E19: Architecting Systems for Hypergrowth with Nate Follen of Ramp

Today's guest is Nate Fallen, who is the head of business systems operations at Ramp and a former business systems leader at Intercom as well. And I'm really stoked about today's conversation because Ramp is obviously a generational company. They were the fastest company ever to $100 million ARR, and legendary investor Keith Rabois called them the, quote, best run private company on the planet. And so today we're going to dive into what they're doing different on the revenue side, how this business is run, what business systems looks like at Ramp, and how AI is being used specifically. So, Nate, really excited to dive in. Thanks for joining.

Hey, Kyle, thanks for having me. Excited to do it.

Hello and welcome to the Revenue Leadership Podcast by Topline. I'm your host, Kyle Norton, CRO at Owner.com. And every Wednesday we dive deep into the strategies and tactics being used to drive success by the best revenue leaders in the world. So join me as I sit down with revenue operators to discuss actionable frameworks that you can implement in your business today with no fluff, no sales pitches, and no platitudes.

This episode of the Revenue Leadership Podcast is brought to you by Superhuman. Superhuman is an AI-powered email app that helps sellers respond, collaborate, and close faster. With features like inbox triage, automatic reminders, and collaborative email drafts, reps can filter out distractions to focus on what they do best—selling. Get the email experience that keeps deals moving. Try Superhuman free for 1 month at superhuman.com/revenueleadership. Com/pavilion.

I gave a little tiny background, so give people your 60-second intro, a little bit about yourself.

Yeah, so, so I came background in the life sciences, actually did an undergrad and grad degree in genetics. Um, from then on, just decided, you know, big data was going to kind of eat up the roles and jobs, and genetics wasn't necessarily a fun workspace working in a lab. I like people too much, so joined a as a 5th employee at a small startup in the benefits space, kind of a competitor to Zenefits back then, and was kind of doing everything, a lot of go-to-market systems work, sales forecasting work, working with insurance carriers and doing some product updates, and got deep into the Salesforce ecosystem. From there, did a few other systems manager roles, joined Intercom where we were moving from kind of a product-led company to more of a sales-led company moving upmarket. And then from there, moved into the restaurant tech space, which we'd love to chat with you about over at BentoBox for a brief stint before they were acquired. The company was then Fiserv and it was a little bit much more slower moving than Ramp, I would say. So I had the opportunity to join the team over at Ramp. Seemed like a great position for my background and have been here for about 3 years.

Very cool. And we have a ton of personal crossover. Owner and Ramp are both Redpoint portfolio companies. We're both very excited customers of Momentum. So I feel like, uh, we got a lot to dig into today. And so as I mentioned at the top, we're going to talk about how this business gets run, what's unique about it, um, spend a bunch of time talking about AI, both what you're doing today and how you're thinking about the future. And so we'll start with just an intro question. And I mentioned Ramp is one of the fastest growing startups of all time, uh, monstrous valuation, only of over $8 billion, only a years in. So what the heck is it like to try to build business systems and infrastructure at a company that's just growing that fast?

Yeah, it's— short answer is it's a lot of fun, frankly. The reason why I joined was the, the growth trajectory and the talent and people that I had the opportunity to work with every day. Um, with that comes with every day or week, there may be different priorities. So, so shifting priorities and using and my experience understand what should we work on today to help drive revenue growth for our go-to-market teams. And to be honest, over the last 3 years, and I've talked to people, there's, there's ramp years we call them, is how long you've been at ramp. They feel like kind of like dog years, um, quite a bit longer than an actual year. But it does feel like a few different companies I've been at, uh, in my time here, thinking back at kind of where the state of systems and go-to-market was when I joined versus just a year later and a year after that is, has been quite a big shift. And do you guys still count in days?

I've heard, uh, your CEO mention like the day since founding. He's like, it's day 1,100. Is that a thing?

That's definitely a thing. I think I had my days pulled up. I'm around 1,000, so coming up on that 1,000 mark. Uh, Ramp, I think, is 3,300. I think we have a— we have a quick go link that I can, uh, pull up here. 2131. That's where we're at today. 2131. Nice.

And one of the things that I took from a bunch of Eric's interviews— Eric Gleiman, who's the CEO— is, is just the intensity and speed at which the company is trying to operate. And so what is— what does that do to you in business systems? Because sometimes business systems infrastructure speed is not like— speed is not the optimal— is not the right thing to be optimizing for because you need to build for longevity and scalability and stability. So how do you personally try to manage those two things? This like intense urgency, which seems to be a superpower for the company, but also trying to be really scalable and not be like rebuilding stuff every 4 to 6 months.

Yeah, I think you kind of have to lead with empathy, understand like, hey, is this something that we're all aligned on making a change. And the great thing about Ramp is that we can build that alignment really quick. It's a pretty flat organization. You can send out a quick Slack message or get a Slack channel spun up where everyone's just aligned on a new project or a shift in data architecture or a shift in measurement on, on a part of the funnel that will be super impactful. And then once folks are aligned, it's really just what's blocking us from making this change? Is it a reporting issue? Because to your point, you don't want to change what you're reporting on from board meeting to board meeting and change metrics that are key and not have a throughline of how the business is operating over time. So there are sometimes where there's trade-offs. We don't want to change everything every week, um, but we want to build systems and data capture in a way that we can. So, so I think speed of change is actually one of the, the quiet superpowers that's not talked about quite as much, but the ability to say, this is a great idea, let's make this change at scale at, you know, over 1,000 employees now, and we can still do it really quickly, is is one of the great things about Ramp that I haven't really seen at other companies.

So what's the secret there? Like, what's the secret to moving that fast?

It's just everybody really being on the court and passionate about what they're doing. We're a very Slack-heavy organization. Like I said, very flat too. So people are very autonomous to, hey, this is a good idea, let's make that change. On the system side, it's hiring great talent. It's people that have used business systems before can call out gotchas or, or when we're building in a way that, you know, really isn't that scalable. Um, and what the callouts for are trade-offs for if we're building something in kind of a hacky way that will work for the next few months, but now we're at scale, just having that called out and understanding that that's a piece of debt that we're willing to take on for, for the outcome. Um, but yeah, the other secret is just hard work. So staying, staying on the court, staying engaged and caring about what you deliver.

Yeah, makes a ton of sense. One of the things I found interesting about how Eric was talking about the engineering org is just how much autonomy they could push down. You know, it seems like you guys have built the engineering teams into quite small pods with really specific mandates, and the teams were small. And so is that similar in your neck of the woods as well?

Yes. Yeah. Our team right now on the go-to-market system side, my team is 5, including MarTech, and we have a couple other business systems engineers. So that team I think is 3 or 4 that do kind of Salesforce Apex code and things like that. And yeah, I think our mandate, you know, as a company is to try to save our customers time and money and headcount is a big cost really. So if we can build a bit larger, like enterprise-ready scalable systems with a small team that's, that's very accountable for what they build, that's great. And if we're building things that really need to be built, we need to hire more headcount to support them. That's something that I would need to kind of justify. So we do like to keep lean teams, relatively simple systems where we can, and hire the right people that can maintain and prioritize asks.

So smaller, smaller, more high-powered teams, just like top, top talent.

