Inside Exa’s Run to $100M ARR: Building Google for AI Agents | with Marcus Holm (CRO, Exa)

0:00Marcus Holm: Exa is for AI agents, what Google is for humans. You just passed 100 million ARR at Exa. It's been growing at nearly exactly 10x per year for almost three years now. Marcus Holm, Chief Revenue Officer at Exa. Now over half the web traffic as identified by Cloudflare is actually agents and bots. Elon had a tweet not too long ago who said he thinks agents will search a thousand times more than humans. What's the secret to selling successfully to developers? There's nothing I'd rather not do than go to dinner with a stranger who wants to sell me stuff. Over the years, my favorite question has become, share with me an example of a time when you overcame the odds.

0:36Marcus Holm: Having executive level alignment is the unlock in most deals. Anyone who's worked with me will ever tell you, like, oh, Marcus is always asking, like, who's the highest level person we're engaged with? Product used to be the bottleneck. Now it's go to market. When I joined, I think we were 60 employees and now we're almost 200. And it was pretty much all engineering when I got there and here. And then now it's like, GTM is now bigger than engineering. All roads lead back to sales.

1:12Marcus, welcome to GTM Now.

1:15Marcus Holm: Thank you, Sophie. Great to be here. Great to have you here. And now previous to joining Exa, you were actually the president of LaunchDarkly, mature org, over 5 ,000 employees, built out go-to-market function. Exo is quickly becoming a very, very hot AI company. But what made it the company that you wanted to join? Yeah, I think I've developed an affinity for infrastructure companies in figuring out what's that critical underlying infrastructure that can power a lot of applications on top of it. And so as I started to reflect on getting into the AI space, I think for me it was more or less love at first sight.

1:52Marcus Holm: Like when I met the founders, Will and Jeff, and kind of understood from my time at LaunchDarkly and other dev tooling companies that the rise of agents was really taking place and that they're becoming an increasing portion of the knowledge workers and they had to be able to go retrieve information from the web. I thought to myself, like, oh, this market's going to be huge. It feels like these guys have done a phenomenal job of developing that grassroots, bottoms-up admiration of the developers, which is hard to get. You can't just fabricate that. It takes a lot of grit and taste and patience and social posts and documentation to gradually win that.

2:30Marcus Holm: And so I'd say Exa just checked a lot of the boxes in terms of large addressable market, good product market fit, good culture, high velocity innovation. And the developers loved it. So when I was asking around my customers, like, what do you guys think of Exa? Like, API is tasteful. It's super easy to get started. I thought, yeah, that's a ride I want to be on. Tasteful API. That's a green checkmark. Yes. So you listed four criteria there. Is that generally the framework you use to assess companies? I do. I think generally when you meet with venture capitalists, there's kind of two questions of how big can the market get and then what's the likelihood of this company taking an outsized portion of that market.

3:08Marcus Holm: So I try to apply that same mental model. But I think there's also a lot of importance, especially get older and older and balder and balder. You recognize, OK, I think culture is really important, the people I'm working with and getting a sense for that voice of the customer. What are they saying? and I think that also informs like the likelihood of the company taking an outsized portion of the market but yeah I think those are the simplified criteria that matter most when you're jumping heads into a new opportunity. Yeah those are great criteria. What was it like when you met the founders themselves?

3:40You mentioned it was a good experience.

3:42Marcus Holm: It was high energy right because Jeff and Will are they're younger guys they are themselves super energetic and vibrant and so So when I went to the headquarters, I felt like I'd walked into Never Never Land in Peter Pan. Yes. Where I'm like, wow, there's like young people everywhere and they're like flying around. It's like kind of like this big open Coliseum format in the office. You can see everybody like yell across the open space and there's food everywhere and snacks everywhere and ping pong tables. So it kind of embodied the classic Silicon Valley startup experience. Yeah. And I think that infectious energy rubbed off me pretty quick when I met Jeff and Will.

4:21Marcus Holm: And I was asked, I'm like, what's the audacious mission you guys are embarking on? And like, tell me about the customers and the traction so far and what you've learned. And just found them to be very curious, very intelligent, and also very intentional about where they're trying to take in those few years. And it felt like my skill sets were very complementary to their skill sets. And they were looking for a commercial partner to help on the ride. So I'd say from the, in fact, it's funny. So I came home from the first interview and I was having several conversations with different companies. But my wife made this conversation.

4:53Marcus Holm: I can always tell the way you describe the different gigs and you're like, you're torn because there's some really great opportunities out there and there's some amazing founders. But I think there's like the sixth sense of like the way, even if it's unintentional, but the way you describe an experience, people can kind of tell your bias. Like, yep, that's probably the one. And I just had like this emotional reaction to it. Like I thought the tech was cool and the traction was immediate and obvious and the upside was immense. And I just felt like I could do some good work partnering with these guys on it.

5:21So she knew. She knew. She knew.

5:23Marcus Holm: Because then after like a handful of other processes, she's like, yeah, it's good that you're doing that for comparative purposes and trying to figure like. And you also don't know who's going to actually give you an offer, right? So like you're dating a lot of people hoping they like you back. But in the end, only some or none might. so but uh i think in in my mind after the first few interactions of seeing in fact one of my favorite things i recounted to her i was like yeah like i in the third or fourth um phase of interviewing process we went out to dinner the founders now to try to test chemistry and get to know each other and then came back to the office at 10 p.m and all the engineers were still plugged in still grinding and there were people playing ping pong like wow this place never sleeps like it feels like they're planning to win and that was compelling i love that i love that i mean it's like the dating analogy of you have to date to know, but then you also know when you know immediately, but it is good for comparison purposes.

6:15Marcus Holm: Totally agree. Yeah. Well, I mean, clearly a fantastic, fantastic choice, mutual choice, but also choice because you just passed a hundred million ARR at EXA. So a huge congrats. I mean, past a hundred million, you've been growing at nearly exactly 10X per year for almost three or three years now. So first a huge Congratulations. Did you guys do anything to celebrate? Yeah, thank you. We did. We've had some music. There's some champagne involved. Great. It actually coincides really well with our annual company kickoff trip that's happening this next week. We're doing an offsite to a cool place.

