Embracing the Unconventional: Sam Mallikarjunan’s Journey from AI Skeptic to Advocate - Ep 23

Sam Mallikarjunan - The Guest

...the first time in history where the capability of a technology has exceeded our imagination for what we want to do with it. We're talking about building the internet of agents. The CMS made it so everybody, regardless of skill and and money, could be part of the internet of content. And somebody, and in this case, it's us, needs to work on building a way for people to be part of the internet of AI. the internet of agents.

Kyle James - The Host

Today, I'm talking with Sam Mallikarjunan, the GM of Agent AI and former head of growth at HubSpot Labs. In this conversation, he discusses the evolution of AI, its democratization, and how it can empower individuals to create and innovate. Sam shares with me the importance of creativity, the emotional aspects of adopting AI, and the vision for Agent AI to empower anyone to build and monetize their own agents. This is a visionary episode for all builders and users of the platform. It's one you're not going to want to miss, so let's get into it. Well, Sam, I'm so glad that we got to chance to do this and to sit down and and and record a session here because you and I go way back. It feels like I've had a lot of guests um from the Agent AI team recently uh that go way way back, but um not everybody here likes know knows you like I do. And I always like to start these things like your origin story, right? Like going way back to HubSpot and kind of weaving that in because I think that'll set the stage for how we get into this like, where's Agent AI wanting to go? What do they want to be? What why are they thinking this way? And I think a lot of your origin leads to kind of that discovery process. So I'll shut up now and kind of like help tell everybody, who are you? Why are you here? What are you doing?

Sam Mallikarjunan - The Guest

Yeah, I mean it's it's fair. The story of my life is essentially um doing things I'm not supposed to do uh and somehow having it usually work out. Yeah, you were actually the first person I talked to at Hubspot back in the day. So, I had no college degree, no like serious previous experience, and I was applying to Hubspot which was one of the coolest tech companies at the time, a bunch of MIT Harvard people. Um and so I knew I didn't have a chance. And so when you know you're not going to be successful, it really frees up your ability to be creative. It's kind of in the middle where you have hopes of success that you kind of, you know, have fear of creativity. Yeah. Uh so I built a website called https://www.google.com/search?q=hiremehubspot.com. Um I signed up for Facebook, Google, LinkedIn ads. You get $100 of free credits when you do that. Uh and I targeted everybody who worked there to register for the free webinar on why you should hire me. Got leads for about a third of the company at the time. Uh company was like 150 people. Um including you and you were uh I I I also remember by the way, didn't you try and claim the referral credit for me or something like that?

Kyle James - The Host

Of course I did. Back in those days, back back in those days of tech, like a lot of companies are like the best hires they could make were people that were referred by other people at the company. And I don't remember what hubspot It was thousands of dollars. It's like, of course I was. So I connected with you on Facebook. It was a Facebook campaign, friend you on Facebook. Hey, let me feed you what you need to know. Make sure you you mention my my name in the process. But I backed out of that, but hey, you know, thousand bucks.

Sam Mallikarjunan - The Guest

uh then yeah, finally back in: 2018

Kyle James - The Host

Well, and and I got to tease you this part. You might not know this part of the story, but like I was, I think I was the first one, the southern boy to make that jump up to New England to move into the cold winters of of Antarctica, which is Boston, Massachusetts. Um, so there wasn't a lot. So you were coming from Florida and I could see that. I was like, oh, like, I've done that transition. Let me kind of help people because I'm trying to get other people to do that. Um, they actually brought me in to help recruit Kip for that reason. because he was in North Carolina. I was in South Carolina.

Yeah. So there's a couple of them and I was like, I was trying to like bring the southern hospitality up. Um, literally I still have business cards this day somewhere in a drawer that says, uh the inbound gentleman because I was very much the southern gentleman of hubspot. So, um, there you go. Easter egg for everybody of of like, We should probably uh get into the point unless of you and I like um pontificating about old days. I want to make sure I I hit because you've worn a lot of hats, right? Hubspot, you've been a Harvard professor, um, Flock, um, one screen and and I want to hit in one screen specifically because you and I spent time there together. You kind of founded that company. onescreen.ai and it was kind of your first like big swing at trying to do something with AI because I'm sure it absolutely had to influence how you think about or thought about how you wanted to build out agent AI here. So like talk a little bit about that, right? It was still yes, it was beginning to COVID, but it was still very early in the age of like starting AI companies. So like, what happened there? Well, I know what happened, but like, how did that work out? Like, what did you learn to kind of help come into this role here?

Sam Mallikarjunan - The Guest

Yeah, I mean I learned a lot of things. So, um, one, onescreen.ai the for those who don't know, the entire idea was that internet marketing is incredibly saturated. Like we've spent trillions of dollars, millions and millions of people. It's become a game of millimeters. Um, ironically, things like out of home advertising are really impactful and shockingly affordable. Right? Like instead of sponsoring a conference. I think we spent like 30 grand to wrap cars, drive them around, have people holding projectors. Like you can really do some good out uh marketing at events, account-based marketing. People usually think about it for for consumer goods, but but it can do a lot more. It was just it was very hard to plan, hard to buy and hard to measure. There was no directory of where things were, much less the ability to know like was this actually successful. So, you know, we partnered with Northeastern University, proved objectively that it can drive e-commerce transactions. It was pretty interesting.