And one other thing that how we structure is some teams will kind of have a shared resource of systems that different parts of the company can go to with new asks. We are very funnel aligned. So we have systems engineering team or business systems analysts on my team that are aligned to different parts of the funnel like martech, SDR, AE. So there's really one person focused on their, their micro workflows and workflows that they need to work to do their jobs. And that really helps, um, kind of focus all the efforts to being more like a product owner or product manager where you're actually dealing with vendors or different SaaS tools that may be external, but you're ultimately the one responsible for how the AE is efficient in their role and choosing the project work that will make them more efficient.

No better way to build empathy, just set them deep in that org. We're actually just in the process of growing our RevOps team in the same model. You know, we've moved to Agile. The RevOps teams are now aligned to those specific functions. We don't have AE versus BDR, it's sales, launch, CS as the key stakeholders and, and driving that depth, that depth of alignment, that depth of sort of stakeholder engagement, I think is going to, is going to help us move a lot faster.

Awesome. So you have an operation, operational person and a systems person kind of paired. With those parts of the funnel?

Well, we're so much smaller than you guys. We're like 200 employees. So it's, it's one product owner for each of those functional areas. And they are running— they're doing requirements gathering, scoping, solutioning, working with the stakeholder to figure out what we want to build. And then most of our development is done on contract. Both onshore and offshore. And as we scale, we'll probably get to a model more similar to yours where you can have those pods, you can have different resources. But I think the business function alignment means that that individual develops enough depth in the craft to make really good decisions. I find— I have found in my career that the best RevOps leaders that I've worked with have a combination of the technical acumen, but also business acumen and a depth in that craft that they're supporting. My head of RevOps was a former sales leader. He was a second-line BDR leader, sales manager before that, and then transitioned, which means that the decision-making is so much better because I've been in other situations where I've partnered with a more purely technical individual, and the technical answer was great, but it was an answer that had no hope of adoption. You know, it's like, oh, let's just slap some more required fields on that object. Let's just, like, split this out and ask the reps to do this, this, and this. And that's oftentimes set up for failure. Versus if you have an individual with, like, deep role empathy, you can understand. It's like, okay, we're dealing with SMB reps. They're moving really fast. They're probably not going to fill in 6 new fields with a lot of depth. So let's go automate that for them. Like using Momentum, like let's go fill it in.

So, or even worse than that, if you ask them, they'll fill in a totally different answer and then just, oh, they're just, yeah, putting in crap.

And then like you, you're, you know, less than you did before you had no, no information. Fields. Yep. So let's talk about scale and like scaling systems to keep up. So you've been there almost 1,000 days. How much has the sales team grown in that span of time?

Geez, I think we've almost— we've probably tripled since the time I joined. I think we— yeah, we're upwards of around 400 now, I think.

Wow. Okay. And so in business systems, what have you guys done to live up to that very heady, that very heady statement being the best-run private company on the planet? Like, how are you doing that at GTM?

Yeah, I think that the process is kind of what I outlined. We stay very close and in our lines. So if there's a systems issue, every leader knows who to go ask and can escalate. My way, and I see my job is really help prioritize and get out of the way and make sure that we're still building in a way that's going to work for the entire company and funnel, such that we're not building totally bespoke products for different parts of the funnel. Now, how our sales leads kind of love our go-to-market systems or enjoy our go-to-market systems is having real clarity as to what's happening in the field. And you hit on momentum a little bit. That's been a game changer for us. In terms of allowing the individual sales leads to get the information that they may want, uh, without having to have an engineer go make a bunch of changes for one person for, for one alert. So that's been a great tool of saying, all right, sales leads, we can actually turn around your ask here in about 10 minutes, um, and maintain it very easily over time versus, you know, that's a big internal build or we're just going to do a one-size-fits-all for the entire sales org. So as we've evolved our sales org, we've wanted to learn different different things about different teams. And it's been one great tool of just sharing information of what's happening out in the field.

And, and, uh, just going one step more strategic, like, how does the sales team run, um, how does the sales team run in a way that is like different than others that you've seen?

Yeah, I think they're They're all like the way we hire, we hire for slope. Everyone's very, very curious. I've heard Eric speak to the importance of curiosity. So even if it's a new sales tool that's out there, if it's a new product that could make them more efficient, people will dig into that and help write up a value justification for doing a trial. We're very experimental as well. So sometimes the answer is not a blanket yes, we're going to shift to this. It's, All right, let's pull out a tiger team of AEs. Let's test this new, um, gifting is one example. Let's test gifting out at a certain part of the funnel to see if that accelerates our deal, uh, our deal cycles. So pull out a tiger team, did some analytical, uh, got the numbers behind if it was influencing revenue, if it was influencing our close rate or timeline to close a deal. And then we ended up seeing really good results and adopted it. So it's one of those things where if we have a good idea, let's, let's pull out a tiger team. Test it out, and then roll it out abroad, like to our broader teams.

So you're very much operating like a product team, have a hypothesis, goes tested. Yep.

Yeah, yeah. If something's broken, we have to do kind of— we have to do just general maintenance and support and communication around, you know, what's broken, how— what the timeline of fix will be, reporting, stakeholder intake. And then we operate in sprints. Okay, makes sense.

So what have been some of the biggest business systems investments you've made to drive efficiency, productivity, performance, and sales? Like, what have been some of your big wins?

Yeah, I think some of the big wins have been, we kind of hit on Momentum and the AI, just getting the most out of our transcripts, being able to pull specific pieces of information. Send them over to the product teams or channel teams or sales leads. A recent one has really been just, we went deep on our win reasons, lost reasons, and we look at why we're winning deals, why we're losing deals, but there's not much we can do to, in the very near term, to adjust that. We try to kind of rinse and repeat, make improvements in our product, make improvements in our sales process. But we really want to catch those deals before we lose them ideally and pull in the right So, so what we have for managers now is a KG buyer alert or a deal risk alert where, hey, this person doesn't look very engaged. Let's, let's bring in the manager and see, you know, what are some of the reasons they're not as engaged? Um, and let's see if we can, um, nip that or like get to a better outcome before the deal's lost. And where did you build that? That was, uh, that's built in Momentum Slack. Okay. And so the trigger is.

Well, tell me about the triggers. Like, what did you set up to identify a cagey buyer? I love that.

Yeah. Cagey buyer. No, it was an idea from the field. So it was, hey, I would love to learn if someone's being totally unengaged. Let's notify the manager to get involved. And the prompt is actually very long. So a lot of, with the new AI, you actually have to kind of tweak the prompt quite a bit. Here, we fed it some examples of looking at real calls, saying like, Some examples include this, this, camera off, like other things like that, that, that could feed into it. And emails is the other kind of key piece that Momentum offers, which is people don't always break up the over-face-to-face conversation. Oftentimes it's, it's coming through an email, like not right now, maybe we can connect in a few months, we're very busy, et cetera, which is totally fine. But, but, uh, leading into forecasting and if there's any way to accelerate deals or multi-thread, it's great for the manager to at least know that.

That should just be put to bed. Yeah, that makes a ton of sense. So momentum being a big area of investment, what else have been some of your big wins that you've made infrastructure-wise?

Yeah, our tech stack is pretty straightforward. Salesforce, HubSpot on the marketing side, and then Gong. So Gong is kind of our source for transcripts today, and we use that for forecasting as well.

