No Priors: Artificial Intelligence | Technology | Startups - Revolutionizing Customer Success with Agency’s Elias Torres

Episode Date: December 19, 2024

Today on No Priors, Sarah sits down with Elias Torres, CEO and founder of Agency, an AI agent for customer success teams. Elias shares his journey from growing up in Nicaragua to founding several companies, leading engineering at HubSpot, and selling Drift for $1B. He also discusses his work consulting with OpenAI in 2022, which deepened his understanding of the business opportunity LLMs presented and inspired him to start Agency. In this episode, Elias offers a unique perspective on the future of AI and customer success, explaining how current software has fallen short and his vision for a new generation where customers have direct relationships, spend less time on tasks, and have software working invisibly on their behalf. They also discuss the evolving landscape of hiring as teams shrink, and Elias reveals his ambitious plan to reach $1B in revenue with fewer than 100 employees. Sign up for new podcasts every week. Email feedback to [e-mail] Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @EliasT  Show notes: 0:00 Introduction 0:34 Elias’ journey to entrepreneurship 2:36 Growing HubSpot to IPO and founding Drift 6:19 Consulting with OpenAI, learning about LLMs, and diving into AI 9:35 Founding Agency to focus on customer success and AI-driven solutions 11:40 What will a customer experience look like in 5 years? 15:48 Company building in an era of AI, as a 5th time founder 18:32 Reducing headcount while raising the bar on hiring 20:35 Key challenges in building Agency and crafting a standout product 23:06 Addressing software flaws and transitioning to an era of intuitive, self-operating solutions 26:27 Timeline for the next-gen software revolution + the power of building from first principles

Hello and welcome to No Pryors. Today we've got Elias Torres with us, repeat entrepreneur and CEO of agency, which is working on enabling every company to make their customers successful. Elias is no stranger to entrepreneurship. He's founded four companies, led engineering at the Juggernaut SaaS Company HubSpot, and most recently sold his last business, Drift, for more than a billion dollars to Vista Equity Partners. Alias, welcome. It's a pleasure.

It's been a dream of mind to be here with you. You're doing company number five. Can you just talk a little bit about who you are getting to this place? Yeah, absolutely. The journey, I think for me, is what's interesting bits about it. I'm from Nicaragua. So I came first generation, could not speak English, right? Imagine me at the back of a McDonald's, you know, reading the printouts to founder number five,

three times with Sequoia, you know, and great outcomes, IPOs, et cetera. It's been an incredible journey, right? I worked at IBM for 10 years, and then I wanted to be an entrepreneur. I just could not be inside of IBM. I needed to be free. I needed to have a chance of huge impact. And so here I am. I'm in Boston.

That's another interesting bit, right? I'm not in the valley. I came to Florida from Nicaragua, and then I had a shot at IBM in the Northeast, and I've been here ever since. You know, one tidbit I will insert because I feel like the tech community has somehow discovered that smart kids who end up in tech often do math competitions growing up. This has been going on a long time, guys. Oh, yes, you were one of the math competition kids.

I was a math competition. I wish I would be like an Andrus or something like that, right? But yes, we're talking, let's say, 1992, 91. This is Nicaragua. But I get picked somehow to represent the school. nationally at these competitions, right? I placed third in the country. It's not a great accomplishment,

but the point you're making is that math is fundamental to this and the ability and the thinking is applicable. And when I came to the United States, I'm in a low-income town in Tampa, Florida, low-income school, public school. And somehow, again, I don't speak English, but somehow the math teacher says, do you want to be in math competitions?

Miu Alpha Theta. And I'm like, sure, I'll join the nerd club and I'm like just had a blast there. I did have a lot of trouble with the word problems but I was able to do well in the other ones. I want to talk a little bit about the last two journeys because HubSpot is a company that everybody knows

that is like an important public enterprise SaaS business now and you've done leadership at different scales like coming out of even the success of HubSpethers, spot where you were leading engineering, like, why go start drift? I think my whole life has been a journey of not having any clue what's ahead of what's possible. Like, I mean, that math competition I had in Nicaragua to go to compete in Mexico, representing my country, and meeting kids from all over Latin America was, like, mind-boggling.