s in call it like December of: 2023

Um so I mean I learned a bunch of things from that. One is definitely add more uh hedging of your bets to your plan. Like Phil Bovergard, one of our first investors from Impelent VC. He's like add two standard deviations to whatever you're going to put in your model because the classic startup thing, right? to just like say like we're going to make money this way and then just drag the spreadsheet out for five years and just assume that that's what it's going to be. Like add a lot more volatility in that because we almost ran out of cash at one screen back in the day, right? We had to for the second time in my career, I had to do layoffs, which was which was brutal. Um, because it's always tough by the way when you have to tell someone that they're losing their job and it's your fault, not theirs, right? Like granted, I couldn't have predicted the pandemic, which it was when I had to lay off at flock. I couldn't have predicted that exact time as when the zero interest rate phenomenon was going to collapse.

But it taught me how to deal with distributed teams, which I think was really important. Um, it taught me to think about supply, demand, its impact on the product road map, etc. Because we have to build, like we couldn't sell things if people don't know where they are, right? Uh, and similar with agents today. So agent drawing the line out to agent.ai. Uh I originally joined just to help out Darme for like a short-term project just to get it launched for last year's inbound conference. Um, and but it was so successful. We had like 100,000 users in the first, you know, few months. Uh now we've got 2.2 million, I think people signed up a year later.

So it's it's really it's really escalated. Um, and in terms of how I think about team design, one thing I think we did well at one screen was we had fairly senior people uh who instead of just hiring the cheapest possible people which then require constant management and attention, hiring like a smaller number of more senior people. that's worked well there. I think that's working well uh at agent AI as well. And then also just challenging how you're going to run teams. Like we don't have we have one weekly meeting on agent.ai. It's team of almost 30 people. but it's because everybody sends loom videos in advance. So like Tuesday, everybody, if you have anything to say, if you had nothing to say, great, that's fine. But we don't have any mugs that say I survived another meeting that could have been an email. Um because you can watch whatever loom videos you want for people, you can ask questions, etc. and then our one-hour meeting is just a discussion.

So like being able to challenge the foundations of how you not just the product because the product is obviously the most important thing, the the marketing etc. But the same the way you arrange groups of people, like the same group of people arranged differently will will either succeed or fail. I now find that even more challenging and interesting because in the age of AI, org charts are falling apart, right? Sales reps can design decks, marketers can write code, engineers can do, you know, deep user research and economics modeling. Like in a world like that, now how do we design teams? So, thinking about it differently, like being fully remote was everybody being fully remote was like a new thing. Um, thinking really deeply about managing operations and cash. So skating to the puck, right? Like be have the operations insight controls everything more sophisticated than you need. And while you shouldn't be scared because if you are, just don't don't do startups. Uh you shouldn't be scared of risk. Do think more deeply about um the stability of the risk environment that you're in, right? Availability to capital, the availability of your customer's capital, like all that kind of stuff. So, we learned a lot. That team's still cranking. I think they still got like 50 people. They they still do, you know, millions and millions a year uh in revenue. Yeah, you were the uh the first employee, Kyle. So, taught me in there too. Yeah.

Kyle James - The Host

So, let me ask you this Sam. What was the particular spark or turning point that made you like really see AI as not just a tool, but as something that is foundational to the way that we work going forward, if that makes sense?

Sam Mallikarjunan - The Guest

Yeah, so to clarify that the AI we have now and most people watching this will know is is generative AI. Um it's essentially we've gone from AI being able to do prediction, which is kind of like what one screen was, what Facebook ads uses, etc. um to having enough confidence in its prediction that we like let it do a lot of. Like actually just write the whole paragraph, write the whole book. I was really very much an AI doomer three years ago, like when all this first started happening. Um, I thought we would go from having millions of websites to having dozens of companies uh and that, you know, even like my local Irish pub, I love telling this AI story. They just use chat GPT, they feed it all the stuff they want to get rid of every week and it comes up with cool recipes and fun names for them. So they're doing like inventory turnover optimization at like Amazon levels. I thought that was going to be restricted to, you know, Amazon would be your shopping AI, meta would be your social AI, uh etc.

Kyle James - The Host

So you thought it would be controlled by the central power entities of the Fangs of the world, not distributed through open source.

Sam Mallikarjunan - The Guest

Every movie we've ever seen about AI, right? Like how many movies AI have we seen where AI is the like the catalyst for for good, right? And and art does tend to emulate future life especially with science fiction. So I was worried about it. I was worried about because in a world like that, there's not going to be local Irish pubs, there was going to be like Irish pub global LTD, um which is going to be really good at running everything. So that's one one of the reasons I was really passionate about agent.ai was Darme's whole vision was, can we not live in the dystopian hell hole that all of us want to avoid but for some reason felt the need to careen towards unstoppingly.

Kyle James - The Host

Democratize it, let anybody try any model to do whatever they want in whatever, you know, not stuck in a specific ecosystem.

Sam Mallikarjunan - The Guest

Exactly. And that's that's why I joined the project. It's also why I stayed on the project. It's why most of the other people on the team are on the project. Like nobody wants to build a world that we don't want to live in. We don't have to. There's no reason we have to do that. So in a world in which anybody, the good thing is the opposite has happened. I am now an AI boomer. There's this these two guys in Lithuania who built this chess site where which's just a chess master and one engineer and they've got like 50,000 MRR already like they launched earlier this year. Like instead of having dozens of of companies who are orchestrating everything. I think we're I hope and I'm more confident that we're going to end up with tens of millions of companies. So it's actually going to have the opposite reaction.