Okay. I, after, I don't even know how long, 7, 8 years, just canceled Gong. It was so expensive. And I'm actually moving call recording into Momentum. And I never— if you had told me 24 months ago that I was going to cancel Gong, I would have thought you were nuts because it's been such an important part of our stack. But it's wild how With generative AI, like, all of that is commoditized now. Like, you can get any, you know, I use Fireflies, there's Otter, there's Fathom, or Momentum doing it for like a fraction of the cost. I'm just peeling it out. Yep.

That's one of the things that probably changed my mind about in the recent years is there used to be this, this North Star in the systems world of tool consolidation, right? Let's let's bring everybody into one, one centralized tool, especially if you're a global company like, like you have. You really need some tool consolidation to tell the story of your business. And operationally, it sounds great, but it's not always the best thing for your teams. So, so we've really optimized towards what's best for our team. Let's select the best-in-class vendor to, to op— for our go-to-market teams to operate the best and, and be the most efficient. So I almost prefer having really strong module products that you can plug in. They, they do excellent, like they do great at what they, they're supposed to do. Uh, you can build or buy these products and kind of plug them in. That, that leaves you in a place where you're not gonna have to make a, a massive shift if you decide to change one thing. And it's actually, um, you still ends up using the best, best-in-class tools. And call recording is one of those too. I think it's something that is turning out to be a pretty easy build, um, to build a call recording software. And so We love all of our vendors, but yeah, it's one of those that we do analyze each year at time of renewals. Hey, could we build this? Should we buy this? What's the cost and what are we getting out of this tool?

Yeah, I have gone through a similar journey and at one point was like, I want everything in one thing. I consolidated everything in Salesloft and then we started using Salesloft including call recording. I was like, oh, it's just not where I wanted to be. And then I bought Gong. I brought Gong in. We didn't have it at the time. And I go through ebbs and flows, I find. But the nice thing is it feels like more tools have pretty decent interoperability today than maybe before. And if not, it's so much easier to build your own through like Zapier or building your own API integrations. Like we integrate into Salesloft in a bunch of like unique ways from Snowflake directly to Salesloft actually.

Okay. And do you use that data in Snowflake then for reporting or?

Yeah, so we have a bunch of stuff that happens in the Snowflake layer, which is like lead scoring. Routing. So in Snowflake, we have two scoring mechanisms that run, one for like deal, deal value, and then one for propensity. So eGMV and eWin, estimated win rate. And then those two things power what we call iRoute. And so iRouting basically just takes the best leads by these two scores and then routes it directly into the cadence for that sales rep. So the BDR doesn't have to do anything. Like, it, it, so iRoute runs at night. If you've got 60 due call steps the next day, we really want you to have 100 due call steps the next day. So it just puts in enough calls to get you to that number of the highest scored leads that are available. And, you know, that, that's not possible with any, like, off-the-shelf tooling really, because, like, it's our own ML. It's our own ML model for both of those. And we want it to pick up a bunch of other data that's not just Salesforce data. But the Salesloft API ecosystem is good enough where you can, you can just ping the API, have that lead get into that rep's ownership, and then plop it right into the, the cadence. And we've got a very capable data ops team. So we're starting to do more and more of that.

That's awesome. And is the data quality that, that's coming in for your leads? I remember in the restaurant space, people are notorious for @gmail.coms and things like that, or— Yeah. Do you have any suggestions or recs for how to kind of de-anonymize that and score it, or what have you used in the past?

Yeah. So we built a very complicated scraping and enrichment system where we were pulling from all sorts of places on the internet. It was getting synthesized, going into a GenAI model, and then eventually, like, we would get to a confidence level of that person's restaurant name, their name, and mobile phone number. And this is all our data ops leader, Jonathan Shankman, and he built this whole thing. And now we actually use a vendor for it called DataLane, which does a bunch of that now. At like a bigger scale with a higher quality. So our mobile phone record, our mobile phone enrichment rate is like 95%. So when BDRs are sitting down, they have high-quality leads that actually have a mobile phone number. Because if you remember in restaurant, like trying to get to an owner is impossible. But if you have their mobile, it's easy. And especially in the US, those mobile phone numbers are easy to get as long as you know as long as you know that individual's location and name, you can hit the People Data Labs API, return that information, and then we've got good sentiment and disposition infrastructure so that as those reps are calling, they can say, right person, wrong number, no longer with the organization. They can give us that feedback and then we can go groom those leads. I like that.

Right person, wrong number. Okay. Yeah.

I have a slide of our dispositions all laid out if you want to, if you want to see it, because the taxonomy there I find is so important to make it easy for the rep to be like, did a call connect? Did it connect? Did it connect with the gatekeeper or somebody, an owner direct? And on what phone number? And then of those things, like, what is my What is my result of it? Not interested, call back later, pitch out. If you can make that really straightforward for the reps, then you get high-quality data. It sounds like you have a similar philosophy, like optimize for the rep experience because then you get good, then you get good insight.

Yeah. I think, yeah, that's, I'd love to see that because I know we have like bad data, but that could mean 3 things. So if there's like, we could split it onto 2 things, right person, wrong number or generic number, but we got a hold. We do kind of split out the outcome of, of the call as well as disposition, but always room for improvement. Yeah.

How do you think about trying to, to understand and build things that are rep-centric that like reps are really actually going to be excited about versus like, oh shoot, here's another thing from RevOps that's going to get in my way.

Yeah, well, most of our project work comes in some collaboration or an idea either from RevOps or a customer or a sales leader. So we try to first understand what do we want the experience of our customer or prospect to be and how can we make it the ideal on-ramp, pun not intended, for them to get onto Ramp. And so what does that look like? Does it mean a call cadence with a, you know, a, um, with a person that, that knows their industry. Um, can we set the SD up with a playbook right away that helps them understand the industry that they're calling into? Because Ramp's unique in that our TAM is massive. You know, we can, yeah, almost every company. So where it's challenging is if we have newer sales, um, sales teams or sales folks as we're scaling up, they have to learn about the automotive industry, the restaurant business, um, very technical CFOs and things like that. So in high-tech businesses, so I think Eric has talked about how we have potato farmers all the way through Shopify. And so businesses are so variant that it is nice to use AI and tools like that to surface the right playbooks to help them understand who they're talking to because nobody, you know, it's just terrible customer experience to jump on with somebody that knows the ramp product but doesn't know their business at all or how they operate financially. So that's what what our goal is to try to first optimize what we want the experience to be and then back into the tools, um, that we use to make it happen in the process, um, to make it happen.

And so what are you doing to try to build the, the segment empathy? Have you segmented the teams? Are you giving them tools to like know this stuff better?

Unpack that. This is where I think we do a lot of experiments. We say like, all right, we're, we're in a multi-product world right now. It used to just be kind of Rampcard was our core, Now we have, I think, over 5 products with, with a new product launching what feels like every couple of months. And we have one, you know, upcoming that, uh, that's exciting. And so, so with that, you know, it's, it's, we need to enable the team on it. Um, luckily the sales team is always excited about our new product launches. But what we're doing is really just building out these playbooks, making sure they're in front of the sellers and embedded in their systems that they use, like Salesforce, um, uh, Highspot enablement tools. And then we're also feeding those with competitive intelligence and understanding, okay, they're evaluating other tools. Let's, let's speak to that and have these kind of playbooks and competitive intelligence in front of them at the time that it's useful, because nobody's going to spend a Saturday just reading through new, new marketing material or new, new product enablement material. They would like it at the time of the call. And we also use AI to kind of ask questions about our product. If we get— I have a note here on AI sales engineer. Instead of saying, we'll take that as a follow-up and we'll connect in an email, we can actually just send a Slack bot, hey, does our product do this? Yes, it does. Here's the resources. Okay, we can answer it on that call now rather than looping in, you know, looping in an engineer.