6:49Marcus Holm: You might have heard of this island called Hawaii, you know, just saying. Maybe once or twice. Once or twice. Yeah. So it's been amazing. I mean, it's nostalgic of earlier days, like when I was at CloudAird and the Hadoop Big Data rise. You felt a lot of tailwinds and momentum. And then when I joined Heptio with Craig and Joe, who were the Kubernetes founders, they came out of Google. And that was critical infrastructure. And just seeing the demand just kind of pretty much come from a place where there's uncertainty. Like, is this Mesosphere going to win the Orchestrator Award? Is it going to be Docker Swarm?

7:17Marcus Holm: Is it going to be Kubernetes? It wasn't clear. I think similar here, like with the genetic search. Like, it was unclear, like, okay, how much demand or volume will there be for agents to do retrieval? And then what are they going to choose when they get to a place where they're at that fork in the road? and then go EXA. And over the last six months, just watching the growth and the revenue and the adoption of API signups has really gone crazy. So it's been a lot of fun. Yeah, that's fantastic. Well, congratulations again. Very excited for you all. And I know Will, one of the co-founders, a while ago, he called that 2026 would be the year that we'd see more agents actually searching than humans and said it could go as far as 1 ,000 to 1 in terms of ratio.

7:58We are now in the end of 2026. Was he right?

8:03Marcus Holm: Yeah, he nailed it. I mean, I think now over half the web traffic as identified by Cloudflare is actually agents and bots. And when you look at the query volumes of agents on the web, it has exceeded humans officially here in 2026. And Will's in good company around his prediction around 1 ,000x growth by 2030. In fact, I think Elon had a tweet not too long ago who said he thinks agents will search 1 ,000 times more than humans around that time frame. So, yeah, we're bracing for it. Like, I think we humans are going to put a lot of agents to work, and they're going to do everything from booking travel to booking restaurants and shopping for us.

8:40Marcus Holm: And all that requires web search. Definitely. Yeah. And for anyone unfamiliar with Exa, I think a lot of people are. But in case anyone isn't, you can give us, like, just a couple sentence, super simple overview of what Exa is. So I explained to my mom and dad that Exa is for AI agents, what Google is for humans. Perfect. So you and I enjoy going to Google a few times a day to get the information we need. It's a productivity tool. And agents, because they need an API or an MCP to go run their searches, that's not a Google thing. So they need their own Google universe to go to. And that's what X has created for them.

9:17Yeah, it's fascinating to think of that world where it is agents searching instead of humans. And I've heard agents prefer a pure experience. So no SEO, no ads. What does that mean for SEO and ads? Are they dead in the future?

9:32Marcus Holm: It will be fascinating to watch because the monetization vehicle right now for search is definitely ads. And you can look at the revenue and you just think, okay, that's how you make money in this business. We're really trying to avoid that model. We're trying to focus on consumption type model. And, of course, we're more enterprise focused than consumer focused today. And so I think there is going to be a bit of an innovator's dilemma for everyone in the space where you're doing SEO to try to figure out how to guide traffic to your site. And you're charging ads to take space on your search engine.

10:04Marcus Holm: But now in the world of agents, I mean, that's clutter and clickbait and confusion for them that can lead to hallucinations. So I do think the monetization model is going to evolve in the years ahead. And the amount of human queries will decrease because you and I will be leveraging assistants to do a lot of that for us. Well, and similarly, people can use Exa now with no API keys and things like that. What does that mean for the salesperson? Well, I think in this world of low barrier to entry, ease of adoption, the sales role is definitely more focused on the expansion motion where you get a good bottoms up developer adoption, kind of classic PLG muscle.

10:46Marcus Holm: and then sales folks in a position to understand the use cases, the adoption patterns, and the more consultative you can be as almost like a jungle guide to tell a customer like, hey, I noticed you're doing this and this. Have you contemplated trying to optimize your queries in this way? It's better, faster, cheaper. Or other groups and customers like you are using these use cases. Have you guys thought about using XF for that? And so I think to the extent you can help be a connector of groups and of use cases, you can really accelerate that expansion and adoption and impact that classic metric of net dollar retention, which we all care about.

11:24Marcus Holm: It's like, how fast can they expand to us? What's time to value? But I think in terms of the initial land deal, I think that the role of a salesperson is morphing, at least in our business. I think when you sell to developers, that front door has got to be more of a trial experience. They can just get started without talking to you. And then if it turns out the product's good enough, then they'll waste their time talking to you. Right. So it's really that kind of top of funnel initial. What about, say, five years down the line? Is that still true or are agents procuring software entirely on their own?

11:53Marcus Holm: I do see a world where agents are procuring software. I think it'll be a human in the loop for more substantial purchases. So I'd envision you kind of have a threshold of spend where you're cool outsourcing that to the agent. But then more meaningful thing, let me see what you're about to buy. Is this the vendor that I want? And let me ask my peers real quick and understand like the ROI of this or the business case. And what's the adoption plan for this? Make sure it doesn't become shelfware. So five years from now, I still have a hard time imagining we completely empower the inmates to run the prison.

12:24Marcus Holm: But who knows? We'll see. Yeah. Yeah. I mean, I agree. I think at the end of the day, too, there's so many soft skills that people want to assess around a company. They want to know, will the developers at a company like Exxon be there at 10 p.m. cranking away on the computer? Or are they going to be checked out and things like that? And so which company is going to take you into the future? Now, that is very true if it's a human perception on the other end. For yourselves, when you have agents serving and searching, who is your end customer? Who do you consider your end customer? It actually spans all three levels of agent and human and company.

13:05Marcus Holm: Meaning, I'd say the humans are initially choosing the API to integrate into and funneling the agent that way. and the agent is using the tool to discern the sweet spot around price and latency and accuracy and then the company ultimately has got to foot the bill for it right because as adoption goes crazy you need to have checks and balances and a justification for it but i think sort of that top down and bottoms up motion become pretty important because to your comment a moment ago I think the technical win is only a part of the puzzle. You also have to demonstrate the ability to collaborate, to be responsive, to have a compelling roadmap, to be agile around feature requests, and to have key integrations with the partner ecosystem, to maybe have reciprocal opportunities to benefit from each other's technologies and build better together product capabilities.

