The AI Agents Helping Home Services Book More Jobs

1. 0:00 – 1:28

   The AI Workforce for the Physical Economy

   1. GTGarry Tan0:00

      [upbeat music] Today, we're sitting down with the founders of Avoca. They're a Winter '23 YC team that, in the last two years, has made it to eight digits in revenue, and actually growing so fast, it's very rare to get to 2X growth in a single 90-day period, and here we are. Welcome, Apurv. Welcome, Tyson.

   2. TCTyson Chen0:26

      Thank you.

   3. ASApurva Shrivastava0:27

      Thank you. Really happy to be here, and excited to be in this podcast.

   4. GTGarry Tan0:30

      So off the bat, obviously, incredible growth. You know, what is Avoca?

   5. ASApurva Shrivastava0:35

      At Avoca, we are building the AI workforce for the physical economy, starting with home services. So the idea is think of, like, the entire world of you have your HVAC businesses, your plumbing, your electrical, all of these service businesses. We think that the entire industry is about to change with how AI is going to be working with humans to get jobs done, and that's what we're building.

   6. GTGarry Tan0:54

      It sounds like one of the things that they really sort of struggle with is, you know, someone has a problem, they actually call up lots of different home services, and, uh, nobody responds.

   7. ASApurva Shrivastava1:04

      That's exactly the starting insight that me and Tyson found, which is that both our backgrounds... So I grew up in Michigan, Tyson grew up in Pennsylvania. Both of our backgrounds, we were working in companies like this, where we saw that actually phone calls is one of the biggest ways that all these businesses get revenue. And so our insight was, if we can build an AI agent that goes and answers the phone, we can be adding value to these customers in the most meaningful way, which is new revenue.

2. 1:28 – 3:25

   Finding the Right Market

   1. GTGarry Tan1:28

      And so there's definitely something special about home services that you figured out. Did you expect to sort of end up in this particular vertical? You know, it sounds like you had some experience there.

   2. TCTyson Chen1:38

      I think home services was definitely, uh, you know, one of, one of the things where both Apurv and I, uh, you know, from our family backgrounds, we had, uh, that experience. So my mom was actually an acupuncturist, uh, one of the first ever acupuncturists in York, Pennsylvania. And growing up, I was her one-man CSR army, so helping her answer a lot of phone calls. So we always realized that this was, uh, an idea that, you know, made a ton of sense in terms of, uh, automating the, the, the role of the receptionist. I think what really, uh, made us discover home services was we actually explored a variety of different industries first. We looked into restaurants, and through kind of talking to a bunch, like door-knocking, we actually scouted 100 restaurants. But home services was the industry that found us in a way, in the sense that in home services, a single phone call can be worth $20,000 to $30,000, whereas in the world of restaurants, a single call is only $30, $40. And 85% of the revenue of a home service business comes from the phone calls, whereas for restaurants, it's only, like, you know, 20, 30%.

   3. GTGarry Tan2:46

      I want to double-click on something that actually I think is a really powerful model broadly. Obviously, AI software starts off as software. In your particular case, there's a public company out there called ServiceTitan that is just the software. But, you know, they're only 1% of wallet spend. I mean, when you're talking about these service providers, you know, they want to do the service. They don't really want to have to manage a call center or deal with all the support needs that, you know, but they are sort of forced to. Can you talk to me about, like, you know, if software in these markets, they only spend 1%, what's the expansion? 'Cause you're not gonna settle for just the 1% given AI

3. 3:25 – 6:59

   Why AI Is Bigger Than SaaS

   1. GTGarry Tan3:25

      today.

   2. ASApurva Shrivastava3:25

      It's exactly right. I mean, for ServiceTitan today, we see spends of, like, 1%, 1.2%, 2% in that area, sometimes even 0.5%. It's all around that number. And the interesting part is that that's generally what people have spent for software. But the idea with what AI is trying to do and why we like to think of ourselves as bringing in the AI workforce is that you are now not only thinking about expenditure as just software expenditure to help your business, you're now thinking about, "Okay, how much labor do I need? What operations do I need to run my business, and how does that work from that?" And so if you actually look at customer service or customer care, you look at inside sales, those units are actually 3% to 5% of total expenditure into a business, whereas software only captures one. Now, if you look at marketing, marketing is such a huge portion of what you need. It's how you get all your leads. That's 7% to 10%. So if you actually just look at some of the different kinds of AI workforces that you can hint to, you're not just trying to see how much of that 1% you can get. You're now hitting into this 15-plus percentage that you can see how much impact you can make into that.

   3. GTGarry Tan4:25

      So that's, I mean, 15 times as big.

   4. TCTyson Chen4:27

      We initially started as a singular wedge, which was the AI CSR, like replacing their, uh, you know, a lot of their call center needs. Uh, but now, you know, over the course of the last six months, we're seeing insane ACV expansion, sometimes on the order of 5 to 10X. And if you look at some of our top, uh, customers and, and you look at, you know, the amount that they spend on their CRM versus the amount they spend on us, we're already seeing in cases where we're able to get to two times that, the CRM. So we're already proving by, uh, actually automating a lot of their, uh, workflow, we're able to be able to capture a much larger percentage of, of the value there.

