I'm here with Michael Seibel. He was previous president of Y Combinators at the title and group partner. He was our group partner back in YC Winter 21 when we started the first iteration of our company, which was like a AI note taker. Michael was with us every step of the way. Every time we went into a office hours with him, he would slap us around a little bit and then we would pivot.

You're not pivoting anymore?

No, we stopped meeting with you.

Maybe we can get a pivot right now.

So it was actually his last slapping that made Vapi around two years ago. So we were working on that AI note taker thing I mentioned and Michael generously with his time spent 11 p.m. with us one night for about two hours. I was on a zoom call with my three co-founders and then Michael was just talking for another two hours. And he was like, what are you doing with your lives? You should be doing more. You could move to San Francisco, give everything up, change your location, and just start from zero.

Yes. Become monks.

Yes, exactly. And then within two weeks, we did. We moved to San Francisco.

Most people don't listen.

Yeah. That's true. You create a lot of big changes, but when you follow them, they work. They work.

Today, I want to talk about, since Michael has seen many platform shifts in the past through the many startups that he's advised and his time at Twitch, obviously, I would love to talk about the platform shift that's happening right now with Voice and how you see and can put this in perspective given what you've seen. So back in 2021, when we were working on that AI note-taker, you said generative-smemmemorative?

Yes.

I'm half joking, but seriously, when did you first realize that something transformative was happening and when was your mind changed?

So I think that YC started funding AI companies around that time. It was really funny because none of them worked at all. None of them did any part of the thing they said they could do. So it'd be like, oh, someone take notes, and then the notes are all wrong. Or like, oh, we'll answer the phone at the dentist's office for you, and then the call goes horribly. And so what's really cool about my job is I just get to see, does it work? And I write the pitch. So I know the pitch is a lie. So I get to actually just figure out, does it work? And I would say right around ChatGTP 4 was when people were like, oh, it's live. My thing that I was telling people is now working. And that was when I was like, oh, okay, it's no longer generative-smemmemorative anymore. And so it's not like we weren't funding companies that whole time, but it was definitely funny to watch people struggle with ChatGPT 3 and not really work.

And remember, in our office hours, we were trying to be all big brain and like, oh, what's the vision? You're like, dude, just make an appointment. Just schedule fucking appointments. That is hard enough as it is. He's like, can you book an appointment end to end? Well, kind of not really. Yeah.

Why are we talking about bigger brain things than that? But it's definitely been fun to see now. And of course, there's more to come.

I saw a TikTok recently of someone who built a Rue Goldberg machine where you push something and then someone else falls. And it was a joke about AI agents where it's like you push it. And then the guy comes back and pushes the next thing. He comes back and pushes the next thing. It's like a constant.

When I've seen some of these flowcharts that get created, and I'm like, oh, is this the end state? Like a 200-pod flowchart of death? That's the end state.

That's happy workflows.

I'm sure more is coming.

So early way back when we were doing productivity software, you had to spend time studying the pre-Excel era and the evolution towards Excel. And you're the lesson, the Mr. Miyagi thing that you were trying to do, it was good, it was good lesson. It was that if you look at what happens in the past, it actually pattern matches pretty well to what's happening right now. And what we learned from that in each iteration from MS-DOS to GUI to the internet, every time there's a platform shift, there was a shift and who was the leader at spreadsheets. So does that feel especially similar to what's happening right now with voice and human interface?

It does. One of the things that I've noticed, I fell on this trap and so many YC founders fell on the trap of knowing more about the Series B companies in their space than about the public companies generating billions in revenue. And when I kind of broke out of that mindset, I realized how few companies actually succeed. And I tried to study a little bit more about the ones that did. And around that time, I discovered the acquired podcast, which had been around for years before. And I was like, oh, this is amazing. I actually get to learn about the companies that worked as opposed to my office hours.