14:02Marcus Holm: These all weigh into decisions for customers who are choosing technology to buy, and agents don't care about all of that. Their surface area is just like, which one gets me what I want the fastest? But the macro is much bigger than that and humans tend to care. And I think that's where there'll still be like that tri-level sales cycle and you got to do your best to train the sales team and make sure you're covering off all the bases. Yeah, definitely. That completely makes sense. And you mentioned and alluded to adoption and pricing. Yeah. Now with a consumption model like that, The higher adoption typically equates to a higher pricing for a customer.

14:39What does that do for your sales reps? Like how are you structuring at Exa in terms of let's say somebody comes in at 50K and I'm spitballing here, but three years down the line, that deal is 500K. How do you actually structure your compensation in a consumption-based model at Exa? A quick pause. If you run outbound sales, this is big news. Most reps are bouncing between five different tools just to send one email. But there's a better way. Nooks is the AI agent workspace built for intelligent outbound, where AI agents and your reps work together in one place. Nooks' AI sequencing handles the busy work, finding prospects, writing outreach, knowing who to contact and when, so your team can focus on actual conversations.

15:20If you're evaluating your outbound stack or renewing legacy tools, check out nooks.ai. That's N-O-O-K-S dot A-I.

15:29Marcus Holm: Yeah, luckily our founders are very philosophically aligned with a performance-oriented, rewards-oriented culture, and they want us to have a great sales culture. And so we're in a position where we absolutely do reward the sales team for both landing the customer as well as helping drive the expansion of the customer. So the compensation plan for us is you've got varying commission rates associated with the land and what's pay-as-you-go versus what's contractually committed. because in an ideal world, you want to have as much of that predictable committed revenue as possible. But early on in the courtship, that's not always feasible.

16:04Marcus Holm: They're going to want to pay as they go until they get a sense for what this relationship's going to be like. And so I think you want to create the right incentives for a seller to focus on driving the consumption, getting to scale, getting to production, become mission critical. So then the customer's like, yeah, I can see a future with you. Let's figure out how to get this thing contractually committed, similar to what they do with their cloud partners, you know, do an EDP with Amazon. It's like, okay, now you also benefit from economies of scale because we're willing to give you the volume, discounting, give you dedicated resources, sweeteners around compliance and zero data retention and throughput.

16:38Marcus Holm: So there's levers and sort of incentives for the customers to want to go that route. And so you just want to make sure you have a plan in place where it's not a hit and run because that can be a scary dynamic where it lands the customer, flips it over the fence, and then move on to the next one. And consumption is king. The sales cycle never stops in the consumption world. You're always selling. So I think you have to compensate for that. I kind of think it's the clearest version of what every company and sales cycle should actually be. Because in reality, you sell, you flip it over to customer success.

17:10It doesn't end there. So it truly is a continuous. We've just kind of created these two terminologies that bifurcate the process. But in reality, that exactly what you're describing is the best way to approach it.

17:21Marcus Holm: And it's a brain buster for some of us because, I mean, starting in like an infrastructure world and perpetual software world where you just sold a big thing up front and then whether they use it or not was kind of inconsequential. And then for the next decade plus after that, doing SaaS subscriptions where the renewal would get closer and you'd scramble to be like, okay, is the customer healthy? Like, are they going to turn on us? Like, let's do whatever we can in the next three months to ensure they're going to renew. Totally. And now in consumption world, it's like you're always on trial. Like, you've got to be on your game all the time.

17:51Marcus Holm: Because there may not be a renewal per se or some compelling event. You just got to always be satisfying them, which is like the ultimate form of accountability. So you got to make sure you're answering the call quickly and maintaining your SLAs and shipping new innovation and pricing competitively and making them feel like you hear them and you're giving them deals over time. It's just there's a lot that goes into it. Yeah. And to be able to put that much into it, you need the right people. You need the right go to market. strategy and I mean to scale at the scale that Exa has scaled. You clearly have an interesting and fascinating go-to-market overall.

18:29How do you scale at that level? Like what go-to-market strategy has enabled Exa to just skyrocket like that?

18:37Marcus Holm: Yeah, there's a few tweaks early on that certainly helped the revenue accelerate. And so this whole notion of going from a classic pay-go bottoms-up motion where customers are spending modestly, then you introduce an enterprise sales team and you get a little bit more intentional around elevating the conversation and figuring out who else within the organization could benefit from the technology and you start inching more towards these contractual commitments. So then your average deal size starts increasing and terms increase and you start having better executive alignment and understand their roadmap and where they're using agents and where you think you can give them a really compelling alternative to some of the native search options that might exist from like an open AI or an anthropic.

19:18Marcus Holm: And so I think these are examples of areas where we were able to have pretty immediate impact around net dollar retention. I would say like time to value for the customer accelerating quite a bit, going back for like the second putt, the third putt. Because left to their own devices, like they may grow, but just not as fast. And so I think this is where you compensate a sales team is to sort of artificially condense these cycles and induce the customer to move faster than they otherwise would. and so as you build out that coverage model globally with defined sales plays that seem to work well and resonate and you learn the telemetry data to understand where the customer wants to be taken.

19:59Marcus Holm: Really, like I mentioned earlier, you just got to be the jungle guide. And if you do that well, then your win rate goes up, the deal sizes increase and good things happen. I love how you simplify it because I know it is not that simple in practice. Just that simple. And then the other 60 hours a week, you're just on the beach hanging out. Yeah, there we go. Yeah, I wish, if only. So it sounds like bottoms up model paired with a sales assist at certain points to help expedite and accelerate their adoption. You got it. You sell to notoriously anti-sales persona developers. They typically don't like getting sold to.

20:37So what's the secret to selling successfully to developers? You've done it multiple times.