And in particular, I love the ability of AI to help communication. So most people don't know this, but the the transformer technology that language models are built on, originally invented for language. it's a very hard problem. invented for translation. Uh and so I love that like Lithuanian, I probably shouldn't say this as a US startup person. I love that apparently like Lithuanian, you know, startup founders can go toe-to-toe with VC back Silicon Valley startups uh and culture, language, access to capital because you don't have to raise as much. All of those barriers are much lower. So that's why I'm very passionate about this and it sounds melodramatic when I talk to my friends and family about it where they're like, oh, why do you care about agent.ai so much? And I'm like, because I don't want to live in a dystopian hell hole and I don't see many other teams working in that direction, at least not ones with any funding.

Kyle James - The Host

Yeah, yeah, democratization, anybody can compete, anybody can get access to these models because some of these things are expensive and that's I think that's something that, you know, Agent AI and and Darsha really said, no, we want everybody to be able to play with whatever to figure out what works. Um, it's not just, you know, go install your own deep seek server. It's like, no, if you want access to the $200 a month chat GPT or whatever, like here it is. Um, use it through our platform and call it how you want and and and build stuff on top of it.

Sam Mallikarjunan - The Guest

And the least creative way to use agent.ai is to just use us to get access to all the world's models for free. You can absolutely do that, right? The in a perfect world we want you to be building agents, using tools, having agents build other agents, making things public for other people to use, etc. But the idea of charging for access to those models, when performance has continually gone up, although it's plateauing a little bit now, costs have have continuously gone down. It would be like if we launched a CMS company today and started charging by website visit, right? It's like 20 years ago that made sense because server processing was expensive. Um nowadays nobody would do that, right? Like you don't even charge for the for the builder for websites.

And that's the other thing I'm really passionate about is you don't build a website on Wix and then build a website on Squarespace because your customers don't use the internet they don't care what it's built on, they use the internet. Uh agents are very much isolated and siloed right now. So you can build something really cool on make.com, you build something cool on N8N, um but they don't like talk to each other. So we've got 40,000 agents built on our site, 50,000, like 2,000 of which are public. Um make.com has, you know, who knows how many. But the user shouldn't care. If somebody builds a really cool like collectibles grader agent on make.com and I have a, you know, an agent on agent.ai, like all of that stuff should be able to flow together. So, it sounds melodramatic to say like one, if we don't win, like the human race might not make it, I get it. Uh and then two, if we do win, we're going to talking about building the internet of agents. Like the the content management system made it so that everybody, like my wife has a blog. She had a blog as our dog actually, uh where she just pretended to be the dog. The CMS made it so everybody, regardless of skill and and money, could be part of the internet of content. uh and somebody, and in this case, it's us, needs to work on building an internet, building a way for people to be part of the internet of AI. the internet of agents.

Kyle James - The Host

It makes a ton of sense, right? I think there's a common theme that you're really hitting at that I think anybody that's thinking startups or or in general about what they what they're doing is like, you should understand the mission, you should be passionate about the mission and like that should drive you to want to be successful in your job. And and I'm hearing that from you and like, the good thing is is like the mission here is like if if you succeed like humanity ends up in a better place because like you said, there there's a lot of people that are they're giving it a higher than a coin flip chance that this AI stuff ends very, very badly. And I think those of us that are deep, deep in it are very well aware of that, but it's like, all right, how do we stack the deck against that happening? Right? And democratization is huge in that effort, I think.

Sam Mallikarjunan - The Guest

It's it's the main thing, right? Like cuz otherwise, if if money and technical skill or even being locked into context silos, right? Like, you know, Salesforce makes great AI, uh Microsoft, Google, etc. But they're all locking it into their silos because they're worried about getting killed. That is why I was worried we'd end up with like a dozen companies. But it's because it's the unbound creativity that I love, right? This is I think the first time in history where the capability of a technology has exceeded our imagination for what we want to do with it. Right? So mobile devices, we knew we wanted it thinner, we wanted a better battery, we wanted a better camera, yada yada, right? Uh, but AI can already do more than most people can think of.

Beth Dunn on our team, she built a synthetic audience, uh not just like a blended ICP, that's easy, but a multi stakeholder synthetic audience for her speeches that she was giving at inbound. So she could test it, see like, oh, somebody who's junior might say this, somebody who's non-native English speaker might react this way, etc. You and I have a friend who, you know, is very, very intelligent, but like writes walls of text, uses AI to clean that up, send it to us. So because these he's not thinking about AI is like how can I say more louder, which is how most people are using it on LinkedIn and and email. He's talking about

Kyle James - The Host

cutter the mess in my head in a way that's understandable. Yeah.

Sam Mallikarjunan - The Guest

How can I just be more empathetic? Because does talking really matter if nobody understands what you're saying. And and I love this story too. I got a call from my sister. Uh my my 16-year-old nephew in rural Tennessee, um, she's like, your nephew's been called to the principal's office. I'm like, that can't be good. And I'm like, first of all, why am I involved in this conversation? Uh, it turns out the software they use in his Spanish class had crashed. And while the teacher was trying to figure it out, he just built an agent on agent.ai that was better than the software that they had. Uh, and he got called into the principal's office and then he eventually demoed it for um their school board uh to swap out like this much better software for them. That's my 16-year-old nephew in like five minutes while he was waiting for them for the teacher to finish resetting the thing.