Hmm. Did you guys build that in-house or is that an off-the-shelf tool?

That was built in-house.

Is that what Toby is? I've heard, I heard Eric talk about Toby built out of the Applied AI team.

Yeah, yeah, very, very similar. So it's kind of more built on our product information and things like that. And then yeah, I think Toby is more, is more built on call transcripts. Okay, cool.

And so I'm going to come back to Toby. I've got more questions about how you guys interface with the Applied AI team. But before we tie off this like systems topic, curious, what do you think are the big mistakes that RevOps and systems teams make at fast scaling companies? Like what are the biggest landmines that people should be trying to avoid if they're— I mean, nobody's going to be on the ramp trajectory, but if you're like $10, $20 million, you're still doubling, like that's fast growth. Where do you think it can go off the rails.

Yeah, I think so. Ramp trajectory, it may actually be not hiring fast enough, right? Like, we want— we have people interested in our product. Can we give them the right level of, of, you know, guidance throughout the sales process to get them active, understanding and using our product and adopting ramp? So that— so not only hiring too fast, that's obviously a big misstep where you have to either downsize or rethink your, your targets. But hiring too slow can also leave a lot on the table if you're really in hyperscaling mode.

And what about on systems building? Like, what are the big architecture decisions that you feel like people can get right or wrong?

Yeah, on the system side, the biggest one that I've seen both at Ramp, Intercom, and past lives has been pricing discussions. So, so how often do you build your pricing to customers? And how do you make it easy for sellers to quickly select, generate a contract, send that out? Once you have that paper process identified, can you get it from, I'm going to commit to using it, to signed signature, this deal's closed? Especially in like high volume sale, like we have in our lower segments, our smaller segments. So, so yeah, I think building the right pricing model to meet your customers with what's fair is huge. And I think it's never going to be absolutely perfect, but Where I've seen missteps is changing it back and forth quite a bit or making it too confusing where it's all good intention, but it's actually hard for a rep and a customer to get behind. So if it's, so that's where I've seen, and then there's a lot of systems levers there too. If you make a very complicated pricing model, it's very hard to build into your systems and build kind of a quick and easy contracting motion. Interesting.

And so what about, like, one of the things I see sometimes is like too many tools too fast. How do you think about like what's the right velocity to try to bring in best of breed versus making it feel like too chaotic and not understanding how anything works, like for the reps? Where do you think the Goldilocks is there?

Yeah, I mean, usually POCs or trials are free. So I mentioned before, I like to kind of, if someone's really asking for a tool, there's usually a need there and there's usually something that they either are not getting or don't know about that we already have to help them accomplish what they're looking to do. So leading with empathy and taking that 15, 30 minutes to understand what the, the team is asking for, if it's, if it's a team issue, if it's more of a top-down, like, this is a gap or a white space that we have and how we're reaching our customers, then going to market and understanding how the best-in-class vendors work. If we can trial them, great. That's what we did with Momentum, ended up trialing them against a couple of competitors and, and realized we're going to select the vendors that have the highest slope. We kind of use that as well for people, like, we want to find people not past experience necessarily, but where we think they'll be in the next 6 months, how quickly they can onboard. And I do the same with vendors. So is this vendor willing to jump on it, like open up a Slack channel, be very responsive to new asks? And can they build, build a product that will kind of suit our needs in the future, not just their current product, but in the next year, 2 years? What's their product velocity and shipment like? So, so that's kind of how we select vendors. I do think we can't just go and trial every single vendor. So there's a balance there. It's usually the way we strike that balance is just stack ranking the asks from our teams and saying, all right, I know we want to check out this vendor, but we have these 3 things that are, that are above that. So we're going to work on these this week. Makes sense.

It's so funny that you say slope. This is my number 1 criteria for picking especially early stage vendors because, you know, like we're, we were, I don't know, $15 million ARR when we started using Momentum. And I think I might have been their first like full paying customer. You and I were both early, but Avara we use for AI sim. We were one of their first like handful of customers. Even for Salesloft, Salesloft, I was a first 10 customer. Chorus, Chili Piper. Like I've been early on a bunch of these things and picked right more more times than wrong. And the thing that I'm always screening for is just like, how fast is their product velocity? Because most of them are going to have a good idea. Like, so CRM autofill, there's a bunch of companies trying to do it, and I talk to them all. And the thing I'm trying to find, one, an alignment of philosophy, like, do they want to approach this problem in the way that I think it should be approached? And so like just in this category, Momentum wanted to give me a lot more flexibility to write custom prompts and wire that into a specific custom field that I can just like spin up in Salesforce and then start routing data into versus another vendor I looked at. It was much more opinionated, like it recorded your calls, it filled out medic and it was like only medic and they, it was like out of the box. And for me, I'm much too opinionated about how I want to run things. And so just like philosophically, I could strike those. And then it came down to two companies and I was able to like do a demo, see some stuff, look at how fast they delivered new product over a couple-week period. And then we ended up piloting whether it's like Momentum or Avara or some of these tools. And if that company if I could give them feedback, see them implement it in the roadmap and ship it, then I was onto something. Like, then I knew. So it's funny you say Slope. That's like my identical path here.

Yeah. We had similar, like you said, there's tools like Gong that own the transcripts. They just maybe didn't build the autofill right away. But in that eval, I think very similar, we had some really easy asks. Momentum is very early on the writing emails to send out, which I think we have probably 5 tools that can technically do that today. Yeah. If the other tools haven't done it yet, they should, they're probably building it because it seems like the lowest hanging fruit commodity. It's generative AI. So yeah, people can pick and choose where they want to get their emails. You could probably do it correctly in Gmail at this point. But aside from that, there is kind of this autofill idea that you just, take a transcript, fill some information in based on the transcript, but there's, there's much more nuance to it. So there's, you know, do you want to override? Do you, do you not want to override? What are the parameters for if you override data? Do you want to do a look back in a higher context window? So these are some of the things that Momentum's actually built out, um, which, you know, that's, that's why I'm very glad we, we went with them in the long term, because now we can look at these, these context windows, fill out information, and be much more opinionated and specific of who and when we're updating things. And, uh, yeah, seen a few other AI newer vendors too that, that are very promising building kind of early at the, the user level where it gives individual users the ability to, to kind of dictate how their AI works, but it's still, still a kind of a, you know, interesting way to do it rather than kind of the Salesforce methodology of here you have admin rights or you don't have admin rights. Yeah. Yeah. Yeah. Yeah.

Are you being paid what you're worth? Pavilion's Compensation Report has become an industry standard for go-to-market leaders to benchmark their compensation, equity, and other employment terms. We're refreshing the report for 2025, and we need your input today. To get early access to the results, take the anonymous survey at joinpavilion.com/com. That's C-O-M-P, or click the link in the show notes.