Then I come to the United States, then I'm in Florida. And then you start, like, you know, I don't know, like MIT. stuff like that. So there's always like more. It's just been fun not knowing what's ahead until you get there. And taking HubSpot Public, when Brian Halligan came to me and bought my company performable, so he's like, our goal is to take a public in the next year or two. I had no clue what taking a company public meant. I mean, we had 30 million in revenue, 200 employees when he said those things. Brian says, you're in charge. You and David can do anything you want with product and engineering, we've got to go public. And so on my mind is like, okay, we go in public,

but I have no idea what going public means. That naivete is great, right? We go public, a billion dollar company, three years, everything was great. We went from 30 to 130 million. I said, if Brian can do this, why can I do this too, right? It should be easy to go from zero to 100 million. And that's why I left HubSpot, really. I was so naive, thinking that was going to be easy. Well, you were not so naive that it didn't work, right? So your last company drift, it's sold for more than a billion dollars, and you could do it. And yet you told me you think of it in some ways as a failure, which is like a very odd, uncommon point of view on that sort of outcome. Why would you describe it that way? Look, I'm very happy with my life. I had almost no food growing up,

right? So, like, to me, every year has been a better year than the one before for me. Ever since growing up. So, like, I have zero complaints. But when we left HubSpot, you know, to start a company, I left it at a billion dollars. And I saw it grow to 30, right? And so as I'm building drift, the goalpost moved, I say. The goalpost moved, right? So the day that we sold drift to me, it was a very sad day. It did feel like a failure. It was super anticlimatic. It was literally I was at home in bed, tired. It was like 11 o'clock. We've been waiting to sign the papers, you know, power of attorney. I just got a text from one of the lawyers. It's done, you know. And that day I felt nothing, right? And the next day it felt even weirder and weirder.

It took a long time for all those things to clear up what it meant to sell for a billion in cash, right? A company that we grew to 100 million in revenue, eight years of hard work. But I lost the dream of building a $30 billion company or taking a company public, right? And so I felt at that time I was so tired, worked straight 45 years on my life, you know, and I was like never taking vacations really. And I was worried that I didn't have it in me at the time. If you asked me, are you going to start something else? I would be like, I'm done. It's like, I'm like, how can I start over? Yeah. I know. I know. So I was like, I'm done. It's like, people are like, I want to start a company. I go, that's stupid. Don't do it.

So you took a minute. You're kind of hanging out for a while. Then you started doing some consulting for Open AI and like helping their customers. Like, where do the energy come from again? As Pat says, like I have like, I'm like I have the energy of a teenager. I was a little tired. You're aging in reverse. It's a little weird, Elias. It's a little weird. You too. It's like, we're lucky. What happened was, I was tired. I was in, I was November of 2022. What happened?

Chat GPT. Chi-GPT. And the launch of conviction, everybody. This is very important. At the same time? Yeah, it was... October, yeah. October.

No, you knew. You stole the thunder. So I was in Brazil. Actually, technically, that was the first real vacation I took post-acquisition. Because I worked, I stayed adrift for almost two years. And I was there. And I kite surf, that's one of my hobbies. I have hobbies now, you know?

Nice. Poor immigrant does not have hobbies. hobbies, but now I have hobbies. For the first time, I felt like I get up in the morning, we'll kite surfing the afternoon, we just have breakfast, we feel the ocean breeze, just four of us, four friends, and we were just like kite surfing. And I'm explaining chat GPT to them, right? And I'm like, this is insane. I don't even know how to explain chat GPT. I don't know what an LLM is, October 22nd, technically speaking. I knew Bert, I knew Word of Act, I knew transformers, I don't know this. And I'm like explaining to them why this is different.

So that's the first moment. I had the chat chip between moment where I was like, you know what? Again, I missed the boat. What happens? Why wasn't I working for open AI? What wasn't like, is this it? Is this game over? Many of us asked ourselves that question. People are still asking themselves this question. And so that was the first thing that happened, Catalyst. I've already gone through that journey of a post-exit founder, but I'm already coming the conclusion. The only thing I know how to do is build, and I want to learn. And so I get introduced to Open AI just to do like one of those networking calls, and they tell me, we're drowning, customers are asking us for help. We're 200, 300 people. We don't, we can help them implement

