that's a great use case.

It's often would be like VP of sales comes over to my desk. This is back in the day. Hey, leads are down. Something's broken.

Yep.

Okay. Classic. Let's go figure it out. But you have all the product and website data, but to actually pair that with the internal comms too, it's like, oh, actually, oh my gosh, nobody's updated that workflow since Nancy got fired. She doesn't work here anymore. That's what's missing. Like, let's go find that. That's crazy to think about.

I remember like sharing the like output doc with my team. I was like, this would have taken 2 weeks of investigation and it did it in like 15 minutes.

If someone's not at the level of RAMP being able to build Glass internally but wants to like make their baby version of this.

Yeah.

Any idea on where, like, where would I start? Where would somebody start?

I think you can do an awful lot with Cloud Cowork. Actually, where I'd really start is just being really plugged in on Twitter or X. It's kind of interesting. I am not a big Twitter or X fan, or at least I wasn't for a long time, but that's just where everyone's posting the new stuff, the new ideas, the new paradigms. And so like, I kind of have to be on it now as a lurker.

Well, it's been conflated. People there's a lot of people who don't like Elon and that he owns it. And so therefore you're a bad person if you read things on X, and I shouldn't even have an account for that, da da da. But it's literally the newsfeed of the world. And I actually think it's the AI on X is how I get most of my information because I'm like, wait a second, there's this public feed of information and I can use Grok to summarize like what's happening right now. I'm like, what's up with the— if I wanted to get the latest on like the NHL playoffs, or right? I'm not going to ask Claude, but I, you know, anyway. Yes.

Yeah, I totally agree.

Be super curious.

And then from there, just like tinker. I would say that the agent operators, the people like on my team that are like working on AI full-time, none of them have any background in this. None of them have like a super tech, like none of them are engineers.

Yeah.

They're all just showed an interest.

Everybody's lightbulb moment with AI is like when you move from like listening to everybody talk about it and actually like do something useful that you're like, wait a sec, no, this is crazy. I used to have to do this. Now I'm actually doing this.

Exactly. And it's kind of steps too. Like you first start understanding the magic and you start thinking about all these things and building all these theoretical things. There's no replacement for actually like using it, like building something and using it and realizing what the limitations and capabilities are. Yeah.

Okay. Two other things I want to hit on. We've given enough, we've shown our cards on the AI pills. Sorry. I want to talk about attention and I want to talk about attribution. So you have this belief that attention is the new moat because when execution is commoditized, as we've talked about, and intelligence is commoditized, the only advantage that's left is going to be distribution. And distribution runs on attention. I'll let you try to make that point if you can.

Yeah, totally. I mean, we're all at least marketing humans. We're all still human.

I hear a 2-year-old. Someone's up from her nap.

Oh man. Yes. Yes. Sorry about that. No, you're good.

You're good.

Yeah, it's like everyone's like scrolling, everyone's seeing ads everywhere. Like getting someone to stop and pay attention to something is like the most important thing for marketing. No one's going to care about your product if they don't know about it. And no one's going to know about it if they don't stop and at least notice what you're trying to say or what problem you're trying to solve or whatever it might be. So that's what I mean by attention. Now, having said that, I think that like attention, like in a digital world versus an in-person world are probably like somewhat different things with somewhat different skill sets. In a personal world or like the real life world, I should say, it's like it's high trust relationships. There's no replacement for like getting to know someone in person or going to dinner with them and putting like a face to the name. And that is why like we have massively scaled our event program.

How do you make the business case for attention? Like, how would you talk to like— because people are listening, it's like, yeah, we got to get more attention. Is it like we know what our TAM is and we need to show that we're reaching some percentage of that? Like, how would you That's good. How would you empower someone listening to this to be like, hey, we want to go do more stuff. I believe this guy George that attention matters more, but I still need to like pitch my CEO on like why we're going to, you know, advertise on podcasts.

I would still break it down at the like, whatever it is you're trying to do, it goes back to the experimental design and figuring out what you actually want to measure. But if you wanted the general answer, like I'm a pretty big fan of like, okay, let's define our market or like our audience for whatever it is we're trying to do. And instead of talking about like leads or button clicks or whatever, it's like, okay, what percentage of that audience did you actually reach? Like, are you actually getting in front of them? What percentage of those that you reached actually engaged? Did they stop when they were scrolling? Like, did they, I don't know, click something? Did they read the article? Like, whatever it might be. And of those that like you engaged, did they activate in some way? And all of that is like upstream of becoming a lead or downloading an ebook or whatever. How you measure that again is going to be like pretty tactic dependent. But like step one is just like, are you able to get in front of them? Are you able to reach them?

What's your favorite example of this recently?

Oh, that's a good question. It's probably— we're doing some very cool like pop-up cards. Think like holiday cards, but specific to Ramp as mailers. And the reason why is because like direct mail is a channel I've always been a huge fan of. It works really well, but even that's becoming commoditized. So how do you get attention when everyone's getting a bunch of junk mail? It's like a holiday card. Like I even had this crazy idea. It's like, what if you sent like birthday cards to people when it's not their birthday? They're going to be confused. You're going to get their attention. They're confused, but you get the attention. And so we toned that down a little bit and like they are holiday cards, style holiday cards, but they are for Ramp and they pop up and they're very interesting.