So talk to— let's, let's, uh, actually, before I transition to the AI topic, I wanted to— there's one other thing you mentioned there about like stack ranking. So I think one of the hardest parts about being in RevOps and systems is like you have all of these stakeholders with all of these asks, and they're all important, and you just get pulled in a million directions. So how do you think about prioritization? Like, what's your framework How do you make decisions and just like talk to me about that process?

Yeah, we went to a— back in the core RevOps days, I was sent to an offsite with Michael Dearing, who was early eBay, but he spelled out the Drake's equation, which is an equation for estimating how many, how much life would be in the universe outside of Earth. And anyways, he kind of brought it over to the revenue Drake's equation where You look at revenue and that basically equals number of leads times lead conversion times average deal size and minus churn. And so with that, we try to map most of our projects to, all right, we really want to write these really tailored emails or get these alerts for intent in this way that maps onto number of leads because we'll see more leads if we have, you know, better triggers for our signals for these, these potential buyers. And so we try to map all of our projects onto some part of that funnel and then estimate, you know, how much revenue will this bring us if we execute on it based on team size. So if like we're building something for a much larger go-to-market team versus a team of 2 or a small tiger team, we'll prioritize that. So we kind of do our internal, that's kind of how I think about priority of projects. Obviously in the RevOps world, there's just non-negotiables like, hey, our fiscal year is turning over, we have to update these things, we have to ship new quotas, we have to put these new plans in place. So there's things that just have to be done. And then outside of that, we work with our leads, operational leads and, and go-to-market leads and just say like, here are the things that we're just doing. We have to, we have to have these done. And then here are the things that we think is the order and correct us if we're, if we're mistook in there. And then also we can add some scope to it. So this is a quick win. Let's put this at the top and then we have our cadence. Weekly. We just drop some notes in, look at the stack rank, and prioritize from there.

And you guys, you said you're agile? Yep. Okay. On a 6-week cycle then? We do 2-week sprints. 2-week sprints. Okay.

Then we try to publish a roadmap for each half of the year with like big ticket items.

Okay. Very cool.

On the date, like we do a daily standup too, and that's where we just kind of we sync on like, just let's list the top 2 to 3 things you want to get done that day. You mentioned days are super important at Ramp. We want to not get through one day without shipping anything. So if I notice there's people that have like 10 things, we try to pick some out and say like, let's delegate these elsewhere or communicate that they're not going to be done today. Cool. The other intake session on that is just jam sessions. So we often meet with our sales leads, but we have optional kind of office hours with individual go-to-market folks where we can do some shadows, empty interviews, understand how they're actually click pathing through our tools and navigating. And that actually gives us a lot of good signal on if things are being adopted or totally not adopted and some new ideas of systems that we can implement or improve. Very cool.

The theme that I'm repeatedly getting here is just is just distance to stakeholder, stakeholder empathy, starting with customer, then your user. And it makes a ton of sense. I mean, it seems common sense to me, but it's not, certainly not the way most orgs run. But this seems like very deeply embedded in how you think. Yeah. Yeah.

I've been on the other side of it too, or I've seen companies or talked to other leaders that feel like they're submitting a ticket to systems and it's going into a black box and that's how things run. And unfortunately, at big companies, I think some of these tickets do get lost and are not prioritized. But the closer you stay to the end stakeholders and the more visibility you can share, I think the better and the faster you'll build. Yeah, cool.

So let's get into the AI topic. So we talked about momentum. I don't even think we really like defined what it is, but basically for us it's a tool, takes in all of our transcripts and then fills in CRM fields and then does a bunch of other stuff, alerts. So when I, when certain things happen on calls or doesn't happen, it alerts us. So that's like a big use case for us. What are your other big AI tools and use cases, both off-the-shelf stuff you've built, BizOps, business systems versus applied AI? I'm curious to hear just sort of like the full scope of your AI utilization?

Yeah, our, our go-to-market engineering team, our growth eng team, has been very early with, with their AI use cases. So we see a lot of kind of tools popping up with that end-to-end funnel of understanding a warm lead, scoring, and sending out emails. We've been doing that for a good amount of time. And then like those email responses, making sure we score those and understanding like, all right, is this a positive intent, low intent email? Is it something we should hop on right away. So that's one, one kind of cool use case, or use case we've been doing for quite a, quite a bit of time. And where's this—

what do you do the scoring with?

Uh, we have some, some scoring that comes through. I believe it's, uh, on top of, um, Front, Front App, on the shared inbox piece. Okay. And then we also A/B test that against, against kind of our, our main SD generated emails. So we're not an AI shop necessarily of just only AI emails. Um, we hold people responsible for the emails that they send and try to use AI to enhance their efficiency and enhance the emails that they send out. Um, one other tool we use on the scoring and AI front is Actively. Um, they use LLMs and inference to look through or Yeah, I'm not exactly— let's see, we'll mark that. I'm not too sure. I don't want to speak to their, uh, their exact use case, but, um, we'll use Actively on the SDR side. And, and what they do is they build an outbound scoring model for us. So they'll look through past email correspondence, past deals, look at, uh, data in Salesforce as well as external data to essentially say like, hey, it's been 9 months, um, this last email said they're going to renew their other contract here in the next few months. Let's route that deal to somebody. So that's kind of an outbound scoring model and it, it, yeah, it gets better over time. It looks at who's actually converting from those emails that are sent and that will retrain the model. So that's kind of a self-learning model that's, that's been really positive for us. Cool.

So you've got email response scoring, lead scoring, Where does this GTM engineering team fit, like, in the org? It seems like, because I heard Rahul has the applied AI team, which was doing some go-to-market stuff, and I'm going to ask you about Toby, and then you've got GTM engineering, you're doing a bunch of stuff. Like, how does that all fit together? Yeah, yeah.

So they're— the applied AI team is kind of an overlay, so we can engage. I mentioned that the company's pretty flat. If we have a new idea and we think, all right, the applied AI team could be very much involved in this, this project, or this is project work for engineering team, we'll engage with that team, they'll, they'll ship things very quickly. I mentioned kind of our operating motions for the go-to-market org, and the systems and RevOps org, we kind of operate as a pod. And then we'll pull in additional end resources if we need to. And so that's, that's kind of how, how it works. And yeah, AI is kind of embedded in quite a bit of our workflows. Now, I know they've worked quite a bit with our risk and underwriting process. That's definitely cleaned up a lot of the human, you know, component of that or aided in the human component of risk and underwriting. They're also very involved in our product. So making Ramp more intelligent and I use Ramp every day with procurement and for expensing trips and travel booking. We're doing a company kickoff here in a week or so. And so it's actually gotten quite a bit smarter with AI and it can basically guess what I'm doing and generate—

The 1, 2, 3, that's the best. Oh, it's amazing. We're a Ramp— we're a Ramp customer. That was our other— it was, it was Redpoint Momentum, and we're a big— we're, we're a very excited Ramp customer. My VP of Finance was actually apparently a really early customer at his last gig, but I just noticed I took a picture of a receipt pretty recently and then it pinged me back, this looks like— and it was like a parking receipt or a whatever, and it's gotten it right every time. And I'm just like, this is awesome. I mean, I'm a nerd about it. I really am excited about this stuff. But it was such an interesting application of such useful AI where I was like, oh, this is the best. Like, this is what you bet on. This is why you bet on RAMP, because it's already a sick product today. But like, you know, it's going to get better faster than everything else.