   5. GTGarry Tan5:09

      At the risk of, like, going VC-speak on this, I... You know, one of the funnier things that, um, you know, we hang out with our friends at, um, DST, for instance, and one of the things they were really excited about is actually exactly this idea. From time immemorial, like, software has been sort of trying to get into more and more of the pie of GDP. When it does, it's getting that 1%, which is not very much. This very small percentage of, like, overall revenue that goes to stuff like this. You know, what's great about tech and being, working in tech is The Street, like, they will actually value what we do at, you know, 10X, 20X next 12 months revenue. Like, that multiple is pretty wild. Like, you see non-tech people look at what tech people get, and they get very jealous about it. Uh, but that's actually exactly what's happening today. It's just matrix math, guys. [laughs] Like, it's software that's built in a slightly, you know, more accelerated way because it also uses AI, but it's actually jumping into, you know- Payroll, expenses, marketing, all of these other parts of, you know, share of wallet. A San Francisco-based startup prior to maybe even 18 months ago never would've dreamed to be able to crack that part of wallet share, and then being able to multiply your wallet share by 15X and have it be software that is valued at, like, 20X next 12 months, that's, like, gonna be the avalanche of value. Like, if you're trying to make something, this is actually the wildest story in all of, like, the world right now.

   6. ASApurva Shrivastava6:37

      I completely agree, and actually, I think it's even more crazy when you think about how early we are in terms of the- how mature these AI tools are. Like, we're still in the very earliest phases where... In fact, actually, if you look at our product roadmap, the most important things we haven't even started working on yet. We're still kind of getting into that, and we've already been able to create so much value. And so, I mean, to your point about it's truly order of magnitudes level of different if we can execute really well.

4. 6:59 – 11:53

   The AI Job Story Nobody Talks About

   1. GTGarry Tan6:59

      I feel like we should address, like, some of the stuff around, you know, someone who just heard what we just talked about will say, like, "Oh, wait. Well, aren't these people's jobs?"

   2. ASApurva Shrivastava7:07

      Mm.

   3. GTGarry Tan7:07

      And the weird thing there is, I mean, the attrition rate on customer support is insane.

   4. TCTyson Chen7:12

      In a normal, um, home service business, there's almost a close to 100% attrition rate over the course of two years. So that means that almost no one from that initial cohort will still be there, and it's because the job is just, it's not really fun, right? You're dealing with pissed-off customers all the time. The working conditions oftentimes aren't great, and you're just overloaded with all these types of calls. What we really view our solution on the AI CSR front is actually helping those folks. By having an AI that's able to take, you know, the 60, 70% of what we call nuisance calls that aren't even revenue sensitive, that actually allows their CSRs to spend more time doing the things that they enjoy and actually, you know, have that unique human advantage for, which is, you know, outbound sales, stuff like that. And so I definitely think that we don't really view it as come in and, and, and replace a lot of folks. It's really to augment and allow them to scale in a much more efficient way.

   5. GTGarry Tan8:12

      Have you seen the movie, uh, Office Space?

   6. TCTyson Chen8:15

      Yeah. Yeah, I've seen it.

   7. GTGarry Tan8:16

      Like, the woman who's like... They, they put, uh, one of the characters next to someone, a, a woman who's in CSR and who's like, "Corporate account speaking. Just a moment." And it's just like, all day, that's, like, what they're doing. [laughs]

   8. TCTyson Chen8:29

      Yeah.

   9. GTGarry Tan8:29

      That's what I think of when, like, maybe a form of white-collar torture, if you really think about it. And then, you know, the opposite of that is, like, I was really taken by your story about, you know, some of the CSRs who work in your customers' operations, like, sort of converted from, like, being tortured in that way.

   10. TCTyson Chen8:47

       Yeah. The, a lot of them, after using Avoca for nine months, now Avoca's taking a lot of the calls that are not as enjoyable, and they don't even really want to take. A lot of them have been promoted to different positions in the organization. Some have been promoted to, you know, dispatch functions. Uh, and there's a new kind of role that's very interestingly emerging, which is actually the person that's in charge of training and overseeing the AI agents. And so some of the top CSRs, the folk- the ones that really see this and, and try to think about what they can do to really help the business, they've embraced this new role of being the person that can really pioneer kind of the team lead for my fellow AI agents.

   11. GTGarry Tan9:29

       That kind of rhymes with, um, you know, sort of the revolution the car industry went through in the '80s, where-

   12. TCTyson Chen9:36

       Mm-hmm

   13. GTGarry Tan9:36

       ... you know, Japanese cars went from, like, kind of these rust buckets that people didn't like to buy to suddenly they were the most reliable, and, you know, everyone preferred them, and people were sort of worried like, "Will people buy anything other than Japanese cars?" And the way they did that was actually empowering the worker. So, you know, a Toyota production system famously had this thing called Kaizen, where, like, the people who have to do the work get empowered to redo the workflow of how it works, and then it transforms this sort of, like, very rote, you know, uh... In a factory, it's like, how much more rote could it be? It turns it into play because, like-

   14. TCTyson Chen10:13

       Yep

   15. GTGarry Tan10:13

       ... you are actively trying to make a system that you have control over better and better. And then if you think about, like, what's happening in terms of transformation of the roles that when Avoca comes into a CSR situation, a lot of people are quitting. [laughs] 0% annual retention means, like, literally if you had 10 people, all 10 people would quit that job on a 12-month basis, right? You know, if you took a couple of those people, you know, they were gonna quit anyway, but instead, they're wrangling AI agents, and they're learning more about the customer.