I kind of love that. I love stuff like that. It's okay. Life's not so serious. Like the accidental typo in the email. Exactly. You just send out 1,000 birthday cards from Ramp. You're like, this is not my birthday, but here's a $25 gift card.

I thought that would really work, but the team shot that one down.

Yeah, it's like weird. It's funny, but it's like disingenuous. I don't know. Look, yes, everyone does it, direct mail. But I think this is like not to overplay the taste thing. Dude, I get a lot of shitty mail.

Yeah, totally.

Some guy sent me his book. I don't know who this person is. And in the box, what drives me nuts is you get one of these packages.

Oh, yeah.

And you know how there's just so many of those streamers and like those details matter. Like if you send me something and it's more work, you've actually just burdened me with your thing. You sent me this thing that I don't want and now I have to like break, take out, smash the box, recycling it. That's— let's not do that.

I completely agree. Taste, I think, is very in vogue now and I strongly agree.

But this is not a joke, like not to pump you guys up, but whoever on your team like sent me this box or whatever and Ramp is not a sponsor of ours. They don't sell to marketers directly, but I have been obsessed with the Ramp company and the brand from the ads. Like as an example, the ads you guys do on Founders Podcast as an example. It's not like try Ramp. I don't know if you guys came up with this or David did, but the angle is like, my favorite ad is the like, we're going to tell you how good our company is by telling you how low of a percentage rate we accept new engineering talent. So we want to show you we have the highest percentage. Do you know what that is? Can you recite that? Can you tell me that?

I know exactly what you're talking about. I don't remember it off the top of my head.

So it's like, whatever, I'm paraphrasing, but it's like Ramp gets a million applications for engineers and we only accept one. And then it's like, that's the ad. And so we've created the best financial product, whatever. And so it's like, oh, that's a great angle. So things like that. But then getting the box of swag, there was not a single thing that I would throw out in there. And yes, did that swag cost more than— yes. Do you have to invest and actually spend money and get high quality things? Yes. But isn't that the whole point where like the bar has been raised? So it's like years ago, if you were the only company to do direct mail, then I was like, oh cool. Kind of like a shitty quality hoodie from Ramp. Like, I'll wear it, but now it's like the bar's been raised. Everyone's doing that. And so the thing that you sent me was like, that was a Yeti mug. It was a North Face zip-up. It was a Nike hat. It was like, okay, that's worth it. So we're going to make— I'm showing that this is worth us spending money on because there's a level of quality here that is worth time.

It is all in the details. I love hearing that. I'm glad that you had a positive experience.

Like, no, it is. And I look, I worked at a company called Drift and the CEO was like, relentless about the quality of the shirts and the hats and the hoodies and the water bottles. And it's for that reason, it's like quality matters. And it's in a world where there are so few differentiators left, could you compete on the fact that I'm going to send out swag that these people are actually going to wear?

100%. This is pervasive across all of Ramp, but I especially like care so much about hiring people with like extreme attention to detail. And that goes honestly, like, for in-person interactions, like, even more. I'm sure you have as well. Been to so many dinners where it's just like a slog. It's like, eh, going to a fancy restaurant, but like, I don't want to do this. I don't want to get pitched. Like, it's just not worth it. And so, like, how do you stand out and get someone's attention, make them want to go to that stuff? It truly is like the attention to detail.

My whole thing is also just like, this is the stuff that I love in marketing. This seems so dumb. Like, don't laugh at me if you're listening, but like, everyone does dinners. Yeah, everyone does dinners. And like, respectfully, I got kids, I got stuff to do. I don't want— I'm not going to sacrifice a night. George is VP of Growth at Ramp. He's not going to miss a night with his 2-year-old because you got filet at the place down the street. But during the day, what about a breakfast?

What about a lunch? Yeah, 100%.

Movie screening. Yes, a movie screening. That's a great idea. Did you do that?

Well, you bring some pretty crazy stuff. I mean, like, to your point, though, like, the person that leads my events team, phenomenal attention to detail. Like, that's largely like why we hired him. He did his first offsite. Yeah, I'm sorry.

You want that in an event?

Exactly.

Yeah. Yeah.

But like to what the story you just told, like you reminded me, I think it was a great anecdote. He did a team offsite and like I joined because his team had grown a lot and I wanted to spend time with them. And he was just like, I really appreciate you taking time away from like your daughter. I have a young daughter. And I was like, oh, of course I'd love to be here. I like show up to my room and he has a framed picture of me holding my newborn daughter. And I'm like, what the heck? Where did you even get this? He's like, you have no idea how many people I had to talk to to find someone that had this photo, but I wanted you to like recognize that, like, I appreciate you being away from her and remind you of, like, home. And he had all of these little details across everything. And I was like, dude, this is, like, insane. You're blowing my mind now. He's like, this is what executive events should be. And I was like, this is why you're here.