their own solutions. We don't have the bandwidth. So we're looking for people. We're trying the big firms. And it was an idea they were giving me, and I said, I want to do that. I want to do that. I want to do that. I don't know anything about it. I've never used the GPT API, but here I go. And so they were like, are you sure? They looked me up and they're like, is this what you want to do? And I said, yes. And so I did consulting for them. And I started from the bottom, right? People were like, who are you? I was doing some support tickets for them, et cetera, explanations. But then I start getting contract like the MBA, you know, ticket master. I'm in Dubai talking to customers, Red Bull, and I'm just having a blast with a small team, implementing

agents, LLM apps, and Claibio, one of the customers too. And it's my whole world changes, right? It's like, this is the dream as an engineer to build solutions that are this intelligent. Okay, so you come around to the idea that you have this like capability that you're actually really inspired about. How did you end up thinking about customer success? Like, can you sort of give us a line about, you know, what agency is and then landing on that idea? Yeah, I mean, agency is really the lessons learned in all my prior four companies, right?

Startups, we know how to like build products. I mean, we know how to raise money, hire people, build products. We know how to sell it. Sometimes we know how to market them. But I don't believe we know how to take care of our customers, especially in the B-2-B. I think managing customers both at scale when you have tens of thousands or 100,000 of customers, it's, I don't think anybody has the answer to do that, right? This is something that happened in the past 10 years. And we're struggling as companies, right? How'd you end up

focusing on CS? Yeah, I ended up focusing on CS because one of the, one of the customers that I started, I'm a good friend with Andrew. So Andrew used to work with me, a performable, Andrew Bilecki, CEO of Claibio, Boston-based company, amazing success IPO. I did some consulting for him. And one of the ideas was, how do we help scale the customer success organization, right? At Drift, I built in 2019 the first AISDR. But now I was helping, how do we help customers at scale? Because he has hundreds of thousands of customers. So when I started solving that problem, which was a little bit of applying my experience between marketing and LLMs is when I realize

the amount of help that we can provide as a company to help scale benchmarking, reporting, inside generation, idea creation to the customers that they have at scale. Because it's something that human struggle, right, to do that in a very detailed manner. And so that's kind of where the idea was born, right? It was like, wow, I can do this for all B2B companies.

What do you think CS looks like five years from now? It's not been a focus of a lot of AI applications today. One would argue that investors and CEOs often don't think of it as like super strategic. You can be better or worse at it. But does it become much more powerful somehow? I think CS is everything. I think the word CS maybe is the misnomer. I think that that's where people are getting lost.

I think, no offense, right, but, you know, VCs are like saying, oh, here's CS, here's a category, how many software companies are in it, what's the TAM, how much is the revenue, add up the valuations, is that something that we're interested in? I'm not necessarily interested in CS, right? I'm interested in customers. I've been in the pre-sells for a world for like 15 years. I know that like the back of my head. I'm going to try and true customer-obsessed founder, right?

All my companies, I'm the one. who services the customers. And when I went to HubSpot, I went from 20 customers to 5,000. And I was crying because I couldn't, they call me and they would say, Elias, help me get on the screen and help me fix HubSpot for me.

And I would say, no, I'm sorry, I have 5,000 customers. So the moments where I stopped servicing my customers, like if they were the only one, my soul gets crushed because the company shifted, right? And so that's what I want to solve today, right? So that, it's not CS, what I'm trying to solve, is how do we serve as the customer? How do we give the power, the agency to the customer itself?

I give an example, like, I like to have personal relationships with the business that I work. Ah, yes, this very much, yeah. Right? Like, for example, I have a guy that in the winter I have to store a car or two, right? And so I can text, he has a warehouse, he saved his space for me, and I can text him. he lives down the street, and he can come and pick me up, we can take the car drop off, and he's super flexible to me, to my needs, right? I'm in charge of the relationship.

Yes. I go to my barber. I just book two haircuts on Saturday for my son and I. My second son says, can I come? And I say, you text, Mel, and you go and say, like, see if he can squeeze us in. So he texts, mails, okay, I'll squeeze you in in the two slots, right? And so I like that experience. can we have that experience with like a service now today?