Like that's a fascinating world. This is a fascinating world to live in. I don't I don't think it has to be terrifying and I think we've fortunately passed the inflection point of where it's going to be terrifying. I think it's too open, it's too connected now. I think the genie's out of the bottle and the major enterprise players uh who might who don't necessarily care if the world of small businesses and startups dies because they sell to enterprises. Um, if people worry about like wonder what our raison d'être is, right? Why do we do all this kind of stuff? Uh in a world in which there are no small and medium-sized businesses, there is no hub spot. Right? So even though we're not a part of hub spot, that those philosophical missions are very much aligned.

Kyle James - The Host

So speaking of that, one thing that I know that you've really been like thinking out of the box and thinking of this new world with the team, um is encouraging individuals to like build agents to build up their work or like automate what they're doing. A lot of companies or or individuals are almost afraid of like this this secret thing, this helper that they have getting exposed where I think you're very much approach it the other way like, no, guys, if you could go spend half a day to build something that automates what you're doing, it speeds up and you're able to accomplish other things. It fundamentally shifts everything. Like I I'd love you to like unpack that a little bit and explain it because I think it's such a fascinating like it's flipping like switch for people to flip in their head, um but it's but it's coming, right?

Sam Mallikarjunan - The Guest

Weirdly the most popular one liner that I have when I'm speaking uh or even when I'm doing interviews and such is explaining to people that right now we're employees are in an impossible situation because we want them to be creative and innovative and take risks, but they're constantly worried about losing their jobs. Capital markets have changed, right? Like we have to manage efficiency and capital different in businesses. Uh and they're all worried about AI replacing their jobs at the same time. And this is when we're asking them to take risks. Uh and also their bosses are giving them crazy vague advice. They're saying things like, you know, just go figure out an AI strategy. How do we infuse AI into this? like a bunch of BS like that.

Um, and kind of the nice thing about being me is I've reached I've done enough in my career that I feel self-confident enough to tell people that I don't know like what that means. A, your boss doesn't know what it means. I promise you they don't because I told the same thing to people on our team. Like Meredith, I'm like, you know, she wants to uh get uh who's our head of marketing? She wants to get an agency to help with video production. I'm like, okay, but can we get one that uses AI? And she's like, I'm like, at least I'm self-aware. I know that what I'm telling you is vague, but like we understand the spirit of what we're doing.

The other thing is like a lot of people are using AI in secret and they're doing it because they're worried their bosses are going to say like, oh, cool, you know, you've replaced yourself. Now I can lay you off and just use this AI that you built. I really want to do this stunt. I don't think the marketing team is going to let me. I want to like do a hackathon with 10 professionals where we just automate their entire job, have them show it to their bosses. I would love to meet the manager who says, wow, we're in the middle of an AI transformation wave and decides to lay off the person on their team who has demonstrated their ability to use AI to create leverage for their company. I just don't think it's going to happen. And if it does, whoever that manager's manager is should fire that manager and bring the other person back. Like I was even going to offer to sponsor like three months of severance or whatever in the case that there was a psychotic manager who actually did do this.

Um, but I think there's a lot of fear. most of the barriers to adoption for AI are emotional. All right? It's there's a lot of fear, it just feels overwhelming. We all have a lot on our plates, uh etc. But yeah, like uh Rush me on our team, um automated part of her job, wants to help with business development, wants to help with finance, just no background in finance, but it's fine. You know, Harry wants to help with with product stuff. Um, again, like this is what what I mean when I say org charts are falling apart. I do still think people need managers like the middle management layer is dead thing is a little overkill. because middle managers should be doing more than managing project status. That's what Asana is for. It's not what one-on-ones from with an hour for your key employees are for. It should be about like developing the person and helping get them to say the things that that they may not want to say.

Um, so the death of the middle management layer is isn't where I'm going with this. But because like I think people need that. But I think that the death of the fixed org chart is here. right? Where sales reps um can well you're just your your class I mean we used to name our families after what we did, right? Like Smiths were blacksmiths, Wayne Rights built wagons. Now we can't even name our companies after what we do, right? Because if we call it hubspot at the beginning, it would make no sense with the software that they or something besides hubspot, it would make no sense what they do now. And I think now we're going to find that we can't even describe the work that we do in such narrow limited ways.

So that's something I'm really thinking about right now is how do I maintain the support that people on a team need for growth? How do I maintain alignment? How do I have clear decision makers still? Because that's very important. Um, but acknowledging this fact that I want our team to because we have the privilege of of making mistakes and being inefficient because we all know each other really well for the most part, we've worked together for a very long time. I want our team to be able to show everybody else like, here's here is a safe model. You can all now go do this because Sam didn't fire us when we screwed up like the 900 ways not to build a light bulb, the 900 ways not to do our jobs, not how to work uh together with each other, etc. Like I want us to figure that out um because fortunately the the team trusts each other because we've been together before.