20:42Marcus Holm: Yeah, I'd say the majority of my career has been focused on the developer persona. And I'm not a computer science grad, but I always had an affinity for consumer technology and understanding how it works. And I was an early adopter to try new stuff. And so I think when I got into enterprise sales and started working with the technologists who are shipping the winning features and building the software and the hardware, just getting a sort of a window into what inclines them to want to engage the salesperson. So I think it starts with curiosity. So if you ask them like, hey, what are you likely to respond to and not respond to?

21:17Marcus Holm: And it turns out a lot of them do not love the initial request of, you want to go get coffee? Hey, you want to go get lunch? I'm your new account rep. You want to go get dinner? Like, no. Like, in fact, there's nothing I'd rather not do than go to dinner with a stranger who wants to sell me stuff, you know? And so I think over time you evolve and recognize for them it's all about initially creating value around what are ways I can be using your platform that would be more efficient or what are things I haven't thought about or what are documents that you can point me to or who are some peers you can introduce me to who are going on the same journey I'm going on.

21:53Marcus Holm: and so as a seller um i really like to focus on playing matchmaker thought leader kind of classic challenger sales paradigm if i could help uh kind of stretch them in new ways and become credible not because i'm smarter than now because that rarely if ever happens but just by introducing a new perspective and then they're like oh this person's not just like a money grubber who's like very transactional like they're they actually understand because they ask good questions what that I'm trying to accomplish. And if I believe they can help me achieve my mission, then now I guess I'll go to the dinners and I'll grab coffee with you and be open to the networking stuff so I can build enough of relationships so that I know how to best support them.

22:35Marcus Holm: But it's a human hacking exercise that's very difficult. And for most people, they try, and they're like, nope, I'm out, and they run the other direction. But for me, I've just always had a passion for developers. And then I've got a few family members who are in that world, and I actually have fun dinner conversations of like, so what do you guys like and not like? Like what's the right approach with your type of persona? That is very helpful to have an insight source to bounce ideas off of. Yeah, absolutely. It sounds like earn the right and the trust to be able to pitch any kind of networking stuff.

23:06Marcus Holm: Yeah, and you gotta be willing to develop enough technical acumen that you sound credible. Yeah. So if you show up to meetings and you say, hey, my name's Marcus, this is my SC, they'll take it from here. And you're not gonna snicker about that, but it's pretty common. Yeah. You know, conversely, if you show up and introduce yourself and say, hey, having worked with other developers building a similar application as you, I've noticed they get stuck at this, this, and this. I'm curious, like, how do you overcome that? Like, okay, this guy's throwing acronyms and terms that I wouldn't expect a salesperson to know.

23:36Marcus Holm: And then they'll share. And then as they share, you know, the guard comes down and they're more willing to go through discovery and qualification, which initially they're allergic to because they just think you're trying to sell them something. And so I think you got to put in the work, just make sure you understand what are the trends, how's the technology, what's the architecture, what are the integration points, how do I get started? And if you come off that way, then they're much friendlier. Those are great tips to cracking the code to selling to developers. Yeah. And we mentioned the scale at which you grew at Exa, you need, I mean, many things, but two things primarily that is kind of consequential to our discussion here.

24:12Yeah. The go-to-market that we talked about. And the other is the right people. So you need salespeople, you need individuals on the team overall that take that approach, that have that thoughtful nature, that can best understand the language, speak it, dance the dance, and really be in the trenches and earn that trust with developers. How do you seek out, and this could extend, yes, to Exa, but to all the other companies in your career where you're selling to developers, how do you seek out the right salespeople for that?

24:42Marcus Holm: So this was another reason I was excited to join Exa because our founder and CEO, Will, describes the attributes of an Exin. It's HUA. You've got to have humility, urgency, and agency. And so I think with a lot of the top sales performers over the years, I've been attracted to those who have the technical aptitude. So you have to be reasonably intelligent to sell this stuff. And you've got to have the curiosity to get credible around it. but then you gotta have the drive, the grit, the tenacity to just keep picking yourself up again because you're gonna get rejected most of the time. So if you're easily hurt or offended and developers ignoring you causes you to be down and out, probably not gonna last long.

25:26Marcus Holm: So that grit is really important. And then I think track record is often a predictor of future performance. So looking for the folks who've been to those sales club trips and are often stack ranked towards the top of their team. and I think when you study the habits of those performers, there's a lot of consistency around. They consume podcasts like yours on the regular. They read a lot of books. They tend to interview a lot of their peers who are high performers to understand how they do it. They binge watch gong calls to learn how their peers, in fact, it's always fascinating when you hire a new rep because I'd argue within the first one or two weeks, you already know.

26:08Marcus Holm: Like, are they going to be your top performers? Are they going to make it or not? And it's just all those early behaviors. Yeah. Like, are they just completely engorging themselves in the business? Are they kind of waiting for it to come to them? And I think over time, you pattern match that enough times that when you're interviewing candidates, you can ask kind of layer two, layer three questions where you want examples of how they do that in their personal and professional lives. And how you do anything is how you do everything, I say. So try to get a sense for their habits and what they have sort of committed to their craft to get really good at something and how that ports over.

26:41Definitely. Definitely. Yeah. I like that saying, too. It's a very, very good one.

26:46Marcus Holm: Yeah. I want to dig into the sales side a little bit more here because you mentioned gone calls. Yeah. So when, let's say, a rep's hired within the first two weeks, are you checking how many gone calls they're watching and all those metrics? Ironically, not just me, but my founders were also paying attention to that because they're engineers and they're data oriented and they're curious to see how much are they studying and how are they showing up in their calls? How much are they talking versus listening? But yeah, certainly for our frontline managers and our sales leaders, that's a critical part of the ramp.

27:20Marcus Holm: So you've got a new hire boot camp. You've got certifications. You've got some peer mentoring that occurs from the ramped reps to the ramping reps. but just watching what they do in those first few weeks and how quickly they blaze through their onboarding or they're like kicking the can and procrastinating and you're like oh cutting corners already yeah so yes i think you got to have some pretty clear gates of uh what does success look like and then try to help them because you want to have as high a hit rate as possible and people being successful and retention is is key and so um those activities in that agency it tells a lot Yeah.

27:57And you want to have as high hit rate as possible, but the reality is it's near impossible to hit 100%. So when you know, like how late is too late to let someone go when they're not the right fit?