I love that too. It's like, why do I have to write out a whole message. Let me just hit one, get on with my day. That's one of my favorite features. The other is just like putting your email in for a receipt rather than anything else, and then it will just pull it straight in. You have to do anything. Yeah, that's great too, especially at restaurants. Um, um, but yeah, so, so they're embedded all across the company and, and we engage them if there's kind of an internal build that we want to do. Momentum is so self-service though, like some of these AI applications or use cases on the go-to-market side, my team will build just directly in those tools and momentum or, you know, work with the actively team. So, so they're not necessarily involved in every single use case of AI, but they're, they're deeply embedded in, you know, risk product and some go-to-market projects.

Cool. What's your favorite, your favorite application of AI that you guys have built and deployed so far?

Oh, I think, I think the newest one or the next one, I guess, should be the answer. But no, the I think I like our context window of why we're losing deals and why we're winning deals is, has been one of my, my new favorites. Um, and the reason is because we have all these fields in Salesforce and that's been standard operating procedure for, for years or since I've been in the system is tell me the reason number one, the sub-reason, the sub-reason, and at the end of the month, let's, let's do a big report and look at the pie chart. But, um, but with this, it gives you so much more context of, hey, here's the sections on each part. They didn't hit this section. You know, there was not any product gaps. It was more of a timing and transitional change reason. So it gives a lot more context of why we might be losing deals. And then we could layer on AI if we really wanted to, to pull that information out back to structured fields, right? So it's not a 1 or 0 thing where we have structured fields or we have more useful, useful information. We could take that information, pull it back to some structure. And score and, and report on it. So that's, that's one thing I really like, as well as the handoff. It just gives our CSN team so much more context on deals, um, when they're onboarding and activating. Champions, here are the people you'll be working with, here are the discussions, here's the link to the Gong call, things like that. So like pulling out— how did you build that?

So that's a dossier that goes to CSM, or this is in Salesforce? Give me that. That's a cool one. Explain more.

We have Momentum call notes where it pulls out both all the med pick pieces of who's involved. And we're just shipping kind of the, the win breakdown, which is context window of last 6 months. Let's look at the emails and calls and let's get the key pieces of information. Which products are they committed to using of Ramp, or what, which products we think or infer that they'll be using? Um, what cards or bank accounts are they using? Where are they at in the application process? Um, things like that. What their you know, just kind of some core features that the CSM team usually uses to do it, run kind of a productive onboarding and kickoff call.

And so that's a flow and a prompt you built in Momentum? Yep. And then it pipes into Slack or Salesforce?

Both. So we were looking to DM, we'll probably DM both the CSM and the account executive so that they're on a thread together. We've explored some deal rooms. We haven't really went live with deal rooms quite yet, but the DMs are something we've been Using with a lot of success at the top of funnel, like SD, old world was SDR would hand off a deal to AE, AE would work the deal, and then, you know, they would have to ping each other back and forth on how it's going. Now with Momentum, we have that, uh, a shared channel and just group DM them both and say like, here's the context of that call. Here's how it went. Are we going to move this to an SQL? Are there next steps? Um, and they can do it straight from the buttons in Slack, which is nice.

Very cool. Yeah. The handoff use case is one we haven't done yet, but it seems so obvious now. I'm feeling like I'm an idiot, like summarizing the entire deal. We typically use Momentum just to pipe everything into Fields. Like the, instead of the Slack summary and a big chunk of stuff, like I'm trying to break down into discrete components. Like tell me about this part of the deal. Like, a good example is, so we charge a 5% service fee to the guest when they check out, and that's part of our monetization. That's how we keep the cost low for the restaurant owner. But if the restaurant owner doesn't know that on a sales call and they learn it on onboarding, they're going to be pissed because they feel like you left something out. And so I built a prompt to be like, did the sales rep specifically mention the 5% guest service fee? And then there's a checkbox as a custom field. On the opportunity record and any, any deal that doesn't have that checked, then it can get flagged and notified to a channel with the managers, for example. And so my strategy has been like, put it all in, put it all on the object, put all on the account object, the opportunity object. But you've got me thinking about like, should I then pull that all back out in a summary to hand to an onboarding person that might have like broader context. That's cool.

That's a perfect use case. Another similar one that we've been using is we don't necessarily need to ask every call, but our teams are very interested in which countries people are operating in. Are they multi-country or operating internationally? And where are their employee bases? So, we're pulling that out and saying yes or no, if this was explicitly discussed or you can infer it, please enter an answer. If not, Don't say anything. It might come up in the next call. So that's where it's been really nice to be able to tailor those prompts and say, all right, sellers, we're not going to block you and have you fill out this data that you'll probably put bad information on because it was never discussed, right? Like a lot of times you just don't discuss these things, but you still want to progress the deal. So those are a couple of really good examples of where I think it's really useful.

Yeah. One of the things we've done there originally I wanted to just have Momentum to bring it to the rep and they have to go like edit and enter. And the adoption was like pretty good, but a lot of people weren't even just doing the like, you know, click to accept. And I was like, well, you know, I don't want to just have AI do it all because it's going to be wrong some of the time and I want to make sure the data is good. But then I did the numbers and we ran, we ran an experiment and just like having it fill in the fields. Automatically every time. And obviously the rep has the chance to go change it. Just, we got a larger volume of high-quality data running it that way. Like, not even, not even asking the rep to, to click accept, just having it auto, autofill, and then they can update. Yeah.

I mean, we, we were still accepting for our core fields just to have some sort of stamp of approval there, but it's been awesome for the fields that you know, aren't P0, P1, but we want data there, um, those will just, will just auto-write every time. Um, yeah, so that's been, that's been super helpful. The other, um, use case that, that we're kind of auto-writing for is the— oh shoot, I lost it. I had, I had, I had one other good use case in mind.

Oh, that's okay. And so what about other AI use cases? So, so like we've talked a lot, a bunch about Momentum stuff, what are the other really interesting use cases outside of scoring and anything else to mention?

Yeah, I think there's, we're using a couple tools that are interesting on kind of new information with agents at time. So at time of interest. Clay's an example of one that you can kind of put in information, get an output. Um, ROX is a newer tool that, that we've been using on the go-to-market side, which is, um, very user-friendly and a good way to visualize your books of business and customizable at the user level. So they can, you know, they can add a new column and say, which, which of my accounts are public? We'll go do like some like natural language agent stuff. And oh, cool. So there's a lot of ROX. Yeah. ROX.ai. Similar, very, very new company, but High Slope. And so we've seen some success with them. And yeah, yeah, it seems very user-centric, much better interface than like, than our CRM right now. And it's a bit more configurable at the user level, which is great. Other AI tools. I mean, there's a lot of new AI features just built into our existing tool stack today that been exploring. So, um, trying to keep on top of, you know, what, what outreach is launching that's new, that's AI versus Gong versus, you know, our existing bigger players, Salesforce and Agentforce. And the opportunity there is huge for them. And so, um, determining, you know, how we'll engage with that, try all that, uh, and then also just building internally and saying, all right, let's test something out. Let's, let's, um, let's build, you know, Toby, or let's build a sales engineer that's AI and stuff like that is, are kind of fun, exciting builds that our team is, is able and willing to do really quickly, which is, which is why it's so fun to work here.