Can you text? You know, it's like we want, I want to be able to provide businesses of the future, the ability to put the customer in charge. Yes, at scale too. At scale with hundreds of thousands of customers. That's the essence. Yeah, I was going to say, this is actually something I believe very deeply about venture, which is like I do not want a bunch of, could be conceivably very talented people,

between me and the founders that I owe support and partnership and work to. Because I don't have any of the contacts. I don't care as much as I do. And I think it's not a great experience when there's like four layers of people who are passing around some task for a founder. But it does mean venture doesn't scale if you're doing it in a particular way. And I guess that's true of many customer relationships with like an owner of the product. I think you and I are alike, right?

I want to disrupt how startups work. I want to break the status quo, right? Which is what you just said. You don't want four layers of people between you and the founders, right? And so the same is with me. At agency, there's only going to be one email address. Elias at agency. That Inc.

It's very dangerous, man. Everyone will email. I will always take care of all the customers, right? That's the company of the future. because at DRIF, I had 800 people, like 250 in the customer organization. That cannot happen again because it wasn't working, right? That number of people didn't necessarily solve the problem, right? And that's something that every other company is struggling with.

And so I want to maintain that relationship. And so the only solution out of this is by leveraging AI. I want to talk a little bit about how you think about company building now, both in the era of AI and then also like fifth time around. You don't want 800 people. You say like, I think I can get to a billion, you know, is that value? Is that revenue? I would think revenue. Yeah, that's a better one. Value is too easy. Yeah, okay. The goalpost move. It's a billion in revenue. That's a good goal. It's a billion in revenue. With a hundred people. How do you not

hire the other 700 people? We have to question everything. I'm a big fan of Elon, right? I think I heard him speak about like, you know, he just thinks about building a rocket and he says, well, how much does it cost to build a rocket, right? And instead of just saying, like, how much the parts cost at the existing marketplaces, right? It's like, well, I just build a part, right? It's like, it's a metal. I think the same way, right, is, I have to produce something

that has high value and that is very rare, right? There's two ways to make a lot of money, right, a billion dollars. You either have something that a lot of small businesses can acquire for free, right, with very little marketing costs, or you sell something that is very, very expensive, right, to enterprise customers. And so I'm picking to choose something to solve very, very, very big problems for big enterprises. Second, I think I want to challenge, like Paul Graham says, do things that don't scale.

I think that that's game over for that statement, right? I think now we only have to do things that scale. I already know all the things that I could do before that not scaling. I could always throw our body at something and be like, okay, let's, you go do that. I think you said it in a tweet recently, right? and an X, and you're like, can we just hire an intern, right? And then I think Pranav or somebody says, well, we can throw a devon at it, right? It's like, that's the thinking that we need to have, right?

Do not do things that don't scale. I already know that they will work for a year and then they're going to break in a year. I now have to solve the things right from the first place, right? What are the fundamentals, right? Customer intelligence, you know, customer communication, understand the problem deeply, get pricing right, get pricing better, build the right relationships with the right customers, who's going to take you there, right? You know, I'm building a company at a much faster speed.

It's only been a few months, but I'm already much further ahead than drift was almost three years in, right, when he comes to understanding my passion, what I'm trying to solve, how big the problem is, the market, my position as a company, the team, the team, the engineering organization, the infrastructure, the branding, the marketing. I'm moving at a speed that I just never felt before. And it just feels natural, partly because of AI and partly because of my experience. What advice do you have for people who, I think, like, one reaction is like, oh, man, if you can get to a billion dollars in revenue with only 100 employees, is like,

I want to be part of the 100 employees, right? Like, what makes those people valuable and special in their way? work, like, how do I end up in that last hundred folks? You're asking for something very deep right here, but I'm going to say as much as I can because I really don't care anymore. Let's go deep. Look, as a founder, I see founders all the time doing this. By the way, I was part of three incubators this time around.

I've never been part of an incubator until now, until my fifth time. And I got to spend a lot of time with a lot of first-time founders. And it was amazing because he gave me a little refresher of what I was like, you know, the first time around. And so I see a lot of founders of keep making so many mistakes in hiring. I've made them all too, right? Too many times you get excited. You meet somebody and they say the things like, I want to work for you. I'm going to make your company get to a billion dollars in revenue, you know, 25% faster than you think. They stay all this right things and they're available and they give these pitches. And I no longer buy any of those things. I've hired every executive,

every C-level suite that you can think of multiple times. I have hired from the best companies. I have hired people that went from zero to 200 million in revenue in four years. I've hired them all. I know them all. I'm much more disciplined in my interview process. And no one joins agency until they have work for me as a contract. I'm going to put you to work.