I think voice is particularly interesting. I funded a lot of voice companies. I think hilariously the reason why I think voice is really interesting is my mom was great on the phone growing up. My mom loved QVC. My mom was the kind of person who could get on the phone with Comcast and get us free cable. The phone was my mom's lever on the world. When I came to the Valley and I interact with all these engineers, I remember just talking to people who never wanted to be on the phone ever. And I was like, if one phone call could save you three hours of work, shouldn't it? It's like, no, that's failure. The phone calls failure. So in my company, in Twitch and Justin.TV, I was the guy who would pick up the phone sometimes. And seeing what kind of experience you could create with LLMs plus the phone, it was like, oh wow, these are the experiences that I think make something feel special or luxury.

Theoretically, we've all read these Clayton Richardson books about rich people get stuff first. And the middle class gets it. As I become rich, it's weird to see how many humans I get and how much interaction I have with them on the phone. Just the other day, I got a human who messaged me, hey, there's some wine that's really expensive that I think you're going to want. And I was like, great, let's talk on the phone about it really fast. And then I made an obscenely large purchase. And that wouldn't have worked over email. That wouldn't have worked purely. There was a connection there that wouldn't have worked on a website. And so when I see you guys and all these other voice companies, I'm like, you're going to unlock these experiences and change how people interact with the internet. And I think that's what's really cool. We're just talking back there. The idea that searching something on Amazon and then going through pages of shit that you know is probably going to not be what you want, that's not the future. That's the past.

And what do you think? You mentioned all these guys to do things. What is it about that human relationship that makes it to enable this better experience or the ability to make that trust, to make that large purchase?

I think when you're making a big decision, you want to feel like you're interacting with an expert. And I feel like the internet at this point has said, go do your own research. It's like, here's an explosion of options. Go do your own research. If the wrong thing shows up in the mail, you fucked up. And I think that this set of products going to be built with voice AI are going to be completely different. I want to get to know you. I'm not trying to pressure you to know a purchase. I want to understand what would be best for you. If I get you to buy something that sucks, you're never gonna talk to me again. And that's like bad math. Whereas like that's a lot harder to see when someone comes to a website, oh maybe they didn't wanna buy. And it's like no, like they signed up, they talked to you for an hour, they bought one thing and then they left forever. You probably screwed them. So I just think that these, I think the luxury experience, the other thing that I think about is that I think the internet optimizes for fast too often.

I don't know many of y'all but like can you go like to go on vacation? I love to go on vacation. I like to research where I'm going. I like to kind of fantasize would it be more fun to be on a beach or over here do this, do that. And I realized that time is entertainment. I was talking to one of the early folks at Zillow, and they were like, yeah, our whole product is entertainment. How often do you buy in a fucking house on Zillow? But you can dream. So I'll go back to the douchey experiences. I now have a standing phone call with my travel agent every 10 days. Every 10 days. Every 10 days. We're for an hour. We dream about the vacations I want to go on.

For an hour?

For an hour, every 10 days. Yeah. One day you'll have, one day you'll power this. But that's exactly the same as like going on TripAdvisor or reading the blogs or going on TikTok and mindlessly scrolling the travels. I want to spend time doing something versus I want it to be one click. I think there are so many things we want to spend time on that the normal internet does not really help with.

So I'd love to take a zoom now at, I guess, more tactically, what we can do as companies and as founders. I'm sure you've seen a lot of folks chase the next wave and fail. I'm sure you saw .com. I'm sure you saw social.

I didn't see .com. I'm not that old.

As I said it, I was like, I don't know. I was like, where are you?

I saw social.

You saw social?

Yeah. I was there for social.

You probably saw a lot of stuff, like Friendster.

Didn't see Friendster. Too young for Friendster.

Really?

I was a Facebook user.

Oh, Facebook. Got it. Got it. So I guess what did you see during that time that separated the founders who ended up winning and those who didn't?

I'll say, you're asking a tricky question. I would say I would give a different answer in the consumer world. The lessons I learned from social were far more about how people choose to spend their day and whether they're spending their day with you or somewhere else. I think that in the B2B world, there's two lessons that I keep on kind of relearning.

