The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch
00:14:58 · Matteo Franceschetti, Harry Stebbings
[00:14:58] S0: Out of home. Everything depends on the scale of the business in that country. So we try to do a campaign, an out of home campaign in The Middle East, but we're not ready yet. So if we were doing an out of home campaign in New York, where we are decently known, it would perform very differently than the same campaign in The Middle East where we just launched a year ago. And so those are the things I'm learning that I didn't know. [00:15:23] S1: When are you big enough to do out of home? Because I think one thing that we've seen in the last, you know, two years, so to speak, is tech companies realizing the importance of brand marketing. Out of home, football shirt sponsorships, you name it, and it supposedly performing well. How big do you have to be to do out of home? [00:15:39] S0: I don't think there is a hard number, but you want to be a brand that is decently well known. And so I make it up. If 100 people see that ad on the street, I don't know, 5% of them knows the brand. And so that is one of the other things that we monitor, that is this aided and unaided brand awareness, to understand how many people know the brand. And if they know the brand, they know what the brand does. [00:16:05] S1: How much did you spend on out of home?
00:16:05 · Harry Stebbings, Matteo Franceschetti
[00:16:05] S1: How much did you spend on out of home? [00:16:07] S0: Nothing, no. But we created a track now that goes around Silicon Valley, and so you can book the track, it comes to your home, and you can try the product live. So this track has a bedroom built in in the back of the track, but now the track also goes to specific events or to specific areas during the weekend where there are many people, and we are seeing pretty good success because there are a lot of people that wanted to try our product, but they couldn't. And now when they see the track, they just get into the track and try the product and bite. [00:16:39] S1: That's a phenomenal channel, and it doesn't... And you're gonna have, like, high conversion rate on that. [00:16:43] S0: Yeah. It's doing pretty well. [00:16:45] S1: That's fascinating. I... I'm I'm also really impressed that TikTok works for you because TikTok is freaking hard. [00:16:50] S0: Yeah. It was now working in the past, then we kept iterating, and it just started working six months ago. [00:16:56] S1: How big is the marketing budget today? [00:16:58] S0: In the hundreds of millions. [00:16:59] S1: In the hundreds of millions?
00:16:59 · Harry Stebbings, Matteo Franceschetti
[00:16:59] S1: In the hundreds of millions? [00:17:00] S0: Yeah. But the story I can tell you is, for example, before we had an email marketing team, now we have zero. Everything is done through AI, and email marketing makes a very large amount for us, and everything is run by AI. How? So what happened is, we were forced to death, meaning we had a team of two, then the person leading the team was leaving. And at that point, my co founder, Alexandra, jumped in and said, Okay, let me see how I can use AI to really make this work. And within three days, she was able to build multiple bots that now run all our email marketing. And so now we have a team of zero, and email marketing makes close to 100,000,000. [00:17:39] S1: I just want to understand because we have an email newsletter, and there's people that do it, and it's still pretty average, think. When you say bots, what do [00:17:47] S0: you build So your for us, obviously there is a calendar of what emails needs to be built and created and sent. This is based also on the history, so we have historical data. And our AI suggests, okay, when to send what email with [00:18:03] S1: And when what's you your AI, it's like your models, it's using Yeah, Grackpod, it's using
00:18:03 · Harry Stebbings, Matteo Franceschetti
[00:18:03] S1: And when what's you your AI, it's like your models, it's using Yeah, Grackpod, it's using [00:18:07] S0: we probably, I mean, we should check, but I would imagine we are using Cloud for that. Then internally, we have probably by now, hundreds, if not thousands of agents that we use internally at Eight Sleep. So now, started measuring what I call the human employees, but then there is also the AI employees. And probably the AI employees are, I mean, without probably, we are probably three, four X bigger if you start including all our AI employees. [00:18:33] S1: Where is it most significantly impacting the company? [00:18:36] S0: Growth, because email is like that. Now, also, all our digital marketing, our team received a report every morning from our AI agents suggesting what changes to do or not to do, and they approve or reject. And so also, our marketing business our growth is run by a few people right now. [00:18:57] S1: I worry that in a world of infinite content supply, discovery becomes impossible and we just get flooded. Do you worry about that too? [00:19:05] S0: Definitely. But at the same time, I think that the personal agents will look for the best product, and so if you are the best product, you will still win. So, yeah, so it will be probably for humans, right, the amount of content will be too much, and so we'll need to think about how we show up to humans where humans go. That is how I think. But then you need a different strategy for the agents to make sure that their agents find you and recommend you.
00:19:05 · Matteo Franceschetti, Harry Stebbings
[00:19:05] S0: Definitely. But at the same time, I think that the personal agents will look for the best product, and so if you are the best product, you will still win. So, yeah, so it will be probably for humans, right, the amount of content will be too much, and so we'll need to think about how we show up to humans where humans go. That is how I think. But then you need a different strategy for the agents to make sure that their agents find you and recommend you. [00:19:30] S1: We said there about kind of the amount of agents that you have. Engineering is one where consistently across the board, everyone's engineering teams are being changed unbelievably. How is AI changing your engineering team at Eight Sleep? What are you using? What are you not using? How do you think about that? [00:19:45] S0: Well, our engineers stopped coding around a year ago. Since then, they didn't code. What they have is hundreds of AI engineers that they code for them, and that is how we can achieve what we are achieving at scale in 35 countries, including China, including Middle East, including Europe, where there are a lot of regulations in all of them, it was a fairly tiny team. What tooling do you use? Really Claude. How much do [00:20:10] S1: you spend on Claude a month? [00:20:11] S0: It keeps growing so fast that I know. Is it [00:20:14] S1: like 5,000,000? [00:20:15] S0: No, it's lower than that, but it's still in the millions.
00:20:15 · Matteo Franceschetti, Harry Stebbings
[00:20:15] S0: No, it's lower than that, but it's still in the millions. [00:20:19] S1: And the thing I'm trying to understand is Mark Benioff said that they spend 300,000,000 a year on Claude, and they have a 6,000,000,000 budget for engineering. 5% of engineering salary is being spent on Claude in particular. Yep. And my question as an investor is will it stay at 5%? In which case, we can justify the $2,000,000,000,000 IPO, but it doesn't have massive expansion from there. Is it a case of the 5% is just the start, baby? We're gonna go to 25%, in which case, material has load up on Anthropix IPO. Or is it the case that actually we're going see the commoditization by open models, most of workflows are going to get taken by them, and that 5% could get down to 1% as we see efficiency? [00:20:57] S0: Yeah. I think there will be two vectors in the opposite direction. The usage will increase, and so the 5% will become 50%, but the cost will go down. And so net net, I think it will go down. I was listening to an interview to Sam Alman recently, and he was saying that in the early days, to pay for one hour of electricity at night, you had to work five hours, just to give you a sense of how expensive electricity was in the early days. And now nobody cares about, I don't even know what's the cost of electricity at night in Miami where I live. And so I think the same thing will happen. [00:21:30] S1: Do you find it a bit scary that intelligence is now free and entirely distributed? [00:21:34] S0: Yeah. It's something I [00:21:36] S1: A choice aside, like.