And I am going to be real exactly how I am with everybody in the company today. And if you don't deliver at a level that is world-class in that one or two-week time frame, there will be no room for you, right? And I'm focusing mostly on engineers right now. Everybody has to have a clear role of what they do. It's very difficult to be part of this 100, right? What's on the other side then? What do you think is going to be the hardest thing about building agency?

Building agency is always, always about product. is always about change management. Everybody thinks that, oh, I just put an LLM on it, generate some stuff, summarizes a meeting, and, bab, bah, here's the next CRM, here's the next customer platform, right? No, no, no, no. It's like being in the weeds with a cut,

building the product is the hardest part. LLMs are solving one tiny part of the problem. We still have to build the products that solve as specifically. problem, the solution, right? If everything was just passing it to the LLM and it does it. But we have to transform organizations from very large organizations struggling. It's a lot of chaos and in the customer and the post sales organizations that we need somebody to go in and understand them, talk to them, nurture them, see what's broken and slowly transition into the future,

right? It's not something I've talked to CEOs of public companies and no CEO. comes to me and says, I want to change everything right now. Just throw it away. Let's swap it. And so most companies don't understand that change management is the hardest part to get through. Building a product that people can use, gaining trust from an LLM. I mean, people are, people like, it's okay if you get the summary of a meeting from an LLM. Like, nobody cares, right? It's like, whatever, it's great. Look at this. I can read faster.

But to trust the output of an LLM to send an email. to a million dollar customer. It's like, you know, when you present that to a CSM, they're like looking at it and they're like, they're questioning every word. You have no idea. They're questioning every word in that email. Yet when they're writing their emails,

they're not questioning their words, right? And they're just as bad. And so it's like, we have a long road ahead for this. This is a problem that there's not one solution to it, right? This is not one LLM or one foundation Alam is going to be able to fix. I don't know.

AGI will solve everything, but whatever. I don't, it's that. I'm not worried about that. I'm really worried about what is it that I want to build as I listen to the customers, right? To my customers. And how do I take them where I want to take them, right? And then do they want to come along in the journey with me? I want to take the last few minutes and talk a little bit just about the software industry.

Because, you know, for somebody who, been building software for like 30 years, you're pretty anti-software now, like, as somebody who's quite dismissive of like, oh, I've just been putting shit in databases and taking it back out. Like, explain yourself. Like, what, you know, why do you, why do you think that's so irrelevant? People are like, oh, my God, like, AGI is going to, you know, enslave humanity. What's going to happen, right? I'm telling people, like, they're enslaved today. We make fun of it. But like, who wants to be a sales rep, you know, that after you finish a meeting goes and puts like a task to remind you, to remind the customer to do something else to put. I mean, like, it's just like, I just think it's utterly ridiculous that we just, like the emperor has no clothes.

Nobody wants to say anything about this. Our software is shit. Everything is ridiculous. I mean, there's no good software out there. Uber is great, right? Because you need to go somewhere. You call it, shows up. And it takes you there, right?

That's good. Fantastic. That's the one software product, Elias likes. Okay, Uber. I just want people to realize that why do you want to enter all this information and all the systems and if it doesn't do anything for you? For example, at agency, should I install a CRM? Should I install Salesforce?

Why would I buy Salesforce is a monstrosity of a software that I would have to hire somebody else to go configure for it? And then I don't know how to use it. And then that person that is going to configure it for me, do you think that person knows how to do business better than I do or knows my customers? They don't. And so they're going to tell me some, you know,

antiquated way of working with my customers and what to do. And then I have to hire people to put stuff into the software. And then I have to have people to manage those people to monitor what they put into that software. At what point did we talk to the customer? I mean, like, I'm telling CEOs, you should just throw away your instance of Salesforce. Just throw it in the trash.