I think the first lesson is that the best B2B companies figure out how to make more money for their customers. I interact with so many YC companies that I'll tell me a version of the story where it's like, we do this thing, and we charge $100,000 for it. And if a company were to build it themselves, they'd have to pay five engineers, and now it costs $1 million. And so look at this ROI. And I think that's the worst argument for a company in the history of companies. That is a bad argument. As opposed to, we build this thing and this company can generate 10 more million dollars in revenue because of it, or 100 million, or a billion. I think that if you look at the public B2B software companies, they have compelling arguments on why using their products makes their customers more money. They can basically say, we take dollars away from your revenue. Like without us, you wouldn't be making small revenue. With us, you make huge revenue. We just take this part. And I think that's hard.

The second bit that I've learned is that you have to care about your customer to make them more money. Because making them more money usually happens after they sign the deal. Like, it's when the implementation goes really well and like six iterations later and you have to understand their business because maybe they don't even know how to use your product really well. So many founders in my mind are like, I just want to give the customer a tool. And if they're an idiot, they have no idea how to use the tool. But if they're smart, they'll use the tool and make a lot of money. And I hope they don't bother me. That doesn't work.

If I were a startup founder in the room, I would just be, my KPI wouldn't be how much revenue I'm making. My KPI would be how much revenue my customers are generating because of my product exists. I think that you'll have no problem monetizing if you make your customers a lot of money.

And then how do you think about on what dimensions companies should compete on and which ones they shouldn't nowadays? Maybe as a startup and then maybe also as an enterprise.

I meet too many founders who say, because there's a unicorn in my space or because this company exists in my space to raise a bunch of money. Well, it's actually really funny. I hear it explained as like a reason to be in the space or a reason to not be in the space. And I think it's largely irrelevant. I think the sad fact is how few companies win. I feel like so few companies win that you really shouldn't be focused on your near peers or the other companies in your space. I think you should just kind of assume that like almost everyone in your space is gonna die. It's like that's the math, but even the well-funded ones. And I think that like if you assume that and then you think how do I create the most value for my customers? I think that you're now thinking about the right question.

As opposed to what we did in our startup where we had three competitors, all of us were failing, and each time one of us launched a feature, we'd be on their website, and we'd be like, that might be the feature that's working. We'd all copy it. So there's a bunch of failing companies copying each other's features. The exact wrong way of doing things. But they'd raise more money than us, so they must know something. Obviously, because fundraising means anything. So I think that for me, I wouldn't be thinking about competition. I also would say going back to studying new trends, just as often the first player wins as they lose. None of you use AOL, none of you use Netscape. And so being discouraged by it because someone seems like they have an insurmountable lead — that's not useful, not useful. Big folks die all the time.

And then as you think about being the smaller company, taking on the incumbent, what is the advantage or moat that the smaller company has?

It's been interesting. I've started doing some work with the San Francisco City government. As you can imagine, they are a large customer of a variety of companies. What I see kind of blows my mind. They spend a lot of money on software. The experience they get is just raw manure, horse crap, shit. It's very justified to not have faith that software is going to help them. Because most of the software that they've been delivered has been delivered by a salesperson trying to get a commission. It doesn't work that well.

I think the superpower of startups is that instead of a salesperson trying to get a commission being in that room, someone who gives a shit can be in the room. And it is so different what happens to a customer when a company cares about their outcomes. To me, that's the superpower of every startup is like, if you can figure out how to get the people in your company to care about the customer more than they care about themselves, their pay, their bullshit, you can take down the grades. But if you can't, if your employees are just like, what's the stock price? What are we doing our next round? I want my vacation. These customers, I hate the customers. They're the ones filling my inbox every day.

I'd love to ask you this question actually from the other perspective. When it comes to the, because we have some folks and engineers and teams from more larger companies, software engineers, and they're seeing these cool hot nimble startups go out and do these crazy, agentic things and they're desperately trying to do these agentic transformations. What's your suggestion to them to stay nimble and keep up?