28:10Marcus Holm: Yeah, I think clarity is kind. So as long as you communicate to them, this is the onboarding plan and here are the certifications and gates that we want you to clear. And a lot is very much within your control. So you're kind of assessing the will part first. And then throughout that, you can assess the skill level of like, okay, is this pretty much plug and play for you? Because you've clearly done this at other places. or as it turns out, you're not so adept at pipeline generation or getting to the economic buyer or building champions. And so then you try to focus on your weekly everboarding activities, how you can help refine their skill set so you can be successful.

28:50Marcus Holm: But I think if you've had the conversation around the will gaps and you'll see pretty quick remediation there, I think you're probably wasting your time because you can't really teach motivation. But conversely, if you have a conversation around some skill gaps, often if they're high will and they meet a lot of those other attributes we talked about that you want to vet out in the interviewing process, then they're a great investment to keep developing them. And watching someone kind of dust themselves off a number of times and turn a weakness into a strength is, I think, one of the most gratifying parts of leadership.

29:26That's amazing.

29:28Marcus Holm: Yeah. And that's once they've been hired. What about before they've been hired? What is a question that you are without a doubt asking that interview that is helping to tell and predict that future behavior? I'd say over the years, my favorite question has become share with me an example of a time when you overcame the odds. And then they'll often try to ask me a question that will, I mean, like at work? No, it could be personal or professional. Just like what comes to mind when like the world was stacked against you and people doubted you could do it and you went out and you did it. and it gives you so much motivation.

30:03Marcus Holm: When the days are hard, you can reach back to that and say, remember when I did that thing? And just the stories you hear are fascinating because sometimes you'll get a really vanilla response. Like, there was this deal that I thought I was going to lose and so then I doubled down with my manager and I won. I'm like, okay, that's great, but that's super transactional. It's not that interesting. That's literally your job. And that's probably half the, I'd say half the candidates answer with the deal. and then the other half were like oh I grew up with a single mom and we were homeless or on food stamps or like I had to get a job at 14 or I had cancer when I was 25 and I went and did all these things and now I'm in remission I said screw cancer I'm gonna beat this thing and like I just love hearing stories like the underdog thing like when I'm supposed to be defeated but I refuse to be defeated and I rise up like those people I gobble them up yeah I feel like that's your probably personal bias too of having overcome a lot of things in your life.

31:02Marcus Holm: Absolutely. Yeah. Cause I just think in a startup environment, when you're in the build phase and it's a lot of evangelical selling where you're trying to educate a market around the way things are now going, it just requires a certain degree of stick-to-itiveness and grit and determination. And there's gonna be plenty of days where the customers will tell you you're wrong. Like that's not how things are going, but you got to be able to manifest that reality. And so I want to see an example of someone who I was like, nope, I'm sticking with this thing. I'm going to make it happen. And that, I mean, you find that in a person.

31:34Marcus Holm: It doesn't honestly matter that much what job you give them. They're going to find a way to succeed. 100%. Like you said, you cannot teach motivation. Cannot. It's innate. Yeah. So that's a way of telling. I love it. I love it. And, I mean, you're assessing, you're growing. Like, who are you hiring right now at EXA? What's the go-to-market look like moving forward now that you've hit 100 mil? Well, we're busy across all go-to-market functions, right? So when I joined in the last six months, it was pretty much like, okay, we had just raised a Series C at a hefty valuation, and we clearly had product market fit, but most of the revenue was inbound through the PLG motion.

32:11Marcus Holm: So now to achieve escape velocity and win the market, you need to rapidly build out global capacity. You've got to figure out your customer engagement model across SDR and marketing of how do you develop top of funnel. And then you've got to figure out your pre-sales motion of how you're going to develop the technical win by way of demos and POCs and workshops. You've got to then figure out your sales motion around the process from stage zero to seven to close and winning a deal in Salesforce. What are the activities, the assets, the resources required to win each stage? and then you got to figure out your post-sales organization like how do we implement how do we expand and drive consumption and instrument the customer health so we get early indication of expansion or contraction so there's a lot that's been and then the ecosystem partnerships yeah because when you're building out the world's information there's a lot of publicly available information on the web that you can index and crawl but there's also a lot of really prized first-party data that you want to add your index from partnerships and so these things are all super critical to harmonize into a cohesive go-to-market strategy so that the customers can ultimately get the best possible experience of getting the knowledge they need.

33:23Marcus Holm: Because I think the world has been pretty spoiled by OpenAI and Anthropik and these model providers in the world of intelligence and sort of reasoning. But in the world of knowledge and retrieval, there's been a deficit there. And so we're super passionate about solving for that gap. So as the reasoning keeps getting better, the retrieval layer of knowledge is accurate enough that we're not hallucinating and can believe what we're seeing. So needless to say, you've got your hands full. Yeah, I've been spending a lot of plates for sure. Yeah, yeah. Hiring all the roles for all the functions. So for anyone listening to this podcast, Sophie, they want a job across anything, go to market or not go to market.

33:59Marcus Holm: I'm happy to plug them in where they need, but we're going to be on a hiring vendor for the foreseeable future. Amazing. Well, we'll have the careers page in the show notes for anyone who is indeed interested. Yes, please. That'd be great. I mean, that's fantastic. You've got an incredible growth trajectory ahead. Very excited for you guys and all that you're building. I mean, prior to this rocket ship that you've joined and you've been building out at EXA, you've got an incredible career. We've mentioned it a couple times at different points. But overall, you know, LaunchDarkly, Fortr, Cloudera, all these experiences.

34:27Like, if you look back at them, what are some of the lessons that you've taken there that you're now bringing with you to EXA?

34:35Marcus Holm: Yeah, I think I've learned that technology is not enough. and you need to pair that up with a really good execution model for customers to consume and expand with your tech, meaning there's sometimes a misnomer. And having worked with a half a dozen technical founders, sometimes there's a misnomer of like, hey, if the tech is so good, it'll sell itself. It's like classic field of dreams, but build it and they will come. For sure. And you've seen these PLG companies get to a certain size and scale, but I feel like they inevitably all hit a wall one day. They're like, okay, we've got to do sales-like growth.