Kyle James - The Host

Man, there's so many threads I want to pull on there, but I want to tell a story then dig into the like generalization comment there. Like one, like you remember was it three to five years ago, there was these couple of stories that came out in the news about like software engineers that basically outsourced their jobs to China and they were like doing like three jobs or something like that. Like that's kind of what we're talking about here, but we're actually talking about agents doing it. What what was interesting about those stories is all of us that read it are like, heck yeah, like that's what the guy should be doing. But it was the companies that got all up in arms and I think that's the fear now, but but those perceptions are kind of changing because why shouldn't they? right? And and like I hear you saying is like, that that was a thing and I think it's coming back but it's coming back in a slightly different tilt. You want to you want to react to that?

Sam Mallikarjunan - The Guest

Yeah, I very much remember those days uh and that that was my reaction, right? It's like, why does the company care that the capital that they deployed to get a certain amount of work done is getting that certain amount of work done. Yeah. Uh and it it's again, it's an an emotional reaction and it's very short-term thinking. I remember one of my or like early jobs, this 20 years ago. I built a when Facebook first opened up their API and before I realized social media was evil, I built a uh something a spreadsheet that would just pull the the feed into it. So if anybody looked over my shoulder, I looked like I was working on a spreadsheet, I was actually just on Facebook. Um, like my manager at the time should have wanted to use that creativity for more constructive purposes than just hiding what I was doing.

Or and even in the early days of HubSpot, man, I was the king of freelancers because our expense policy back then was use good judgment. Yeah. Right? It's like you can spend money and you just had to be able to justify, you know, that even if it didn't work out that you had used good judgment when making that decision. And so I would outsource all kinds of stuff. That's how I grew my career, right? I got to lead our expansion into LAN because I hired I like a 20-year-old uh college student in Nicaragua to teach me about Spanish Spanish language and culture and stuff like that. Um, we had designers and freelancers, you know, like that is I think a very good thing.

The companies that reacted poorly back then are likely the companies that would react poorly today. But even back then, those companies were were on their way out. They were going to die, right? If you are trying to impose an order that does not make sense just because you can and it's what you're used to, you are going to lose eventually. Um, if there are companies that are upset that people are using tools and AI to help them help them be more productive, that's those companies are going to die. If you're at that kind of company, I know it it will result in short-term pain. But if you automate your job and show or even significant parts of your job and show it to your boss and they let you go, you should thank them because that company has you know, three years ago I would have said companies like that maybe have 20 years to go. Now, I would say companies like that have less than three to five years and they're just not going to exist anymore.

Kyle James - The Host

Wow. Wow. So, all right, coming back to the the age of generalist now. The interesting thought I had kind of hearing the way you describe it is like, you hire somebody who is a subject matter domain expert on something, they should be able to automate that thing, right? Because they they they know what great is in this thing, right? So if they could automate that and then go start learning other stuff, the cross domain, the ability to see and understand things across domain becomes super, super valuable. And and that way if someone needs that subject matter expertise from this person on that thing and they've just got an agent thing go ask for it, it unlocks everybody, right? And I think that's kind of what you're alluding to, but I'm just like playing out in my head like, this is maybe how this actually works in real practice.

Sam Mallikarjunan - The Guest

It's an interesting nuance and I hadn't actually put it that way before, um, which is, yeah, people pay you generally for specialized skills. With the current state of AI technology, the only people who could really replace a job or even meaningful chunks of a job are those people with those specialized skills. Uh so it it does make it kind of even weirder. I remember this this woman, Diana Urban on the marketing team back in the day. and she like took a week off, taught herself jQuery. Uh she was not a coder and she invented the slide in CTA on the hubspot.com website. Now it's a product feature, right?

But what if all of the people like that could have been doing that stuff? And and that's where we live today. So if you wanted to do that, you don't need to teach yourself jQuery, you can just kind of describe the impact that you want to have. Um, so it's again, as we think about career management, as I I think much more about mindsets and personalities. Um, I I'm very transparent about what I do to our team by the way, because again, they're all fairly senior. but I've intentionally designed the team to have a mismatch in personalities, work styles, and things like that. So, um, people who are very organized, execution-oriented versus people who are very, you know, pie in the sky kind of thing. Because when it's no longer about the specific workflow itself within the bounds of like where you fit on a chart, where you are on a funnel, that kind of stuff. Um, the creativity is I think going to be the thing that's most interesting and creativity comes from conflict.

Um, and again, for companies that are listening, if you think you're going to lay off all of your people and replace them with AI, here's the thing. Everybody's going to have access to the exact same tools for the exact same price. especially if we have our way to say about it and for the most part, a lot of this stuff's going to be free. Uh, if everybody has access to the same tools, the only differentiator is your people. It's you. So, I suspect instead of software companies with 80% margins and services companies with 20, 30% margins, we're going to end up with a lot more blended sales and services companies, uh, where maybe they have 50, 60% margins, but the differentiator is you. Like the the chess example. I think their site's like called https://www.google.com/search?q=chessdoro.com, right? It's it's worth checking out. They've basically made it so that this this grandmaster chess person can coach hundreds of people, analyzing their games and and creating content and stuff like that using AI. So not only did they build the platform, but they're scaling out like that kind of impact to the point where I don't really know if I need to pay for chess.com and chessable anymore, right? Big, big companies with, you know, billion dollar kind of, you know, uh market segments.