Yeah. So talk to me about Toby. I would love to hear more about what that looks like.

Yeah. So that, that was a bot built, I think, in a weekend that, that was embedded in our Slack instance. Ask a question, get, get a response based on some transcript information. And so So yeah, I think it's been adopted pretty well. I'll have to like come back to you on that actually. I might mark this section and just say like, I'm not— I'm trying to get more context on all the Toby use cases, but I don't know.

All right, well, I'll delete the Toby question.

Okay, appreciate you. I think it got refactored to our AI sales engineer. I'm just trying to figure that out. Okay, makes sense.

So how do you think about build versus buy? Like, what do you— you got way more AI, internal AI capabilities than most companies, but when do you decide to go, like, get a tool off the shelf versus have these internal teams build? Yeah.

This is another thing I've kind of changed my mind about over time. I used to, I used to be very much in the camp of let's put together requirements doc with a checklist of which features we're going to need. And let's do a build versus buy analysis. And oftentimes it was much cheaper to buy. So we do a bunch of demos at a bunch of different companies and decide to either build or buy. And now I think in the last few years, we've really just went both routes. Like why not both essentially? Let's build and buy and let's do a little bake off with an experiment of we have one engineer trying to build specific features. Could we maintain it? What feature sets do we need? But let's also engage with, you know, best-in-class vendors and do some demos to understand why they built in a certain way, what they did build, how their customers are actually using it. Would we get additional benefit that maybe we didn't capture on an initial spreadsheet? Because oftentimes you build these vendor analysis sheets, these are really cool features, but you find 10 more that you'd want within 2, 3 months, right? Or you find 10 more that you're using quite a bit. Within a few months. So, um, so my thinking on build versus buy is one, yeah, let's look at the ROI of, of the cost and, and try to maximize or reduce how much we're spending on software tools, but also maximize that ROI of the tool. So if there's a few different features that we would use that will increase our ROI, I'm very much in the buy camp. Um, but if there's tools that we want to have real, real control over And they're kind of a differentiator to our, how we operate, then, uh, then buying is usually the right move. And, and often it's, we're gonna do both, or we're at least gonna put some people on it and do like both with both engineering and, and the go-to-market team. So that's how I've been thinking about it more now is trials are often free. We can see how things work and, and determine the build versus buy.

Yeah, I've really liked the fact that you can pilot tools so much more today than 5, 10 years ago. Yep, you can just plug it in, let it go, see what it's like in production. Especially today, there's so much vaporware, so many people being like, oh, my AI does this and that, it'll tell you every single call, and then you run it and just like hallucinating gobbledygook. And, uh, being able to, to really like pilot tools and vet them out— this is, this is advice I give to people a lot of the time, is like like just push the vendor to pilot. Like most, like if you have some sway, you're a big company or, or it's a sizable deal, most will, most will let you pilot. Um, and it's the ultimate way to just like de-risk like crazy.

Yep, exactly. Um, same page. And yeah, we push for a lot of pilots per year. Definitely worth a demo. And Nothing really takes more than a few hours to get an initial setup, I've found, in terms of software. Now they just connect to a few systems, we get sign-off from security, sign a mutual NDA, and we're off and running. And that's so much faster than waiting 2 weeks to do a demo and they build this massive deck of all their features and get engineers on the call. It's like, let's just try the product out, um, get a small team using it and see if we adopt it. And ultimately, build versus buy, or whichever we determine, sometimes we, we have both or options to to use both and whatever people adopt usually wins because ultimately I think our customers, our internal customers, our sales team, go-to-market teams, if they're using something a lot, there's obviously some value there. So we try to always where we can measure adoption and usage of our tools.

Yeah, it's interesting. I think, you know, like RAMP has been so successful and you clearly just have such significant resourcing that like not a lot of companies do. And that momentum and resourcing becomes an advantage of its own, you know, for, for you like, hey, we're going to look at build and buy and let the user tell us what the answer is. It's such a fortunate position because you're going to get really good decisions. You're going to get the utilization and adoption. And it's interesting how momentum in startups just snowballs. Like you build a lead and then you attract the smartest people. You have more money to be experimenting and give people more freedom and like that grows your lead as a business because it does feel like Ramp is increasingly like separating itself from, from its competitors. I, I, I just as a user of these products and somebody who watches the markets closely, like the, the tech space, like the, the gulf seems to be widening. Yeah.

And to be clear, we don't build and buy everything, but I wish we had the resources to say, let's, let's build, you know, Salesforce or Gong and then buy it as well. But, but no, you You're absolutely right. Like the people that work here are excited to work here. And often the projects of building are actually pretty fun. Like Slack bots, we probably have a few different versions of how we've built Slack bots, interface the transcript data. I know our data team does a bunch with that as well. So building some of these things has happened. Doesn't happen with every software. We're not going to, you know, build a new Ironclad necessarily for redlining internally before purchasing that. But, but, you know, yeah, it is a privileged position to be in. And I think it also accelerates the, the excitement from teams to, to try new things and do new experiments. Yeah, very cool.

So last question before we hop into quickfire. So over the last couple of months, you know, 01 came out in Q4, 03 is now like out in the ether and is being discussed widely in AI Twitter, AIX. Um, how have reasoning models changed what you think is possible, or how are you thinking differently about your AI roadmap with reasoning now, like, really being here?

Well, just one, like, right off the bat, it just makes using these tools so much better. Like, it makes them actually useful. Things like uploading a vendor contract and really pulling out numbers and things like that internally. We have a procurement tool that's just gotten so much better over time. And so with that, it's made kind of our analysis. We just went through budgeting for this year. It's, it's helped with, let's import some contracts, let's consolidate, you know, how, how much we're spending on each of these enrichment vendors. And it just works. So our internal processes have gotten a lot faster for things that might take a person or an analyst a good amount of time. They've gotten faster, um, even writing something simple like validation rules in Salesforce. Like, let's, let's write this, test this for us. I know my team uses that quite a bit. Um, code review. So internally, I think it's made, made things really, uh, really good. Where I see it in market potentially opening up is suggesting things for deals and, and more complex deal shapes, dollar values, discounting, things like that based on your current customer base, your customer data, like what is a good deal shape for this customer? What's, what's their propensity to buy? I think you have, you have that in your models as well of like estimated GMV. I think that will get a lot better if it hasn't already for you. Yeah, but no, I'm excited to see where it goes and I'm excited to layer in more models. I don't have anything very structured today, but I think in the next— if we talk again in the next year, it'll be a whole new world.