35:06Marcus Holm: Oh, we've got to figure out enterprise sales. And I think that's what I've been passionate about is sort of the through line with these different technical founders is like, okay, you continue to do the research and the product development to figure out what's our North Star? Like, what do you think the market's going to do in the next five years? And then I will do my best to figure out how do we synthesize that down into a digestible, simple set of positioning and sales plays that we can run across different personas, across different verticals. And like that jigsaw puzzle to me is just super fascinating.

35:38Marcus Holm: So I think carrying that value sale muscle with you from company to company, because no matter what you're selling, a lot of tactics on how you can accelerate growth and improve customer retention is pretty similar. What's the biggest unlock you've found for that? For anyone who is similarly looking to transition from PLG and layer on that sales assist or sales lead motion on top, what's been like a kind of a lesson unlocked for you? I think the economic buyer alignment piece, so having executive level alignment, is the unlock in most deals. If you do a lot of deal reviews every week and you deconstruct it, in the end, most of them, their biggest blind spot is you're talking to a developer or like a director of engineering, which is great.

36:25Marcus Holm: They're important stakeholders, the decision. But there's often misalignment between what an individual contributor thinks is important versus what a VP or a CXO thinks is important. And so I think a broken record that anyone who's worked with me will ever tell you like, oh, Marcus is always asking like, who's the highest level person we're engaged with? I just think it's so important in terms of securing what reality is, de-risking the budget approval, the decision. And often when you lose a deal and you deconstruct it, the retro is we're single threaded down low. We should have multi-threaded and we should have gotten up higher.

37:01Marcus Holm: And so I think that's an art. And I think when you look at differentiation and stickiness in the world of AI, where feature parity is easier to achieve than ever because of coding agents, I think now all of a sudden execution and relationships are going to be as or more important because that's still a knife fight. Like if you want to win a large enterprise, it's still the hard yards of going in, building a business case, doing an RFP, get the technical win, get the commercial win, drive the adoption plan, do executive briefings. I mean, this stuff is still brass tacks, like unavoidable as inevitable as death in taxes, if we would say.

37:39Marcus Holm: And so I think if you're thinking through differentiation in the future, it's like, man, you really want to have those relationships. And so you want to hire people who both have them and those who can develop them. Because I still think human in the loop on core technology decisions will remain the case for the foreseeable future. A hundred percent. Yeah, I agree. And I mean, we've talked about how you're gauging sales reps because like you said, they need that ability to be able to build those relationships. Yeah. Or maintain those relationships and or maintain those relationships. Multi-threading is a bit of a science and an art.

38:11Like how do you enable people to multi-thread? How do you teach them to multi-thread? if you need to teach them? Or is that something that's inherent to a great sales rep?

38:20Marcus Holm: No, I think it tends to correlate with their tenure. So more senior enterprise reps understand the importance because they've learned the hard way. I think earlier career reps get happy years, we call it, where you go to a meeting with the developer and they say, hey, I love your tech. I tried it. It's legit. I'm going to buy it. Okay, great. Awesome. Now, let me know what's going to close. and then all of a sudden they go dark for three weeks. What happened? Man, they told me they loved it. And then it turns out some competitor talked to their boss and offered 50 % off and claimed they have all these whiz-bang capabilities that you don't have and you got completely sort of blindsided.

39:00Marcus Holm: And so I think it's convenient in terms of like the networking piece or the swagger to take your shot and get high, I think there's an element of that that can be inherent. And like for me, I always felt like I was willing to go get rejected up top a lot easier than most because of my background and like just haven't been rejected so many times. Some people really like have a hard time with the ego hit of taking that shot and missing it. And so I think giving them the confidence that like, hey, failure is a great teacher. You know, success is not. So let's firstly use all these wonderful AI tools that now exist where you can actually centralize all the common connections.

39:37Marcus Holm: you can sort of form an org chart there are just such easier ways now than when I started my career and you're like basically hunting through the yellow pages and trying to figure out like the company phone numbers and figure out like there's no org chart so I'm like sleuthing this thing out on LinkedIn trying to like make my own but nowadays you have the benefit of demystifying pretty quickly who are the key stakeholders and then work the common connections between your team investor teams just whoever you can get there through. And once they experience that a time or two, you get like that dopamine hit like, oh, we got to the VP, we got to the CTO.

40:17Marcus Holm: And that was the turning point and the opportunity where it was competitive. And now it's not because you understand their KPIs and you align to that. I think that's a pretty special part of the career journey. And as people get better and better at that, their win rates just skyrocket. That's a great, I mean, North Star metric to have and focus on. Absolutely. Super interesting. And AI usage. Let's talk a little bit about that on your team or GoToMarkCrate in general at Exa. What has been the biggest kind of learning for you all in adopting AI? What's really driving your AI usage? What tools are you using?

40:56How are you using them?

40:58Marcus Holm: Yeah. We have an embarrassment of riches in 2026, right? So it is challenging sometimes to prioritize what tools should we invest in. And I think our team has had tremendous success with a tool that I thought was mostly for engineers when I joined, this Devon tool by Cognition, right? It's crawling like a weed. Yeah. And it turns out Devon can be super insightful around how customers are using your product and tell me the history of this account over the last three months and help me understand or develop sort of my value hypothesis. so I think the classic like clods of the world the devons of the world decentralizes the worlds it's all about just trying to have aggregation and for all these different data feeds to a place where a rep can quickly get a point of view on the account the status of it what are the right approach angles to take to advance the opportunity I think the role playing is also getting pretty impressive it used to be really cheesy, but now these AI avatars can simulate a pretty decent customer experience, which if you're a seller who's used to having limited at-bats and you swing and miss, it hurts because you might have worked for a really long time to get that meeting and they ask you a few questions that you just were not audible ready on.