And and that's my I haven't written this case study. I really want to because I I want to get in touch with those people and interview them. It's it's fascinating to me that one tech person and one like person with something to say who who understands what it's like to be the person that you're selling to can can build something so quickly that's so much more impactful than companies with thou hundreds of millions of dollars in funding, thousands of people. Like that is that is a really interesting really interesting world. Now imagine that as an org, right? where everybody in your company can do what those people did. It changes how I think about a lot of things. Like I think people should quit faster. I think you should give up on ideas faster, which I've never said before.

Kyle James - The Host

Well, you you're the king of like trying stuff and being okay failing, right? I think like like you talked about that when working with freelancers. I think part of that mindset there is, like, look, I'm just going to hire a freelancer to do it. It's spend a couple hundred bucks. If this fails, it doesn't give me what I want. No no harm, no foul, right? Like that's part of just like try stuff.

Sam Mallikarjunan - The Guest

Yeah, the counter point to that. One screen's a good example, right? Um where we raised, you know, millions and millions of dollars essentially on our credibility because when you're early stage startup, you have no traction, etc. Um and I think like our first $2 million came from former hubspotters, right? each writing like, you know, 10, 20, $25,000 checks. Uh and to have given up there would have seemed like a big personal blow, but also just bad strategy back then, right? Because it's the cost of starting over was so high that usually, if you were to look at it just rationally, staying the course was better than just like giving up and trying something new. Yeah.

In today's day and age, the cost of starting something at all is fairly low. You don't need to raise $10 million. And therefore, both the cost of starting the new thing and the opportunity cost are lower. So like weirdly, I think that if you're running a startup and you're like, okay, this is interesting, but there's another thing that's actually significantly more interesting, maybe totally unrelated, maybe slightly related, do it. It's fine. Like you don't have to stay the course like you used to. I think that giving up faster and really following the whimsies of both our interests and kind of the waves of innovation, time, and economics, I think it's one of the great privileges of the age of AI. You don't have to make one choice at the very beginning of your life and then, you know, you're like my my father-in-law who worked for IBM, had one of the original wooden think signs, worked for IBM for, you know, for decades. So you don't have to name your family after, you don't have to say at the beginning of your life this is what I want to do. You don't even have to say at the beginning of the week this is the work that I want to be doing. That's cool.

Kyle James - The Host

All right, so you alluded to this early. I think this is a good place to ask this now that we're like half an hour into this conversation. What is the long-term, short-term vision of Agent AI? Like if things go wildly successful and 2 million users of it, you know, 50,000 builders now, like maybe that already is. But let's say things continue to go wildly successful. Like, what does success look like? What does it become? Like like people are using this platform, but they're always looking for like, why am I using this platform? What's the long-term, um, you know, can I can I trust this thing long-term? So like share some of that with people because I think there's a lot of confusion and and it just it's just not clear to people, but I think there there's definitely a vision and a plan here and I just love to kind of get that out there for people.

Sam Mallikarjunan - The Guest

yone would like to go back to: 2006

I have debated adding fake pricing to the site just so ridiculous that no one would pay for it, just to stop having people ask me that question because the answer again, it's like it's it's nuanced, right? In the near future, and actually probably by the time this airs, this will be live. Um, we're adding the ability to charge for agents that you make. So you can already require sharing an email uh so essentially lead generation with agents that you make. You will now be able to charge for agents that you make. It's going to make the math very interesting on our side because agents can use other agents. So like the economics of this is is why we've punted so long. It's it's it's quite complicated.

Everybody keeps saying the next billion dollar company is going to be built with three people. I don't want to build that company. I want that company built built on us. So like that's kind of like the can you trust us long term? Yes, right? Like if you look at the the resumes of the people on our team, the people that are involved, the backing we have between Darmesh and Hubspot as a corporation, like we're not we're not going anywhere. And I think the the reaction of the market itself has, you know, has really shown that. We have we whatever amount of money you think we've spent, we've probably spent one 1/10th or 1% of that amount. And however many people you think we have on our team, okay, 28, right? Not not that many. So the market is definitely rewarding us. It's not just because we're that skilled.

In in terms of like strategy, this is this is the challenge. Being a category creator is a good thing, but exhausting. And for any startup founders listening, if you can avoid being a category creator and build something you love that creates value without doing that, please do. Yeah. Like inbound marketing sounds easy, but it's it's really hard. uh to like come up with that kind of stuff. But I think about it in kind of like Shopify. I think has done a beautiful job of this where they've got Shopify, the e-commerce platform. You can build your store on it, right? So an example for that is like allstage.co, uh or https://www.google.com/search?q=metaprompt.com, they're built on agent.ai. Nobody needs to know that. We don't care, right? Shopify doesn't care.

But at the same time, Shopify has the shop app, which has officially replaced Amazon as my primary place to go as like an all-in-one thing. So they've enabled the kind of the one-to-one, here's a seller, a builder, and a customer relationship. And then they've got this deep ability so that if I buy, I mean like half of this stuff, right? Um, you know, are things that I bought on shop.com and but you don't have to be able to sell me everything, right? So they got the best of both worlds. So that's really what I want to see us do is you can build your business on agent agent AI uh and you can build all you can also be like part of an ecosystem, part of a marketplace. Like Etsy enables people to have jobs doing stuff they love that they wouldn't be able to do otherwise because it can be bought with all this other stuff.

So those are the kind of the two uh strategic themes that we have.