   16. TCTyson Chen10:43

       One of our customers, Cold As, a great example of that, their retention rates skyrocketed. Pretty big HVAC business in the Carolinas, and once they started using our, uh, Avoca agents to augment and, and kind of be that first line of defense, they saw attrition rates go through the roof because people are now answering less calls and enjoying that more.

   17. GTGarry Tan11:04

       I mean, I think this is an important story that, you know, people should sort of pay attention to.

   18. ASApurva Shrivastava11:08

       I'd say the biggest thing to even think about, too, is, like, if there was some kind of dynamic where people were losing jobs and they hated this Avoca or hated AI agents, you would see the following in terms of customer, uh, value. You would see that the owners of these businesses love you because maybe you're increasing revenue, and your customer service teams hate you 'cause they're kind of like, "Oh, these, these guys are taking my job."

   19. GTGarry Tan11:27

       Mm-hmm.

   20. ASApurva Shrivastava11:27

       In reality, actually, it's kinda funny. Like, I mean, the owners obviously love us. We're bringing them revenue, but it's also the customer service teams, they are super happy using Avoca, as Avoca has actually been great for their businesses, and they're able to now no longer be, you know, yelled at or be judged by making sure they answer every little call. But now they can go focus on doing more important things in the company, and we actually see that dynamic play out. So definitely at this stage it's, it's super win-win, and we foresee it being even more like that over time.

5. 11:53 – 16:59

   How the Founders Met

   1. GTGarry Tan11:53

      Well, thanks for sharing the good news. I think there are a lot of AI startups out there who will be watching this who are like, "Oh, actually, this is sort of the example for what we should, you know, try to make happen," and it's certainly possible, so thank you for making that happen. Switching gears a little bit, I'd love to hear about how the two of you came together as co-founders. It sounds like both of you were at MIT and you met at a poker competition.

   2. TCTyson Chen12:14

      So this was many years back, and it was kind of a casual, um, poker competition with, uh, just a group of folks. I remember, uh, meeting Apurv. I was getting ready to leave, so I, I, uh, decided to, to go all in with nothing. And then Apurv, uh, looked me dead in the eye. I didn't really even know him at this point. He looked me dead in the eye and he said, "If I fold, will you show me what you have?" And I was like, "Yeah, sure." And then he's like, "Okay, well then I call." [laughs] And then he just won all my money.

   3. GTGarry Tan12:44

      Oh, man.

   4. TCTyson Chen12:44

      Yeah, and I, I remember leaving that day, um, really frustrated.

   5. GTGarry Tan12:48

      Mm-hmm.

   6. TCTyson Chen12:48

      But then, uh, we ended up becoming, you know, super, super close friends, um, lived together, uh, in, uh, in San Francisco and, and, and throughout New York for over five years. Just really got along, uh, had a lot of similar passions around, uh, AI in particular, and we, we worked on a lot of different AI projects together. Uh, even, even living together, we had built this one application. When we were playing basketball one day, we were like, "Hey, who... Like, what is actually our vertical jump?" And so literally the next day, we decided to build, uh, a quick AI application where you take your phone and then you can video yourself jumping, and it instantly tell you the vertical.

   7. GTGarry Tan13:26

      Nice.

   8. TCTyson Chen13:27

      And so we did a lot of these things together and, and then ultimately, um, decided to build something.

   9. GTGarry Tan13:32

      I mean, both of you guys kinda cut your teeth on building pretty impactful things for, you know, Nuro, for Retool. Do you wanna talk about those things at all?

   10. ASApurva Shrivastava13:40

       Honestly, the, there was a lot of really great work that was happening there. I mean, I can, I can touch on Retool for example. I mean, I think Retool, when I joined the company, they were very much, like, under 50 people, like, blitz scaling. And I think all the way from, you know, figuring out what does our motions look like on product, and thinking about forward-deployed engineering and all this stuff, it's like I was one of the first people to kinda be hired into forward-deployed engineering, growing out that team, figuring out how do we see success for customers, how do we scale those operations? Like, all sorts of different kind of work like that. So I think it was pretty great. And honestly, I'd say one of the biggest and most prideful things, and I think we've kind of kept these values into Avoca as well, is that we hired an insanely strong and talented forward-deployed engineering. I think some of the people there have gone on to be other founders today. You have series A, series B founders doing really well. It just shows that when you hire a great team, and you can execute really well, and these people will do great things in general. And we s- we see that as well in Avoca, where we think the people that are working here at Avoca have created something incredible, and they will themselves be future incredible founders one day.

   11. GTGarry Tan14:43

       Yeah, the FDE model is pretty awesome. My high school and college buddy I think was basically the first FDE in practice. Stefan Cohen, one of the co-founders of Palantir.

   12. ASApurva Shrivastava14:52

       Nice.