42:14Marcus Holm: So now all of a sudden, if I've got a really cool avatar that ingests my websites and competitive websites and they understand my value framework and stuff, and I can get in a few at-bats a day so that now I'm in a real-world environment and I don't want to whiff it, I think that's another fascinating area of AI that you can leverage to make yourself more productive. Yeah, we have a portfolio company in that space, actually, Avara, and it's fascinating. AI onboarding or sales rep onboarding is their largest use case. And so it's how do you hire globally? How do you enable people, especially when you're scaling so quickly?

42:49Right now, speed is everything. that has been a critical kind of area for them and obviously can expand to other use cases. But the onboarding has been a huge, huge one.

42:58Marcus Holm: Yeah, it makes sense because the balance sheet is pretty heavily skewed towards that quota carrying capacity and go to market. And if they're not ramping quickly and they're flunking out and you have attrition, it's just so expensive. So I think any of these tools that can drive faster time to value, increased deal size, increased conversion, is money well spent. Because we are seeing, I think, in my observation, people are achieving a lot of efficiency in the engineering world with coding agents, and now one engineer is as productive as five or ten engineers previously. So you can, I think, more deliberately and slowly scale the human element of the eng org, is what I'm seeing.

43:39Marcus Holm: Conversely, the go-to-market orgs are still like endless demand, limited supply. And so it's still hard to get enough sales capacity globally, especially if you want those profiles we talked about. Turns out you're unicorn hunting a little bit. And so if you're going to work so hard to onboard that much talent globally, you really got to be intentional about investing enough up front in the tools and the processes so that they hit more often than not. Because it's just so painful when they flunk out. Yeah, definitely. And it is fascinating. the inverse has just happened product used to be the bottleneck now it's go to market I think Cognition said they're 4 to 1 right now for go to market employees to engineering employees and we're turning that way as well like when I joined I think we were 60 employees and now we're almost 200 and it was pretty much all engineering when I got there and here and now it's like GTM is now bigger than engineering and it probably will be a lot bigger than engineering next year right I just keep getting the gap I think I'll just keep getting bigger and bigger because I had one of your earlier questions around like what does a rep do now It's like, well, turns out in the era of AI, you look at Anthropic and OpenAI and Kershaw and Axel, all these companies.

44:42Marcus Holm: I mean, they're hiring so many go-to-market people. And they are the ultimate AI companies who can automate whatever they can automate. But I think go-to-market teams are in more demand now than ever. 100%. We recently pulled the metrics. This is about two months ago, but we pulled the metrics for OpenAI and Anthropics hiring. and go to market across both of them was by far, huge margin, the largest category. Yeah, I can't, I don't think I would have put it to that a couple of years ago. No, no, I don't think it. Because forever in SaaS line, I felt like it was pretty much equilibrium. And most companies I was at, it was like half tech org, half GTM org.

45:17Marcus Holm: And when you looked at your budget for the headcount, you're like tug of warren, say, okay, I get a hundred, you get a hundred. And now it's like, okay, you get 300, now you get a hundred. Yeah, like tug of warren. And that tug of warren, yeah. Yeah, absolutely. And yet the output is higher from NJORG, but it's just fascinating that they don't need as many engineers to do it. Yeah. Do you think we'll get there on the good market side or is it harder to scale in the same linear way? Yeah, that's a good question. I feel like there's a level of automation that could be achieved on the demand generation top of funnel type stuff.

45:45Marcus Holm: So figuring out better, faster ways to get the meeting. And so I know there are like avatar outbounding and like you have a lot of AI voice stuff happening right now where can you actually get a bot to get Sophie to take a meeting? We'll see. So I feel like that's probably more ripe for disruption or as you think through like analytics, like for the longest time, I've benefited from a lot of analysts in finance and in rev ops to help me understand top of funnel coverage and conversion rates by stage and the productivity of the sales team to help guide our investment decisions. But now more and more, you can self-service that because you can ask the agent, like help me understand productivity of my reps from the last six months.

46:27Marcus Holm: How fast did this person ramp? You know, like what are the common themes we're seeing with people who've quit after three months? And it's just like you can, I think over time sort of self-serve more and more of that. And I think that's where the GTM hiring composition may shift to have the agents do what they can. But I think relationship management and like navigating orgs and getting them aligned around a buying decision, that just still seems so complex and dysfunctional that humans are required for some time. At least hopefully until I retire. So if you after that, until my kids retire. Yeah, yeah, there we go.

46:59Marcus Holm: Oh, that's a wrap. They're in high school right now. Yeah, they're in high school. So I just need like 30, 40 more years of human dependence for sales. There we go. There we go. Well, we're trending that way. It's never been more important. It's true. You know, agents that add in general workflows are taking up such a large part of, you know, the bottom of the funnel. But it's almost creating more importance on the go-to-market side. Yeah, for sure. My poor kids thought there was no way they'd ever do what dad does. We're going to go be engineers or go be architects. and now they're like, man, AI, it looks like we might have to go into sales.

47:29Marcus Holm: I'm like, yeah, I'll run the back to sales. They truly do. You only have a business when you have revenue, guys. Come on, pay the piper. So we'll see how this plays out. I love it. Well, Marcus, this has been fantastic. I have a couple last questions for you. Sure. Joining X out, what was the biggest surprise for you? How much communication and decisioning is done over Slack? It's just so fast and furious. I mean, I've used Slack at each company for the last decades. It's not a new tool. In fact, I remember at Cloudera using it for the first time. Like, this is amazing. You know, so like almost 15 years ago now.

48:01Marcus Holm: But then, like, I would say AXA is so frenetic in the Slack communications and the decisions are like a five-minute SLA. And in fact, it's so fast. If you blink, you might miss it. You're like, what did we just decide? Like, what are we doing? Oh, we're shipping this for this customer? Awesome, right? So it's kind of one of those surprises where like, wow, that's innovation velocity. Like, you guys just get the request and you just carpe diem that thing. And then I think that's honestly what I hear the most from customers. Like, why did you choose exit? Like, you guys ship so fast and you give us what we want.

48:34Marcus Holm: So that innovation velocity has been awesome. But I have to, like, keep my head on a swivel of what I just missed. I went to lunch for 30 minutes. Yeah, you need an agent to summarize your slack for you. I just feel like the world passed me by. Yeah, super cool. Fascinating. It's really interesting, actually. A lot of people that are at, you know, AI need a really fast-growing company to say similar things. Yes. Because it's just velocity. How quick can you move in Slack? It is. It goes down really well. Another quick question for you is, do you get to track what agents are searching for? Like, what are common search queries that you're seeing at Exa?