Kyle James - The Host

That I think that's really important to put it like a people to really understand. It's like, right? Like, yes, Apple made billions of dollars off its App Store, but Apple probably made 100,000 millionaires off of its App Store. Right? Like, that's that's humongous democratization of people that don't work for Apple, but like we're actually able to make a living and a and a company and a business through those apps that they created. Maybe 100,000 is wrong, but like I'm sure it's tens of thousands of millionaires.

Sam Mallikarjunan - The Guest

Or at least because they don't have to be millionaires, right? Like at least tens of 100,000, probably millions of people who have livelihoods doing what they love that they would not otherwise have had. Yeah. Uh and it's a soft metric uh so it's not one that like we report on here, but I I do have a goal for our team of by inbound of next year, I would love for a thousand businesses to exist, even if it's just one person making income that wouldn't have existed without us.

I think about this with managing, right? I know, I I believe if I was hit by an asteroid today, my impact on the world will have been meaningful, not because of what I've done, but because of the people that I've helped and all the stuff that they've gone on to do. It's it's kind of ridiculous actually. Uh and it's the same thing when you're building platforms and ecosystems is like if we're successful with this, like yeah, like we'll make some money, that's that's fine. Um there's a strategic aspect of this where it does matter to hub spot that, you know, and only large enterprises can use or afford AI. Uh so there's that aspect to it. But like it's it's really about, you know, giving that creativity, that entrepreneurship a place to flourish.

Uh it's why we don't like we don't compete with make.com, the other makers and stuff like that, right? Like I need a better example than this um because it it I don't want to cheap the work that our team has done. But I'm happy with our builder being the WordPress of builders. It's good enough for 95% of the people on the planet. If you want to do something more enterprise grade, go to make.com. Anything you build there can work with anything on our site and that's the goal, right? They can focus on building a really, really cool builder and we can and everybody else can build their own tools, their own data, etc. and we can be like rising tides that raise all boats, we can be like the canals between what are currently separated pools.

So that's our real strategy. It's like get the builders to where and that's what we're about to do and we'll hopefully have done by the time this airs. We want to prove that with what we have built already today, you can build something that people will pay for. So our team and Darmesh are going to spend the next like few weeks working on that. By the way, the team doesn't know this, ironically, the team meeting where I announced it is in three hours. Once we've proven that, my hope is that everybody else will be like, oh, that's really cool and start building businesses of their own. Chess, Dungeons and Dragons, uh you know, interview prep, whatever it is you want to do. You're you're only limited by your creativity, imagination and willingness to take the first step. That's the strategy, man. There's there's no salesforce of the world that sells to, you know, there's only 500 companies in the Fortune 500. Uh historically, they've been able to win anyways, but a lot of their advantages, being able to have lots of people who specialize in things, uh being, you know, people having to raise lots of money for marketing, technology being hard. A lot of the advantages of enterprises are going away. and that makes me happy.

Kyle James - The Host

It's interesting because what I'm hearing you say, like John Marcus, go way back to that episode for anybody, um kind of alluded to this at the end of our conversation that what he sees happening is there's a lot of these small businesses that are going to come out of this. one to three to five, maybe 10 people that are very niche plays. And and what I'm hearing you say is like, you know what, we're leaning into that and we're doubling down on that. Like, how do we make that even more possible for people to use these tools and get access to all these models in a way that like stand up a business. Take a shot, try it. See if you can, you know, if you fail, it's okay. But we're going to be right there to help you try to do that. And that that's cool. That's exciting. That's like like that's democratizing. like it is such a huge powerful way that we're we're seeing that, right? Like these monolithic software companies are going to exist, but you're you're not really seeing any new unicorns being formed now because the niches are so tight and and the way that people could solve stuff is so specific in AI that it's great to solve something for 100 people. It's very different to solve something for a thousand or 10,000, but you get very different economy of scales and there's no way that venture capitals and all they're going to fund something that only solves something for 100, but you can make a very profitable business for a few people with that.

Sam Mallikarjunan - The Guest

You nailed it. and that's why I was like, I don't care whether Apple made 100,000 millionaires or 100,000, I guess we'll call them deca thousandaires, what whatever people with $100,000 is. Um, I I would actually rather argue that it's a net benefit to the world to uh help more people like not be millionaires or billionaires but help them, you know, have a livable life for their family. And again, just think about what would the internet be like today if what was it pets.com or something like that? had to raise $20 million dollars just to build a website back in the day. Like, what would the internet be like today if that had never changed? Right? If none of if like Riverside that we're recording this on, if um, you know, uh chess.com, if granola, I'm just looking at like the tabs that I have open on my screen which are many. Uh, like if none of those companies would exist if you had to raise so much money to do it.

So, that's why I'm so excited about this. And again, I know it sounds melodramatic, but like the world is going to be different and I now believe different in a very good way or in full transparency, I would not be doing this. Like there's no I I can do other things, right? If I thought this was going to destroy the world, I wouldn't do it.

Kyle James - The Host

That That's an interesting way to think about it. It it's like a step function, right? We saw like to your point, like pets.com and then this thing called cloud computing came up and anybody could stand up a data center and do that order as magnitude cheaper and I never really thought about it that way before, but like agents and AI is coming, it's another step function in that to like scale out your team with these agents. You need you need to support person. Great. Here's an agent to do support for you.