   13. GTGarry Tan14:52

       Uh, it was before the title was created-

   14. ASApurva Shrivastava14:55

       Yeah

   15. GTGarry Tan14:55

       ... but that's basically what Stefan did. And then, uh, our buddy Shyam, who now runs all kinds of things at, uh, Palantir still, he sort of formalized it into, like, a training program and-

   16. ASApurva Shrivastava15:05

       It, it makes sense because I think for every business, the most important thing is customer value and customer obsession. So if you can get someone who can understand the product and be technical on the engineering side, and also interface with the customer, they are potentially some of the most important people in any business because they're figuring out how does this product actually drive you to success? Truly, it's an incredible model, and I think it's worked really well. You obviously need to get the right people and it, it's, it's just amazing.

   17. TCTyson Chen15:29

       For Nuro, I think one of the most... I mean, I worked on a lot of projects there for, uh, around three years. Some of the most interesting things at Nuro that, uh, probably impacted, uh, kinda my thinking and our thinking of, of Avoca was some of the work I did with, uh, uh, the human-in-the-loop stuff. And so the team I was on, it, it was called, uh, kinda Supply. We built a lot of the internal tools powering, uh, kind of the operations of, of Nuro. One of the things was, uh, this program called Guardian. When the autonomous vehicle was, uh, in trouble, allowing a remote operator to kind of remote in and save the day. That was something that we actually took kind of a similar lesson from and incorporated it into the Avoca product as well.

   18. GTGarry Tan16:13

       So what were the tactical things there? It's like trying to figure out, you know, when should a human get involved?

   19. TCTyson Chen16:18

       The most important things was understanding, like, when and also, uh, the providing the context. Understanding when the autonomous vehicle is in trouble, and ideally anticipating a few seconds beforehand, and then being able to then give the human operator that's going to step in a bit of a preview into, like, the context of what's happening. You're not just immediately thrown into a situation where you have to-

   20. GTGarry Tan16:42

       Yeah

   21. TCTyson Chen16:42

       ... kinda save the vehicle.

   22. GTGarry Tan16:43

       That's funny. I mean, I guess in AI, people are always, uh, doing context engineering for LLMs, but you're doing context engineering for humans-in-the-loop then.

   23. TCTyson Chen16:51

       Yeah, exactly. [laughs] There's so many parallels between the world of autonomous vehicles and, uh, you know, AI Agents we're building at Avoca.

6. 16:59 – 21:47

   The Pivot

   1. GTGarry Tan16:59

      Absolutely. I think you got into YC twice. Uh, the first time you couldn't come over yet, and then second time you guys did. Talk to me about your time in YC.

   2. ASApurva Shrivastava17:08

      So I'd say when starting the company, I would say the most important thing was me and Tyson, we were roommates for so long, we wanted to work together, and we were, we had very similar ambitions and all that. I think in the early days, we were kind of exploring a few different ideas, hacking on things. We, we always used to hack on a variety of different projects. I mean, all sorts of different things. But finally, I think the, the place where we got really excited, where we thought actually YC would be great, was going to be, um, when we were thinking about the fact that... This is actually pre-ChatGPT, but after the model GPT-3 eventually was out, and I think people started talking about it but the hype wasn't fully there. We started kind of realizing, wait, these AI models kind of sound like a human. It's not perfect, and hallucinates like crazy, but it does sound like a human. And for me and Tyson, I mean, from both of our backgrounds, we kind of, you know, growing up in Michigan and Pennsylvania, it's like we literally did the job of what these agents would often have to do or what these agents could do, where we just had to just, you know, talk to these customers, close customers, et cetera. We really had this, like, kind of hunch where we were like, "Okay, these things aren't yet able to speak, but once they can speak, the entire what we call, like, the physical economy, the entire all these service-based businesses will have an insane unlock because of the fact that now you have someone that can answer your phones, that can go reach out to customers, do all this thing." And so I think we had a pretty early hunch. That's when I think me and Tyson were kind of like, "Let's do this." And especially when it comes to YC, I mean, I think without YC we would not have been successful. I think it's just very clear. In fact, when we first got into YC, me and Tyson were kind of like, "You know what, do we need YC? We don't really know what we're working on yet. Like, should we even do it?" And I think we actually decided not to do it, 'cause we were like, "You know what, still working at a company, that's not gonna wrap up by then." Um, let's just, you know, figure it out ourselves. There's some VCs who are reaching out, you know, wanting to invest. We don't necessarily need YC. But that's actually when we were actually doing the actual exploration, is when we realized that there are so many things about starting a company that genuinely we would not have learned. Obviously, the two biggest things I can think of are customer obsession, which is just being maniacally focused on your customer, and focused on momentum, which is just constantly be pushing yourself. And I think YC gets you into that environment. And so for us, frankly, there was many places where me and Tyson were thinking like, "Okay, you know, we have all these job offers. Do we... You know, we've tried for a bit. Do we go back and, like, maybe try again in a few years, or do we keep executing?" And the fact that we kept executing and did YC is probably the only reason that Avoca is able to be how big it is today.