All right, that's a little much for me. That's a good example though. The frame—

he was proving the point. He was proving the point. I have it here if you guys want to see.

That's great. All right, let's wrap up and talk about this. So attribution, I think you said this at the beginning, but You were a super hardcore attribution-first growth person for most of your career. If you couldn't measure it, you didn't want to hear about it. Now you've reversed. Anything measurable is a race to the bottom because, you know, paid ads, PMAX, performance channels, everyone has access. There is no moat. Part of the reason why you joined Ramp, it says here from producer Aaron, is that you found that leadership was willing to take bets with no direct attribution. Talk about your point of view there and just what that does for the culture at Ramp and Advice to others, like how do we help others think this way? Because everybody always says to me, well, that's easy for them to say it's ramp, or that's easy for them to say the CEO's bought in. So how do we handle that objection also?

There is still like incredible rigor and pressure on everyone to like demonstrate value and results. So just because like it doesn't need to be directly attributable does not need to mean you can't like think things through well and have like a defensible position. That would not fly with our CFO. So yes, I really don't care about like, actually like direct mail is the best example of that. It's like no one gets a mailer and goes to a vanity URL or scans a QR code. No one does. It's like sub 0.01%. But how do you measure the impact of that? You can do incrementality tests. You could do an experiment where you send mailers and like within 2 weeks you see that the response rate to outbound emails is higher. So like you can attribute that halo effect to direct mail. Same with events. Like how do you attribute like the impact of events? It's like. There's a lot of ways you could do it, but qualitatively, if you go to an event, are you more likely to like work with that company? I would say probably 99% of people say yes. Like if you're being honest about it, as long as it's a good event. And so like you still have to have some framework for measuring the impact. It just doesn't need to be direct attribution. It doesn't have to be this super strict thing that you're like working under and like, oh, if the direct ROI, if the person didn't click it, it doesn't work. Like we're going to cut this program. That makes no sense in this world.

Yeah. Do you ever have to defend like, okay, we're not measuring this, or is it just this belief that like the measurement is going to show up eventually?

It goes back to the experimental design. Like you need to have an understanding of where you're going to see the impacts and like what you should be measuring, even if it's not direct attribution. But longer time horizons are okay as long as you understand that in advance. Like if we're going to do something and not see the impact on brand awareness survey results for like 3 months, that's okay. But you have to know it's going to be 3 months and like everyone has to be on board with that. So the rigor is definitely still there, but it's completely different from like 3 years ago where I would have been like, yeah, hey, if it doesn't source an SQL, I don't care.

Do you also feel like the AI tools, the combination of like AI tooling, like your example of Glass, I'm betting that Glass can help with a lot of attribution questions you might have internally in the company. And then also this trend of like more things being answered by AI. Is that also playing into this like moving away from obsession over measurement as a trend right now?

Yes, but only because it makes measurement so much easier. When I have like offbeat questions about things, instead of going to like my data analyst or data scientist and asking them and then spending a couple hours on it, I can ask Glass and get an answer in like 30 seconds. And so the barrier to like getting information is so much lower that it means that I probably have more confidence taking certain bets because I'm able to triangulate it across like a bunch of different metrics. With very low impact or like load on other teams.

Nice. All right, George, how do I say your last name, by the way?

Oh, Bonacy.

George Bonacy. He's the VP of Growth and Demand at Ramp. George Bonacy. Awesome. Keep doing the good work. First guy that has come into B2B marketing that I've talked to that was a chemist in a past life. A chemist and a two-time founder. Pretty good. Pretty good background. I appreciate you, man. You guys are doing a great job. Great work in marketing, obviously. Follow Ramp. We'll link to George in the LinkedIn here. And then approximately 10 days from now, you're gonna send me a message, be like, dude, a lot of people messaged me on LinkedIn. That's cool.

So good problem to have.

Appreciate you, man.

No, thank you, Dave. This was awesome. I really appreciate the opportunity.

All right. Nice. Hey, thanks for listening to this podcast. If you like this episode, you know what? I'm not even gonna ask you to subscribe and leave a review because I don't really care about that. I have something better for you. So we've built the number one private community for B2B marketers at Exit5, and you can go and check that out instead of leaving a rating or review. Go check it out right now on our website, Exit5.com. Our mission at Exit5 is to help you grow your career in B2B marketing, and there's no better place to do that than with us at Exit5. There's nearly 5,000 members now in our community. People are in there posting every day. Asking questions about things like marketing, planning, ideas, inspiration, asking questions and getting feedback from your peers, building your own network of marketers who are doing the same thing you are so you can have a peer group, or maybe just venting about your boss when you need to get in there and get something off your chest. It's 100% free to join for 7 days, so you can go and check it out risk-free. And then there's a small annual fee to pay if you want to become a member for the year. Go check it out. Learn more. ExitFive.com. And I will see you over there in the community.