Doing a startup.

Doing a startup?

I've just been in too many meetings with big companies where the executives goal is to reduce the employees to like inbox/outbox replaceable cogs. The distance between the worker and executive gets so large. For the life of me, I was like, oh yeah, but like, if you're an engineer and you spend nights and weekends just building something cool, it'll bubble up, right? Like, you know, like kind of like, I don't...

It's just like when things get big, it's just...

Yeah.

In your opinion, it's kind of hopeless.

I think when companies get big, counterintuitively the number of brains engaged shrinks massively. So many people are just waiting for their boss to tell them what to do. And their boss is waiting for their boss. And some senior VP who hasn't looked at the code base or talked to a customer, that's the brain. And everyone else is the hands. But in a company of a certain size, things can bubble up. And I don't know when that changes. I don't know. But it's depressing because I've definitely been in board meetings where people are like, we need to squash the initiative because we're not accomplishing our goal. We need to control the machine. And they just need to do exactly what we tell them. And I'm like, oh, so there's only two brains now? 1,000 person got many two brains? So to me, yeah, I don't know what to do in that circumstance. I've been shocked at how people can get punished for being a change engine. That doesn't happen at VAPI, right?

No, of course not.

If you do, call me.

We have 40 brains.

There it is. There it is.

And then I guess as you look at this voice market, specifically the voice companies, the startups that you've spoken with, what do you believe about this market that you think everybody else hasn't really caught on to quite yet?

I'll say something really kind of tactical. I think 10 years from now I'm going to have a lot of voice agents in my contact list and my chat conversations. A lot of my communication stuff will be with agents. That's number one. And then number two, I don't believe in exclusively voice agents. There are times where I want something texted to me. There are times where I want a website to look at. I shouldn't be forced into, well, now I have to give you all of this mass. When you're going to a doctor, you have to fill out this massive form. I would like to type that. I don't necessarily need to say that shit out loud on the train where everyone sees all kinds of diseases that I have.

So to me, what I'm hoping is that I'll have a really great concierge, multimodal experience, and I just think there's so many things I'm gonna do with an agent and not with an app or a website. And I already see that happening in my life, but I'm rich, but I think for everyone it's just gonna be like, oh, I've got a guy for that. I don't need to, there's a guy that helps me with that. I think that that's going to be a really great experience. In some ways, I feel like I'm living in the future, and soon y'all will have anything you're into. You'll have a person for that that you think is working on your behalf when you're not actively doing stuff. And they're always available to you. They're always working on your behalf. They'll ping you when they find something interesting. You can always ping them. I think it's going to be so much better. I think we're going to look at these websites and we're going to be like, why did we ever... I remember when Kayak came out, that was the first real awesome flight online thing, and I was like, you can search, and there's filters, and you can do all these cool things, and I feel like that's going to look like the Stone Ages. Here's the flight you want. Why do we have to do all this bullshit? So I'm waiting for all those experiences to happen, and I think voice unlocks them.

And I'm curious to ask you, because you have ChatGPT. From ChatGPT, they tried to make this app store where you could access all these other services through this one unified agent. How do you think about the future where it's one monopoly agent versus everyone kind of has their own?

We had this conversation, right? Yeah, yeah, yeah. We had this conversation, too. Yeah, like ChatGPT just wins everything.

I would say that's what they want you to think. That makes it easier to stop building stuff. I just think that's really hard. It's really hard to do. And I think that I wouldn't bet that there's just going to be one Uber agent on my phone that's my guy for everything.

It's hard to do because of...

Why is it hard to do? I mean, it's hard for me to answer why, but maybe like I would say that I don't use one agent exclusively. Like why is there a difference in experience with the agents right now? And so I guess I just don't believe in the Uber assistant, because I think that like, if the Uber Assistant is 80% as good as the Specialized Assistant, and there's no cost in having the Specialized Assistant, I want that extra 20%. I always want that 20%, especially for important stuff. And I think this is maybe the thing that I've seen is that like, for the super commodity stuff, who cares? At this point for the super commodity stuff, I do a Google search and I get an LLM response and it's like, I'm good. But for anything that needs to be high touch, important decision, I want the best possible. And I think the internet has changed me that like every purchase should be the best, right? Like wire cut, right? Like, yeah, my fucking toilet paper. I want the best toilet paper.