   3. GTGarry Tan19:30

      Yeah. I feel like, you know, tangibly the two things would be, like, obviously you working with a partner is cool. I mean, I'm working with 25 companies right now, and I'm just always reminded of, like, when you're in the maze trying to, like, find your way to product-market fit in the idea maze, you're like, "Do we turn around? Do we, like, explore this other part of the maze?" And then-

   4. ASApurva Shrivastava19:48

      Yeah

   5. GTGarry Tan19:48

      ... you know, being, having someone next to you who's like, "Oh, no, no, keep going." Like, you don't know that yet or, you know... Or it's like, "Oh yeah, it's time to not do this anymore." Like, that, you know... Was that a part of, like, what you experienced?

   6. TCTyson Chen20:01

      Definitely. I think there were so many instances where Apurva and I, we, you know, we're roommates, uh, and we would get so excited. There would be so many of these kind of like, "Man, like, this is it. We're, like, we're onto something huge. Like we, we... You know, this is gonna be huge." Literally, like [laughs] two days later as we, like, discover more about that niche or, like, talk to more customers, then we fade. It'd be really disheartening, 'cause it feels like you're pounding your head against this wall and you're working, like, you know, we're basically working 24/7 every single weekend. I think it was really, uh, nice to be in that environment where there's other people doing the same thing. Every time we had the opportunity to speak with our, with our partner at the time, Dalton, it was really elucidating. I think having him kind of ground us and, uh, it's just like every time we left, uh, speaking with Dalton it was like, "Okay, th- this is what we have to do. These are the actual core tenets to focus on." As Apurva said, I think, like, speed is, is everything. Like, I think there's many areas where we could've probably built a decent business if we had stuck to some of these other verticals like restaurants that we had found some initial traction, dental. But it was really the relentless idea of like, "Okay, this is good, but how big can it be?" Like, when we switched into home services, we had 100 restaurants that were using us. They were liking us decently. We even had some national restaurant chains. But after we saw the opportunity of home services, we just realized that even though we had 100 of these restaurant customers, that wasn't enough. Like, that's, that's not gonna become a generational company. And so having the speed to kind of go quickly and then also be able to say no to, you know, good markets and good ideas to pursue even, you know, bigger ones, I think that really was what allowed us to succeed.

7. 21:47 – 25:31

   Customer Love Beats Market Size

   1. ASApurva Shrivastava21:47

      Maybe one quick thing to add to that is I think the, the thing that I really like to think about when it, when it comes to this is about customer value again. I will say in the early days, you know, I mean, we worked at Retool and Nero. A lot of our friends were VCs knocking on our doors. It's like whenever we would kind of talk to a VC even casually and, and we'd kind of be like, "Oh yeah, here's what we're exploring," the thing that every VC person would tell us that YC wouldn't, that I think YC was correct on, is that every venture capitalist is always thinking about market.

   2. TCTyson Chen22:12

      Mm-hmm.

   3. ASApurva Shrivastava22:12

      Whereas every YC partner that we would [laughs] ask advice for is always thinking of, "Can you sell it? Will they pay for it? How much do they actually like it?" 'Cause I think selling and paying is a proxy for value. And so for us actually what was interesting was that, you know, even in home services it's like in the early days it's not like VCs were always, like, super excited.

   4. TCTyson Chen22:29

      Early days they, they hated it.

   5. ASApurva Shrivastava22:31

      But, but the thing that I think me and Tyson saw, and that's actually candidly because of a lot of learnings we got from, you know, places like YC, is that actually the amount that these home service customers love our product, it's producing 80% of revenue for their businesses are coming through phone calls. So the amount that our product is helpful is just so high that actually let's not look at classic things about, "Oh, what is this market? Should we go work on it?" That's kind of like, that's the VC's job to go do that. Our job is to go follow where there's customer love, where people are gonna pay us more, and that's gonna be the more exciting opportunity for us. And yeah, frankly, I think we would have... Me and Tyson could very well have been stuck in this maze of, like, trying to play the game of VC of what idea to work on as opposed to just following the customers if it wasn't for a place like YC.

   6. TCTyson Chen23:10

      And I think one other YC principle that really stuck out in the early days was understanding the way that you can add value is really directly correlated to the customer pain. And I think in the early days, like, I remember one of our earliest home service customers, ResQair, they had this pain that was so great. When we spoke to them, that's all they could tell us about, which is how much they hated their answering service, and yet they still paid $3,000 a month for this answering service. It didn't integrate into their CRM. It was... It had long hold times, it was inconsistent, and yet they still paid it. And the YC principle that it always takes me back to is, like, if the pain is big enough, the customer will try to find a solution to it even if it's not perfect. And so that's when it was like, that kind of was a bit of a light bulb moment. It was like, "Wait a minute. These guys are spending this much money?" And it's like the [laughs] the, the solution is this bad. Like, there's definitely something-

   7. GTGarry Tan24:06

      Yeah

   8. TCTyson Chen24:06

      ... here to, to kind of tease that, like, tug on that thread.

   9. GTGarry Tan24:09

      Basically all of these things, like, you can read about it, you can watch videos like this and, you know. But when you're actually in the idea maze and you're, like, breaking your brain about stuff, having office hours with someone who's seen so many things basically is like a... It's almost like if a GPT-5 ha- It actually does have this, like, model router that tries to help you figure out, like, "Okay, do I have the answer or do I need to, you know, use medium or high reasoning, or do I need to use pro?" If you're doing it on your own, it's like trying to use, like, GPT-5 Pro for, like, every decision.