Sam Mallikarjunan - The Guest

like that example because in: 2000

Kyle James - The Host

So, as we start to wrap up here, Sam, like I'm curious. Do you if you could, like what is one piece of advice you'd give anybody out there listening who's curious about using their first agent or building their first agent or or my big stick is I think every user should be a builder. But but you know, how would you like encourage people to like like what what is this piece of advice to kind of get going out there?

Sam Mallikarjunan - The Guest

The one piece of advice that I would give, so it's such a weird experience. When I give talks and workshops, is the first time that I've ever wanted people to come in feeling like what I'm about to talk about is way over their heads, totally inaccessible, building AI agents. And then when they leave, they're actually annoyed. They're like, why did we spend an hour listening to him explain to us something so simple? If I could scale that feeling out beyond like getting speeches uh or doing workshops, that's that's the one thing that I want. I I suspect that the impact it can have on your life is is so much greater than you think with so much less effort than you think.

Right? Even if it's just interpreting the morning's news for you. Uh if it's just building a recipe generator that, you know, reflects the dietary needs of your family, as allergies and stuff like that. Um if it's something that helps you learn. These are things, like I at inbound uh last month, man, somebody sell they're somebody who sells to municipalities and they're like, we're not going to read hundreds of thousands of city council transcripts to find out who's in the mood to buy what we're selling. And I'm like, cool, 30 minutes I knocked that up and they're like, you just solved a $10 billion problem in 30 minutes. I'm like, yeah, right? Like that and another thing by the way, like AI will hallucinate, but I find that a a feature not a bug when it comes to innovation because it might come up with things that don't exist, which is kind of the definition of hallucination. Um, but it's because it thinks it makes sense. That's how AI works.

So, asking AI like what it can do, what should you build? Um that's probably a lot of how I use AI. So it's again, it's an emotional thing. Can I make the leap? And and can I build a habit? So all in your mind. Can you make the leap to start? And then just start every day. Look at your calendar and be like, what can I use AI for? Or actually, you know what? I'm going to build that agent later this week. Hopefully, it'll be live by the time this is up. Sits through your calendar and says, what could I use AI to help me with today? I spent three weeks on a project reconciling all of our financial statements, etc. Because by the way, when you're in charge, everything is your problem. So if people who want to be CEOs and hate math, get good at it. We had a hackathon in New York and some like guy from Ernst Young in 90 minutes built an agent that would have done it for me and can do it for me on an ongoing basis. You would think that I would have thought to try to use AI before I spent three weeks on that project. Um, but so it's like, have the courage to just like start exploring, see what you can do, uh and then make it a daily habit to see like what can I use AI to help me with today? Those are the only two things that you need. All the technology exists, the models are there. You don't need to wait for GPT 7.9 or whatever. Right? The technology is all there, the builders are there, the ecosystems are there. The only thing that's missing is is you. It's your ideas, your creativity, and your willingness to to take the first step.

Kyle James - The Host

Love it. Sam, what's the best way for people to connect with you? And and how can people help you?

Sam Mallikarjunan - The Guest

I decided long ago if I'm not going to change my name, I'm going to beat people in the face with it. So you Google anything even close to Sam Mallikarjunan, you'll find me. Um, my website, Twitter, LinkedIn, etc. Um, I will say the best way actually, if you go to community.agent.ai, um, we have a pretty vibrant community of most mostly non-technical people. Uh, and I I like I want to spend a lot more time there because unlike in the early days of hubspot, we thought we had the perspective and we were kind of like sharing it with the world. I think right now, we don't have the perspective. I want to talk about it. Like, if you want to reach out to me after listening to this episode, please feel free, but know that I'm probably going to respond with something like, I don't know the answer to your question. Let's talk about it together and figure it out.

Kyle James - The Host

Any uh any final takeaways, thoughts, things that I didn't ask you that you want to share with the audience before we kind of wrap it up here.

Sam Mallikarjunan - The Guest

Again, I I keep coming back to this being an emotional thing more than a technical thing. For any managers who are listening, hopefully now you have more empathy for what the people on your team are feeling and frankly for what your executives are feeling, right? Like you're upset that they're not giving you a clear articulated strategy, they don't know. So that honesty um between people who are speaking, people like, you know, you and me, but also just like people on your team, I I think it's time for all of us to have that moment of honesty with everybody we report to and who reports to us and just say, like, yeah, I don't know. Let's talk about I'm not going to judge you for not knowing something that I can't figure out either, right? So like let's talk about this together uh and create that psychological safety because the reason everybody keeps saying 95% of AI pilots fail in enterprises. It's not technical, it's not tactical, it's not strategic, it's not economic. It's emotional. It's mental.

Kyle James - The Host

It's so true. I mean, you and I are both very comfortable telling each other and anybody else like, we don't know. And I think that's like the ultimate form of confidence of not feeling like you have to come to something. It's like, I don't know, but let me go see what I can figure out for you. Right? And and like you said, have that in conversations and all. Uh just like this one and and Sam, pleasure to talk to you as always. I I hope that everybody else got as much out of it as I did because every time we talk, a couple of things connect and I think I even got a couple of things to connect for you today, which is always a win for me. Um so once again, thank you for joining us. Everybody out there, if you enjoyed this. I I've said it before, we are on audio now. So find us on your favorite audio channel. Um like, subscribe, leave comments, thumb it up on YouTube. And until next time, keep growing, um keep trying new things and it's okay to not know.