Mercedes Bent invests in consumer AI startups despite VCs pouring 70-90% of dollars into B2B. She tells technical founders to stop avoiding sales and talk to customers instead. She reads sci-fi to imagine what AI interfaces could become beyond the chat boxes everyone copies now.
Chapters
Dalton and Michael Seibel even called consumer ideas like tar pit. But, but YC's latest batch is like over 80%, um, enterprise. What would you say to founders like me who internalize that advice so much?
Yeah, I don't think they're entirely wrong if what the goal is, is to get the highest percentage of companies to the next stage. It's easier to get to the next stage as a B2B company. In consumer land, you get a customer and you have one day or less, you have minutes to show them the value of the product or they're never coming back. I love investing in consumer because I think it's so hard. It's like the Olympics. Every day to just get a basic product off the ground that customers want to return to the next day.
I just thought it's less competitive with B2B because less people have access to that population of users.
You have to— I somewhat, but if you're looking at Silicon Valley venture-backed startups, the vast majority of companies that are funded by VCs are enterprise and B2B companies, something like 70 or 80, or I think it might even be as high as 90%. I saw a report that said only 6% of VC funding went to consumer startups in 2024. They might not know exactly what the world's gonna look like in 15 years, but they can have a hypothesis and be— they should be able to infect the listener on the other side of the table from them with their energy and passion and enthusiasm that that vision is a possibility. I run this Sci-Fi Tech Club on the side, for example, where every month we read science fiction books or movies. And in part, I do that because I always want a portion of my brain to be living in the future so that when a founder comes and tells me about this vision they have, I can see some parallel somewhere to one of the many, many dreams I've read or had about the future. ABC should not be the people creating the future. You have to be very careful to stay on your side and not get too excited about what you think the founder is saying, but listen intently. But we do have to recognize when other people, the founders, are actually telling us about a vision of the future that is very credibly likely to come true.
So, this sounds like an exercise you're doing, like practice your brain. For imagination.
Constantly. Constantly. Yeah, I never thought about it that way, but you're right. I need to do futurist exercises to stay pliable in my mind. It feels like so much common sense sometimes. It's a little hard for me to understand when people don't, are not like understanding no one's gonna buy that product, but—
It's not so common sense, believe me. I talk with a lot of technical founders and almost 90% told me that the hardest part is go-to-market for them. They had, they have no idea and they don't wanna do sales.
I, I, I love go-to-market. I think it's actually one of the more straightforward parts. My dad's an engineer. He taught me coding when I was in high school. I, you know, originally started out going to college as a computer science grad, so I get it. I get where they're coming from, but at the same time, it's like, just go talk to the customers.
Just, just, just go. Just what the heck, just go.
Just go. I had an event the other week that I hosted at Stanford with the head of product at Anthropic and with the CEO of Luma AI and the chief product officer of LinkedIn, like amazing speakers that were discussing with a bunch of us engineers and researchers about what they're building. And it was like the future of product management was the topic. And they were saying, "We still can't exactly get all of the engineers to talk to the customer, even though now that AI is writing a lot of the code on their behalf, the engineers should have more free time to go do some of the product-y type of stuff." And it's like, did they want to? I don't know.
As a builder myself, one trend I am seeing a lot is everything has a chat interface.
Yeah.
What do you think about that?
I'm so tired of that. Yeah. I think this is a function of not having new interfaces and not having new platforms yet, and also just being very early in the cycle. If you think about the internet and when we first got browsers, a lot of what people put on those webpages looked just like newspapers or just like printed out pieces of paper. Same with mobile. We started to understand that the three lines across the hamburger icon meant a menu. We started to understand that the whole webpage didn't just have to be vertical, that you could swipe left and right, and Tinder obviously made this a much more famous UI. I think we're still developing those for AI. Mm-hmm. I really hope we're gonna start getting what I call like user dials for the model, 'cause right now all of the weights are determined by the researchers and the teams that build the, the product. But I want a personality dial. I don't want this like super pleasing, sycophantic AI that everybody seems to produce. I want sassy one who pushes back against me, and I wanna be able to tune that on my side as the user. I think voice, we still have a lot of surface area. I wrote an article called Voice Buttons. How could you communicate some of the emotional context from how a voice speaks to when it's dictated and you could like visually see.
Yeah.
Some of the emotion that was spoken into it when you send a text message, but that was originally delivered as a voice artifact. People are doing so little on the interface side today. I can't wait for the next, I feel like it's still sometime in the next 5 years when we start to coalesce around things. Mm-hmm. The agentic websites, websites that change based on every single person that comes to them. And a lot of people think that assistant is going to be a character. And it's going to be a character we have a relationship with. And so when the internet starts to become not browser, but character, we're a very long way away from that. I think that will be another really interesting UI shift as well.