Like, it's like, what is it doing for you? We're just so busy configuring it and sending data to it. And we don't ever use the data that is in it. And so software has to transform to do things for me without me even asking. Like, we are in the era where, like, hey, Elias, you're going to speak in San Francisco. There's a calendar of a jet blue fly. on my entry under my calendar, it should read it, let me look at everybody that always ask you to grab coffee in San Francisco and send them a message, I'm in town, and I'm in town from

this day to this day, and lock out, you know, two hour slots every day and say, if you want, meet me here, I'm going to be at this coffee shop and let's catch up. Prioritized out my customers that I'm trying to chase and send them two weeks in advance messages that are repeating every three days, right? Why can we have software that does that? That sounds useful, yeah. That sounds useful, right? That's the level that we want. Like, you know, is that when you can text someone, it just gets done.

Like, that's wealth, right? And so we want software to make us feel wealthy, right? Yes, I don't, I do not feel wealthy from my software today. Exactly, right? How long does it take for this to happen? Like, for this disruption of, you know, software that enslaves us to free us. And does anybody get to stick around? Like, does anybody from the old world of the databases holding our data and creating these workflows get to sustain?

I think there's a lot of infrastructure that is needed, right? So there's good news for that, right? We're going to need places to store this information. We need the internet, right? I'm talking at the solution level, the old wrappers of databases are going to be dead. Like, there's just no way they survive unless they adapt quickly. What we're missing is more people like us at agency where we're thinking fundamentally from first principles

and saying, let's build this new type of software. That's the word that we need to spread, right? It's stop, like, I see all this new CRMs coming up, right? And they're like, it's still the same views, the same tables, except they have like AI computed table. We need to fundamentally think software that is almost invisible, right? I think that we need more people thinking this way

and not just like, oh, let me just improve, you know, workday, let me just improve, you know, Augusto or this and just we need to be thinking how do we really solve the customer problems from a different perspective. Laciness, don't bother them, don't tell them what are you going to do, just do it, get verification, right? As for suggestions, learn with the customer, how they're like, what they like, and create that personalized experience. We need more people to do that. Once we have examples of that software, which is what we're doing at agency. I think more people are going to see and say, okay, that's where we should go. And because of the LLMs, a lot of people are going to be able to build things much faster. And then the big

rebuilding will begin. I hope that happens. I hope I get to be part of some of these companies. You know, what you said that really speaks to me, I work very hard. I have some talents. I'm a very disorganized person. And I felt bad about this for like most of the last two decades. But I think if you were more ambitious about the expectations you'd had for software, you might say, like, why am I filing shit anyway and labeling shit anyway? And, like, trying to force all of this into some project management or database framework or whatever for coordination purposes. Like, it feels like we should be able to accept humans as they are. Because I'm like, all right, I am not that operationally disciplined, but there are a lot of people who don't want to be organizing their data all the time,

just like me. And so maybe that'll be the better way. I think you have incredible strengths, right? To be where you are, to have achieved what you have achieved, is because you're special and talented, right? The good thing is that you don't need organizational skills as a necessity, you know, it's not a need for you to become more successful than you are today, right? The good thing is that that's exactly what AI in the computer.

should solve for you, I hate doing, I challenge everything. If I don't need to do something, like for example, when I first came here, I'm terrible at writing in English, right? And then every time before Chad GPT came out, I would be like, maybe I should learn that skill. I mean, you know, you know that theory about the buckets that there's like five levels of something mediocre and then really bad and then really good, that it takes like 10 years to shift buckets, you know, to go from average to, like, you know, top. And so it says focus on the things where you only have to move one level up to be better. Don't try to go from the one that you're the poorest to the top, right?

And so wherever you're not good at, just let it be. And the beauty is organization like, AI should tell you everything, my questions, what to ask me, prepare context, and you shouldn't lift the finger. But you can do your magic interviewing people. and that's AI can do that, you know, because it's you. Okay, just to be clear, I came up with the questions this time, but in the future, I, you know, I very much look forward to being told what to do with the prep the minute before and actually setting my avatar, it's going to be great.

And then you and I can just get a beer and they'll be it. Exactly. But that's what matters. It's the relationship, right? It's the friendship. I think this is great. I think this is good. Okay.

One shot. We one shot at the whole podcast. Thank you for the conversation. I know this. That's great. Find us on Twitter at No Pryors Pod. Subscribe to our YouTube channel if you want to see our faces, follow the show on Apple Podcasts, Spotify, or wherever you listen.

That way you get a new episode every week. And sign up for emails or find transcripts for every episode at no dash priors.com.