   10. ASApurva Shrivastava24:42

       Exactly.

   11. GTGarry Tan24:42

       And you kinda like spin-

   12. ASApurva Shrivastava24:44

       Oh, yeah

   13. GTGarry Tan24:44

       ... overthink, and then it's like when you use GPT-5 Pro for, uh, exactly what happened to you, like we have a customer who's experiencing this extreme pain Like, in office hours when we hear that, you know, I would be like, "That's sick." Like, "Do that." Like, you have, you- now you got something. All you have to do is, like, start pulling on this thread, and you're gonna have this billion-dollar company, which has turned out to be true at this point.

   14. ASApurva Shrivastava25:07

       We're not quite a billion yet-

   15. GTGarry Tan25:08

       Yeah

   16. ASApurva Shrivastava25:08

       ... but we'll get there.

   17. GTGarry Tan25:09

       I mean, we can see the trajectory.

   18. ASApurva Shrivastava25:10

       [laughs]

   19. GTGarry Tan25:10

       It seems like-

   20. TCTyson Chen25:11

       [laughs]

   21. GTGarry Tan25:12

       ... it seems you're writing the stars to me. So you figured out this product. You've got product market fit. Um, it's working. Walk us through. You did Demo Day, you did YC. How much did you raise? Like, you pivoted in the batch and found this amazing thing. Walk us through, like, you're in the idea maze and then to here, you know, especially around the

8. 25:31 – 29:35

   Building an AI Workforce

   1. GTGarry Tan25:31

      product.

   2. ASApurva Shrivastava25:31

      On the product side, basically the day one value was focusing on one problem that's like a hair-on-fire problem these customers have that we can provide the most amount of customer value, and that was building the AI CSR. And so this is essentially someone, like an AI agent that goes in, answers phone calls, books customer, closes revenue, et cetera. And when we realized that that was actually very valuable, we kind of were, was like, our immediate thing was, let's get this into the hands of as many customers as we can, and let's perfect that. And I think when we started seeing that that was going well, that's when me and Tyson kind of realized, we were like, "Okay, wait, this could be onto something quite interesting." That's when we actually focused a bit more on hiring, got a great team and things like that too. But then the next phase of the company started occurring, which is what's happening right now, and I think we're kind of in this phase where we're, essentially the product is significantly deeper than voice. And so that voice, that AI CSR, was the initial wedge. What we are doing now is, like, what we like to call the AI workforce because it's not just that CSR agent. We're doing all sorts of things. We're kind of doing proactive reach-outs. We can reach out via text. We can do targeted marketing to different people. We can have, like, a coach product that goes in and coaches to see how are you doing on call performance, how is your text doing, and things like that. And so now it's like just building each of these agent flows. I would say the biggest reason we're able to do that, and that's what's causing a lot of these, you know, ACV expansions and things like that, is actually because we also have a just an incredible team. And so I will say it's like we're kind of in this, like, complete green area where you have this industry where there's not that many technical people. You have to create an incredible product because now you have this entire AI wave that's about to completely change our industry. And so that's where we were able to kind of have some of the best people working here. And so I think, for example, I'll just highlight a few people because they were also YC people. But there's, like, Bharat, who I think he was YC in 2015. Without him, the company would not be where it's at today. He's, like, one of our product leaders. He kind of goes in and figured out essentially how to kind of help scale some of these, go from the CSR to building the rest of the workflow as well. Led some of that work. There's Dragos, who for that CSR coach, that, like, the coaching product, he's kind of taken a big lead on that, and he was YC 2020. Incredible person there. There's also Ditia, who was actually the same batch as us. He was building, like, an API infra company, and he's just been one of the most incredible engineers, too. And so in fact, actually, we just have this incredible team out, and we are building out essentially the rest of the workforce layer. And so what I think is the most exciting part about Avoca is not even, like, how the product looks today or how much we've grown, but it's how strong of a team we have and how much there is still to grow. I mean, my candid plug would also be if, you know, there's any folks, we love ex-founder people, wanna be founders. We probably have one of the most founder-like early-stage cultures, especially in New York. And so if anyone's excited, I think Avoca's definitely looking.

   3. GTGarry Tan28:11

      I can definitely say, like, working at Palantir very early, uh, jump-started my whole life in a lot of ways. You know, being able to work at the next, uh, you know, future $300 billion company in Avoca, that, I mean, that's gonna be a pretty killer experience.

   4. TCTyson Chen28:26

      And we give a lot of autonomy, too. Like, there's, you know, there's so much to do. We're the leaders in home services, um, but there's, you know, several other verticals that we're looking to expand to. So just the scope of really being able to, to work, uh, here, I think, is, is unparalleled.

   5. GTGarry Tan28:42

      Actually super interesting how you take this sort of CSR function that's kind of a cost center, like cost of doing business, very, very high turnover, terrible job. Like, nobody even really... You know, you know you have to do it, but you don't want to. And then you've sort of switched it, and now it's actually, like, sounds like it's putting more money into customers' hands.

   6. ASApurva Shrivastava29:04

      Core revenue driver.