49:05Marcus Holm: Oh, I mean, I feel like in the six months I've been here, there's already been multiple phases where it's evolved. And so you would think this is like a multi-year horizon. But on a monthly basis, the trends are changing. So when I started, it was all with the coding agents that are looking for documentation, like how to do stuff. and then all of a sudden you start to see a rise of like go-to-market agents that want to look for people in places and kind of like classic entity research that could inform a marketer or a revenue leader on their ideal customer profile and get after it then you started to see research by quants and analysts and researchers and financial services where it's like i'm trying to study capital markets or i want to do m &a on this company and figure out diligence and can exit agent go off and help me do that uh and then now at the last call it month or two you've had the explosion of the general agents or like the personal agents where now the agents want to go look up travel and uh restaurant reservations and they want to do research and read your email and based on that go augment stuff that is interesting to you with stuff they found on the web and then make a recommendation on your itinerary and how to plan your 20th anniversary.

50:17Marcus Holm: So all of a sudden now the surface area has gotten so vast where it can be anything from like food to travel, cost of goods over time. In fact, we launched XS Snapshot maybe a month ago because people wanted time series data over goods. And so since we've been indexing the web for a long time, I can show you what Target was selling an Apple for six months ago or two years ago. and like that price of commodities over time turns out hedge funds love that stuff like had no idea right so i would say um the query surface area just keeps growing which is precisely what you want so i just kind of in my i think i even did a linkedin post yesterday with the rise of the agents yes where it's like now you have every vertical agent across these industries who have different care abouts and so then you're just frantically trying to get them the content and the urls and the documents that they care most about so over time it's comprehensive enough and And so like our goal is by 2028 to have a Google scale index.

51:15Marcus Holm: And we're getting really, I think the team is a phenomenal job of refreshing a lot of pages per day. So that when people are trying to set up a monitor or get notified around a change, they can make a decision based on that. But I have been astounded at how quickly the patterns have changed from what types of agents drive the most traffic. And it's not that the other one's traffic has gone down. Like the traffic from some agents is still relatively, like their growth is pretty linear. and consistent. But then like within a month or two, a new startup hits a scene with a different kind of agent.

51:46Marcus Holm: You know, like what just happened? The traffic just goes kabonkers within a week or two. So what do you think drives that? Because one of the ways that I keep track of this is the large company is like, what ads is meta launching? Because essentially I see patterns around all of these ads. And so that's a very kind of outside in vision view of it. But you've got the inside out. Yeah, well, ironically, the more mature a customer is with us, the less we know about their queries. Because when you move from a pay-go plan to an enterprise plan, we introduce zero data retention. Got it. So for privacy, we don't know what you're querying.

52:23Marcus Holm: You know, our job is to try to give you the best price for performance. Ironically, with a prospect who's not on an enterprise plan, you can see their queries. And so you can like literally see like what are they interested in? What content do they want? And so early on in a courtship with a customer or prospect, you learn a lot about what they care about and then you're doing your best to just foot race down that content and then when they get to a size and scale where having an enterprise agreement makes sense now you're actually a little bit more blind to what they care about but that's where the sales team again going back to like account relationship and management strategy you're having those conversations around what content is most important to you like what verticals do you want and like where your internal resources invested right now versus where do you want a sub processor or partner like us to augment those gaps in your surface area so it becomes much more of a consultative collaboration around making sure that you're filling in their cracks.

53:21Super insightful and help me quickly finish these two sentences. The first sentence is the most underrated piece of go-to-market advice is?

53:33Marcus Holm: Practice what you preach. I think in go-to-market, you're asking a lot of your team. And I'm a big fan of servant leadership or lead from the front. So everyone should understand the pitch and the positioning and the competitive landscape and the pricing and the packaging and the public service announcements. So I think in go-to-market, we have a lot of like former competitive athletes, you have a lot of type A personalities, is super judgmental, discriminant in who they want to listen to. I think letting your actions speak louder than words is an effective way to drive, I think, a healthy go-to-market.

54:10You just answered both my questions in one. That's fantastic.

54:12Marcus Holm: Okay. But what's the best piece of advice and who gave it to you? So best piece of life advice I ever got is actually from my mom. I said she's my hero. And I think she might have ripped this off from Mufasa and Lion King, but I just think you remember as a teenager every time I leave the house to go out with friends she said remember who you are I'm like what do you mean? she's like you know what I mean remember who you are like make good decisions and like decide ahead of time what your character is so that when you're confronted with a hard decision it's not a guessing game you don't do a coin toss like Harvey you know and Batman like you get two faces like this is not what life is like life is a series of choices and if you want to develop you know a positive reputation and be a person of high character and to engender trust with people you work with or you live with or your family, friends.

55:01Marcus Holm: You just got to remember who you are. And so I try to begin with the end in mind and figure, like, what is the ripple effect of this decision? Or when I treat someone a certain way, what is the maybe unintended consequence of that? And so as I've gotten older, I would say it actually has forced me to slow down in some ways. I used to move much faster and then there'd be broken glass. So I think trying to stick to my moral compass of like, okay, like even like the golden rule, right? It's just like, okay, if I do this, this is going to probably have a side effect of this person I never used to think about, but now because I've made that mistake, they communicated frustration to me.

55:38Marcus Holm: And now I want to machine learn on that and pattern recognize I shouldn't do that again. So I would say in life and business, remember who you are. That is beautiful. Beautiful. And we're going to credit of your mom not Mufasa. Joanne Holm is a result of credit. There we go. In fact, Mufasa stole it from her. Quote it. Totally. Awesome. Marcus, if people want to follow along, where can they find you? Yeah, absolutely. Find me on LinkedIn. Find us at exo.ai-refores slash careers. Would love the chance to work with many of you that are listening to podcasts, and we're hiring like crazy, so let's do it.

56:09I love it. Thank you for joining us. Pleasure.