Yeah, you look up the top 10 toilet papers.

Exactly, yeah. And so literally like, I also think for what it's worth that as companies get bigger, they change. And I think that just because you kind of shepherded in this revolution, which OpenAI deserves tons of credit, it doesn't mean you just win everything by default.

Yeah, even though it kind of seems that way. It's like they simultaneously hold a position of like, oh, we love startups and want you to win. But also, we're going to eat all of you.

Well, let's be clear. That was obvious from day one, right? Or maybe I learned that lesson with Facebook platform. It's rare that you see a company that really is incentivized to make all of their partners work. That's rare. So that wasn't surprising to me. If any of you are competing with OpenAI, yes, they're competing with you too. That's good. It's just hard to be good at everything. It's really hard to be good at everything.

If you were to bet on one weird layer of the stack or start up in this space, maybe voice specifically, what is standing out to you right now?

You know, one startup, and I've tweeted about it, that I use more than I ever thought, is this product called Manus. It's not in the voice space, but if they had a voice interface, I would really like it, so you should talk to these guys. I really like deep research reports. I found, even the most trivial questions, if it's effectively free for me to get a ridiculous PhD level report on the subject, I'm like, yeah, I'll take that. And I don't even mind that it takes time. It's like, I'm gonna go have dinner and let's burn some computer cycles on some report that I'm gonna read once and throw away. And so I find that, it's been interesting. I find with my LLM usage that the Google style stuff doesn't retain me. It's the things that I couldn't ever do that retains me. It's the net new that excites me. And I feel all too often that so many of these products are kind of like, we do this better as opposed to just like, no, you weren't going to do that. You weren't going to research. And I started collecting watches. You're not gonna build a hundred page report on the history of this watch thing.

Now you might.

Now you might, right? But that's a perfect example of experience. I would love to talk to that report.

Go on a walk. That's something that wasn't possible.

Yeah. And so once again, with that voice, it almost feels stuck in my pocket versus in my life.

I'm curious, like what you think, like imagining this world five, ten years from now, where we're talking to everything, we have these AI companions in our pockets or as representations of companies. What are the second order effects that come about?

I think products tend to get way better.

Interesting.

Yeah, I think that Amazon thinks they have a personal relationship with you, but like they don't really have a personal relationship with you. At the point where there's an Amazon shopper on your phone that you're talking to, they're not only going to know a lot about you, I think they're going to care about you in a different way. At the point where they're like, yeah, we have Michael's shopper, and Michael's shopper's job is to give Michael an amazing experience, as opposed to Michael, come to the website, log in, good luck, fuck you. So I think that it's going to create a new wave of really high quality, really personalized experiences because instead of having my clicks, they're going to have my conversation. I'm going to divulge so much more information if I get to use voice versus like, oh, he hovered over these jeans. Does that really like them? Or was he on a different screen? They're not getting a high quality feed.

And I think this is what we're seeing across the board is people thinking about this less as like a funnel and more just like, how can I just provide value to these customers? And it's not just value like transactions, but value like advice. Value like help, support, being there throughout.

I think this is the cool. It's gonna be like a coach and an assistant and a researcher. And you're gonna really like it. If you guys build it.

Yes, we're working on it.

OK, good.

So thank you so much, Michael. I'm getting a little hand signal from the back that we're getting close.

There's a huge sign here.

Oh, it says please wrap. Way over time. In large black font.

I think it said that for like 6, 7, 10 minutes.

Oh, I'm so sorry. I was so dialed in. Well, yeah. Thank you so much, Michael, for sharing your perspective on the future.

No worries.

And yeah, a round of applause for Michael. Thank you.