Yeah, it feels like, you know, you mentioned advice on deals. Every product I saw that gave you advice on deals, it sounded really cool. You saw a demo, or especially when you pilot it with your own information, the insight was always like pretty lame and like obvious or not super helpful. But I definitely think reasoning models will be able to do such a better job there. The framework I always use is AI to this point has been the world's hardest working intern. You just give it tasks and tasks and make it do mundane research, and the quality is really good, but it's really at an intern level of sophistication, which is great if you need ground level stuff done. And I use AI all the time for my personal stuff and writing and thinking and But the difference in reasoning model, just like in quality of output. Now I'm starting to think, I'm like, okay, well, I had AI in this little box of simple stuff that's pattern recognition and things that are repetitive. But now that aperture is so much wider. Even like, I got sick over the holidays and I went to the O-1 model and dropped in like my symptoms and a bunch of stuff, and it has a bunch of information stored about me. In memory. And the difference between what 4.0 and 0.1 gave me in terms of an output was like night and day and better than, far better than what the telemedicine call I had was. And I was like, that's when I was really like, okay, this is different. So, I'm excited to think about weaving this in in more more places. I think like complex problem shapes is definitely like on the roadmap now. Yep.

And in Perplexity, we've been using that quite a bit as well. It's like just recent stuff, like, hey, this vendor looks like this is breaking with this formula field connector or something like that. And they'll look at the most recent, you know, published information. They'll even scan Reddit, things like that. So it just makes it so much more useful. And yeah, suggested information, suggested action. I'm curious your thoughts is, you know, as a company leader, it has sometimes felt kind of hokey until it doesn't, right? If it switches and says like, all right, this is the quote, we listened to your call, do you accept it or not? And they just generate that for you. It'd be hard to imagine living in a world where, oh, I have to go generate a quote, we'll do all this stuff, whereas now it can just be accepted. And same, we use another tool called Atrium, which is kind of a sales manager tool, has a lot of AI cooked in, and you can kind of show the Show the step changes over time. If it looks like this is, was your standard and this has changed over time, or the people are onboarding are now sending more emails. So you can, you can see kind of anomaly tracking in real time with that tool. What it, um, as it relates to sales teams. And so that's been pretty helpful as well and getting a quick breakdown, even if it's not perfect information or you have more context in your head. Um, does seem like it's not very useful until you do catch something that, that you didn't otherwise know about.

Very interesting. That's Pete Kazanjee's company, Atrium, you're talking about? Yeah. Oh, very cool. That's awesome to hear. Yeah, I think the thing that has really just struck me over the last few months is just how much of the sales process will eventually be like AI-assisted and driven. Your quote generation is a great is a great call. Or, you know, now really being able to give a rep an assist on, like, a consequential next step suggestion because, like, some of these tools that I won't name, it's just like, give them a call, send them an email. Like, this deal is just not helpful versus saying, hey, this is the type of content that was clicked on. This is the type of content that got the highest click-through rate at this deal stage with this sort of context from the transcripts and emails. Like, I think the sophisticated model sophistication is like almost there to do this in a much more material way.

I totally agree. And that's what a lot of companies are sort of converging on. We've seen good signal from Atrium, from Rocks, and from a few others like, yeah, like actively where the goal is really to give you that suggested action. And we have some larger players too, but that suggested next action is something we shouldn't have our sellers wasting time to go dig and find. And if they can get to a place where they're spending their time meeting the customer where they are and actually being good on the phones and that's, that's much better use of time rather than kind of finding that next suggested email to write or action or follow up. And if that's surfaced, then I think their time can be spent doing more, doing things to make them better at their job.

Yeah. I shared this opinion in another interview because people are like, oh, AI is going to replace salespeople. AI is going to replace this, replace that. And my feedback was actually AI is going to replace a bunch of the parts of the sales process that aren't sales. And there's an argument to be made that Salespeople will do more selling than they've ever done. And actually, the craft of sales will make a big comeback because salespeople won't be as encumbered by all of the BS that they don't want to do. I don't want them doing because it's not very valuable. Updating Salesforce stuff, moving something stages, sending like a basic follow-up email, like admin stuff, hand— like a handoff to onboarding. All of this stuff you start to abstract away when the model quality is good enough. And then what's left? It's engaging with customers. It's like being on the phones. It's actually getting deals done. And so I think as much as we think sales will be replaced by AI, I actually think AI is going to let salespeople do way more sales than ever, which would be the world I want to live in. So maybe this is an optimistic view of the world, but I think that's likely.

I totally agree. Sales and strategic thinking, right? It'll pull people out of the day-to-day task and think like, oh, I have this customer, there's, there's a pattern here. Let's engage in new partnerships. Let's, let's do some, you know, some more, you know, co-branding or things like that. Like, let's, let's build these relationships in a way that, that's useful for both the customer and, and the team. And yeah, filling out Salesforce fields is not an ideal use of time for, for a growing and scaling go-to-market team. So, um, excited to slowly but surely, um, chip away at the, the operational tasks with AI. Yeah. Cool.

All right, let's move into quickfire. So, uh, what separates a good RevOps and systems leader versus a truly great and exceptional one in your mind?

I think, I think focusing on the right things and making things as simple as possible, um, are what make a great leader. I think a good leader can say, all right, we've checked these boxes, we have a pick list of our checklist of things to do each, each month, each year, each quarter. A great leader really takes a step back and says, what do we want to get done and how will it influence our bookings or revenue this year? And then focus on the big projects that make a shift. And also, how do we get out of the monotonous day-to-day kind of requests and asks? And I think that's where you simplify things for the field and you make standard processes and rules of engagement that are clear, concise, And, and there's a simple way for exceptions to be made. So I think that's where you can waste a lot of internal operational size if there's just misalignment or confusion around rules of engagement or things that should be relatively straightforward on the sales team. Cool.

What's the most common advice you give to people in your role and trying to be great stakeholder partners to their revenue leadership?

Yeah, I would say just be very curious. So, uh, challenge the team to understand the workflows of the teams that they support intimately and know them better than the actual people do and how to do things better. So, so I put everybody through onboarding of the teams that they support so they, they really feel like they're in seat. I'd say curiosity is the big one. Make sure you ask why and understand what the ask is before we sign up to do any new work. Cool.

And the flip side of that coin, how can revenue leaders be great partners to their RevOps and systems people?

I think just framework for how to ask for things. Like, I usually use the framework, as a blank, I need blank so that I can do blank. And usually that last piece is what What you're trying to dig out of the conversation is what do you want to do that you can't do today? Or what do you want to report on that you can't report on today? And let's jam on a solution. Cool.

What's the hardest lesson you've had to learn as a leader?

Ooh. I think the hardest lesson is sometimes folks aren't right for the role. Here at Ramp, we've been very lucky to get really top talent, but But when people aren't right for the role, just talking to it, being talking about it and being honest and open is usually the right move early on.

That is definitely a hard one to learn, but an important one. Lastly, what's the best thing you've read in the last 12 months? Best book.

I read the book Latitude. I really liked it. Basically about the invention of a water— yeah, a clock that can work well enough to judge latitude for ships back in, I think, the 1700s. I'll have to mark that one.

Wow. Very esoteric. Very esoteric read.

It's a pretty quick read but really interesting. The problem to be solved is they just had no idea where they were latitudinally. So, something that was an interesting problem that was ultimately solved, but good kind of history book. That's cool.

Well, this is awesome. I learned a ton, and it's really cool to hear how you think about the world and see this function that is ever more important and what you're implementing. So, I appreciate the time, Nate. This was a blast. Yeah, Kyle, thanks for having me. Thank you for listening to the Revenue Leadership Podcast. If you enjoyed it, don't forget to subscribe and you can find a link in the show notes and be sure to leave a 5-star review, share it with your network, and please join me next Wednesday for another great conversation.