   7. GTGarry Tan29:05

      So-

   8. ASApurva Shrivastava29:05

      Yeah

   9. GTGarry Tan29:05

      ... so it's not just like cost center, have to do it. Now it's, uh-

   10. ASApurva Shrivastava29:08

       That's actually, when we do sales, I mean, our... For sure, yeah, you'll get some cost, uh, reduction. We've even seen cases where a company that had 100 CSRs is now at 30. Um, and that, and, you know, a lot of that is natural attrition given seasonality, and a lot of it's, like, folks get promoted. The value prop we always lead with is this is gonna be a core revenue driver. This will be putting multiple millions more, um, by, you know, people spend so much on marketing, by being able to capture every lead.

9. 29:35 – 34:12

   Why Customer Obsession Wins

   1. GTGarry Tan29:35

      You were saying you found a customer, and they're starting to pull the product out of you. I feel like there's a teaching moment there where it's like some people might overthink it, or maybe they'd not spend the right amount of time. Like, it sounds like what you did was fly there and sit with the customer-

   2. ASApurva Shrivastava29:50

      Yeah

   3. GTGarry Tan29:51

      ... and, like, literally work out of their, like, office.

   4. ASApurva Shrivastava29:54

      Yeah.

   5. GTGarry Tan29:54

      You know, what was that like?

   6. ASApurva Shrivastava29:55

      Customer obsession is, is I think a core part of, uh, what makes Avoca special and a core part of the DNA that we've been able to, to carry, you know, s- since, since we started. A lot of the best ideas that we have come from understanding customers at a very deep level, and I think we can only get to that point by having that level of obsession, where literally now even, like, basically all of our, uh, major deployments will have several team members fly to that customer for a week. What I like to call it is, like, all these customers, they're, especially the large ones, they're successful, uh, for some reason, and usually it's like they might have found one or two secrets that no one else in the industry has figured out.

   7. GTGarry Tan30:41

      Walk me through, like, the story of it. Can you place me in, like, a time and a place? What did it look like in that CSR room? And-

   8. ASApurva Shrivastava30:48

      Yeah

   9. GTGarry Tan30:48

      ... who did you talk to, and, like- You know, when you came away from that, it was like how did that influence your first products?

   10. TCTyson Chen30:55

       The one that kinda comes to mind is probably our, our first customer, uh, ResQair. Uh, when we first walked in, there were two owners, Josh and Michael. They had known about us because one of our restaurant customers, uh, just like said, "Hey, you guys should talk to these guys. They're buddies of mine. I showed them the- the AI CSR you built for me." Uh, and, uh, basically all these guys could talk about was how bad their answering service was. So they introduced us to the operations manager, the guy by the name of Andreas, and Andreas is this like tall 6'2" Greek guy. He's like, "Hey, this is my CSR room," and he showed me folks. There's like six people, and it just turns out that the room was buzzing, but the issue was there was too many calls 'cause there was a, a heat wave in Dallas in the summer. And so the, the call room was buzzing, and you know, people were trying their best to handle the calls, but it turns out a lot of the calls just weren't able to be taken. "Well, what happens to all these calls? Like, you know, where do they go?" And they're like, "Well, we have this local answering service that we're using to handle all of our overflow and after hours," like weekends, nights, where they don't have any of their CSRs to take care of them. And so then we kind of said, "Well, like, you know, does, does it work well?" And that's when we didn't even have to say anything. They basically continuously were like, "You know, this is actually the number one thing that irritates us. We've literally switched to the fifth answering service in the same year. Five, five different ones." And it's like, he's like, "It's always the same thing. They start out strong, but then quality diminishes." And so understanding that, then we got deep into the workflow and say, "Well, wait a minute. What does this answering service actually do for you? Does it actually book jobs?" They're like, "Well, no. All it does is it waits two minutes on average to answer. It doesn't actually book the job. It just takes a message." And so we're like, "Wait, so ba- basically you're paying $3,000 a month for a glorified voicemail," which also, you know, sometimes even irks the customers because, like, the CSRs they have are, you know, not the friendliest.

   11. GTGarry Tan33:02

       They're sitting in a 95-degree room, and they're hot, and y-

   12. TCTyson Chen33:05

       And those answering services are answering phones for, you know, ResQair, the vet, the librarian. Like, it's not personalized at all. And so yeah, that, that, maybe that's a bit of a, kind of the inner look into what that initial interaction was like.

   13. GTGarry Tan33:20

       And then what was it like, you know, when you went back, made a version that, uh, you know, and you went back and showed it to them? What happened?

   14. TCTyson Chen33:26

       There was these moments where it felt incredible, and I think that's... It was not perfect. You know, in the early days, AI's hallucinating. There's all these sorts of issues about how does it handle this edge case? How does it deal with this workflow, this workflow? But in the cases where it did well, it was incredible. And then it was up to us to go figure out how to, like, really fine-tune that product. And then if you look at a lot of the other agent workflows we've built or things like that from there, it's really just came down to being like this is customer X. Nothing in Avoca is built until customer X or customer Y is using that and they're happy with that. Until that point, it's like there's no fe- you're not feature complete until a customer's happy as one of our tenants, and I think the reason we have that is 'cause we hear our problems from them. It's our job to add as much value as we can to customers, and we hope that that'll show increases in our contract sizes and how much revenue we'll also get,

10. 34:12 – 37:22

    Growing to Eight Figures

    1. TCTyson Chen34:12

       too.

    2. GTGarry Tan34:12

       These maxims are, like, easy to say, very hard to do, and you're doing it.

    3. TCTyson Chen34:15

       Yeah.

    4. GTGarry Tan34:16

       Um, and you know, from here it's sort of resulted in this absolute, like, growth explosion. Like last year when you hit, uh, you know, your first seven figures ARR overall, and then, uh, now you've basically doubled to eight figure-

    5. TCTyson Chen34:32

       Yeah

    6. GTGarry Tan34:32

       ... ARR. Like doubling to get there in the last 90 days is very explosive growth. So how'd you do it?

    7. TCTyson Chen34:40

       It was a combination of things, um, but I think one of the big reasons was it kinda goes back to the old, like, YC axiom around like don't try to get 100 customers that like your product. Try to get three that love it. And so we went, uh, to the lab and, and just, like, had three essentially early design partners that we absolutely poured ev- our heart and soul into making something that they loved. And the interesting thing about home services that it's extremely interconnected. A lot of the early, uh, traction was referrals. You never think about it, but in home services there's this influencer network where folks like a Josh Campbell from ResQair, they are like kinda super connectors.

    8. GTGarry Tan35:25

       Mm.

    9. TCTyson Chen35:25

       Everyone looks to them to say, "What's next?" Like, "What is the next big technology?" And so, you know, I think Josh personally connected us to, uh, something like 10 of our first, like, 30 customers.

    10. GTGarry Tan35:37

        Amazing.

    11. TCTyson Chen35:38

        Then we started getting, uh, a l- lot of, uh, chatter about us on these Facebook groups where all these contractors are from. And then we layered in, um, kind of the conferences, the trade shows. With virtually zero sales and marketing, we were able to get to that first, like, you know, million, two million. From there on it was just continuing to, to kinda scale. The last half of it has really been driven by the enterprise and, and private equity adoption. Because these, you know, P is consolidating this, uh, industry, uh, a ton, we figured out, like, how to, how to get that motion down, how to do enterprise sales, and then once we had that kind of continuous inbound from, from the mid-market, and then we were able to understand and really figure out how to attack the enterprise, that's when things just, like, continued to boom. And again, like, you know, we, we've been growing fast, but we really think this is just the beginning. Like, this market is absolutely massive. The exciting innings are, are yet to come.

    12. GTGarry Tan36:35

        Yeah. Basically do unscalable things, uh, to get to the first million dollars a year in revenue.

    13. TCTyson Chen36:40

        Yep.

    14. GTGarry Tan36:40

        And then, uh, the trick is scale it after that. [laughs]

    15. TCTyson Chen36:43

        Yes, exactly. I think, yeah, a lot of unscalable things we did with the Facebook groups and stuff. I think we would personally ask our customers that were, like, really happy anytime there was a AI-focused thread-

    16. ASApurva Shrivastava36:53

        To kind of talk about us and, you know, that led to a lot of things. Yeah, I think scaling, yeah, we ended up starting to scale sales quite heavily. But the key is, I think the thing that we did right is, is scaling before you truly find PMF. And I think that's one of the biggest mistakes that can kill your business. We literally got to around a million, maybe over a million before, uh, we brought on, uh, kind of our f- our first actual sales team. And then it was like, "Hey, we figured out how to do it," and then we

11. 37:22 – 38:54

    The Vision Beyond Home Services

    1. ASApurva Shrivastava37:22

       scaled it after that.

    2. GTGarry Tan37:22

       Yeah, that's the right way to do it. Like, o- often people try to scale their sales team having not even done the sales themselves. Yeah, [laughs] like, without founder-led sales, you can't do it, and so-

    3. ASApurva Shrivastava37:31

       Totally.

    4. GTGarry Tan37:32

       If you're right about going from 1% wallet spend to 15% wallet spend in this, you know, very, very large home services market, uh, you know, what does Avoca look like, you know, next year and five years from now?

    5. ASApurva Shrivastava37:47

       It's almost like there's this concept of system of record that, you know, everyone talks about when it comes to CRMs and things like that. There's almost, like, going to be this... What Avoca will represent to these industries is almost a system of intelligence or system of action, where your core workflows, how you close your customers, how you do your marketing, how you do all of your core operations work is going to be centered around Avoca, and it'll be the biggest platform that we think these people will use. Be significantly more valuable to these customers than I think even a ServiceTitan could be. But I think, yeah, ServiceTitan has its place, too, and it's a great company, but I think Avoca will have a great place as well. And I think it's going to be very big, and I think the most exciting thing also, by the way, is that home services, how that's gonna be done will also be very similar to maybe adjacent industries or other industries, too. And so when we think of, like, Avoca, we're already hitting a market size that's just so massive. And the fact that it's just the beginning with that market is just so exciting for us to hear, where we think that if we execute flawlessly and do really well, the opportunity for this business is... Being a unicorn is not at all the goal. The goal is to see how great we can be. It's like, can never say anything about valuations and pricing, but quite big.

12. 38:54 – 39:04

    Building a Generational Company

    1. GTGarry Tan38:54

       Well, that's, uh, all we have time for today, but we'll see you guys next time. [outro music]

Episode duration: 39:05