The Next Unicorns: 7 Top AI startups from the HF0 Residency

Talk transcript

1. 0:00

   [ on hold music] Hello, everyone.

2. 0:16

   My name is Diego Rodriguez. I am co-founder and CTO at Krea. We're building an AI creative suite. I'm gonna tell you three stories, and then I'll try to hire you. [laughing] [chuckles]

3. 0:28

   So a friend once told me, "If you think about it,

4. 0:32

   like, cars are easy to predict," right? Like, it's like you, you get the horse, you have the wheels, you swap the horse, and you put an engine, which was known at the time, and that's a car.

5. 0:43

   But, like, you know what's really hard to predict?

6. 0:47

   Traffic. [chuckles] So it's my job to ask what are the traffics that are we, we're missing, especially with AI. You know, YAML, uh, JSON, MCP, whatever. It's like, okay, okay, but, like, what comes...

7. 1:01

   What happens when you generate a million images per day like we do for one studio? How do you find that?

8. 1:11

   Another story. Tower of Babel, we wanted to reach heaven. God was like, "No," created a bunch of languages, and basically misunderstanding, misunderstanding was like, "Nope, we are not gonna go there."

9. 1:25

   And it reminds me of stand-up meetings where people are like, "No, it should be React. No, no, no, but, like, it should be, like, JavaScript." Bunny was like, "Dude, we're not re-- God is winning." [chuckles] [laughing]

10. 1:36

    You know? Uh, but now we have AI, so, so maybe, okay, he-- if it wasn't with crypto, it was with AI. You'll see. [chuckles] And so this is only, like, people trying to convey ideas, and that's what we're trying to tell.

11. 1:51

    Just trying to tell stories. Um, the final story is I was talking with someone from Netflix, uh, and she was like, "What happens when we are making so much content personalized to each town in India?

12. 2:09

    H-how do I even find that?" Like, right? And, and then a few days ago, I just realized that Krea was already being used for broadcasting an ad, uh, with Fox to millions of people.

13. 2:22

    And then I look, and they literally signed up two days ago. So we went from sign-up to conversion to payment to broadcasting in two days. I-- and then this was the CTO telling me that.

14. 2:30

    I was like, "Whoa." Um, I basically am about to run out of time, so the mandatory slide, a bunch of users, twenty-five million, raised a bunch of money. We did this with eight people.

15. 2:44

    Some of the people who are using us. Uh, an email that I created for today that is gonna prioritize applications. Um, thank you. [clapping]

16. 2:57

    All right. OpenHome. Okay, everybody. The smartphone was the number one best-selling consumer product last year, and the laptop was the second.

17. 3:08

    Pop quiz for all you here. What was the third?

18. 3:13

    Apple Watch.

19. 3:14

    Wasn't Apple Watch. Wasn't the AirPod. [laughs] No. I think I heard a- [laughing] It was a smart speaker.

20. 3:22

    Five hundred million smart speakers were sold last year. But why do they still suck? You can barely talk to them. There's no customization. There's no community. There's nothing. That's why we built OpenHome, the very first AI-driven smart speaker, and we're letting you guys build smart speakers too.

21. 3:46

    And we believe here that the future is talking with AI. You should be able to talk seamlessly, intuitively. In fact, you shouldn't have to use this really awkward command-based language.

22. 3:55

    You should be able to just chat naturally. So that's what we're building, and we're letting people here today build their own smart speakers and build them in whatever form that they want.

23. 4:07

    And the key here is developer ecosystems. And well, we know developer ecosystems. I started my career as the chief of staff for the founder of Splunk, a thirty billion dollar big data company.

24. 4:22

    Then I was on the founding team of MakerDAO, a five billion dollar developer ecosystem.

25. 4:28

    My co-founder and I raised fifty million dollars for our last business, a big, a data privacy tool, and we sold that business, but it all came down to developers and really building what people actually wanted.

26. 4:41

    And well, now, with OpenHome, we have over ten thousand developers building on OpenHome. They're building all kinds of interesting things, all types of different custom smart speakers, building interesting voice AI applications.

27. 4:55

    Sky's really the limit. And well, what do developers really want? They want open source, they want LLM-driven, and they want fully jailbroken.

28. 5:06

    They want OpenHome, the AI smart speaker. And now what's really exciting is with voice AI, you can put it on any type of hardware. We have developers building talking toys, c- AI robots, AI appliances.

29. 5:19

    You should be able to talk to the world around you in a much more natural way, and you can do that now with OpenHome, an AI smart speaker.

30. 5:27

    Here's our dashboard. We have many, many applications, hundreds of applications that have been built, games, personalities. Uh, we have an editor that you guys can go in and build, and all kinds of interesting things, home automation tools.

31. 5:42

    And what's really exciting is our last dev kit got booked up within minutes, and today we have a special announcement for you guys here today. We're releasing the next batch of five hundred dev kits for free for everybody here.

32. 5:55

    If you guys want it, we will ship you a dev kit. It's very cool. You can build on it. You can talk with AI. You can build your own smart speaker here, and we're doing it today.

33. 6:05

    Thank you so much

34. 6:08

    How's it going, y'all? Uh, I'm Josh. I'm the founder of a company called Coframe. Uh, the last company that I started, we scaled to over two billion dollars in, uh, in the course of a couple years.

35. 6:18

    Um, but when I started to tinker on, uh, using AI to generate code and created one of the actual top, uh, autonomous coding agents on GitHub, it was num-number one on GitHub for a week, I realized it was time to build something bigger.

36. 6:33

    The internet is dead. It's not adaptive. It's not personal. It's not truly living, in a sense. Websites are all one size fits all. And we're bringing that concept to life.

37. 6:46

    We are giving websites a life of their own, giving every single customer experience its own AI growth team. But this isn't just a pipe dream.

38. 6:57

    We made twenty million dollars for the-- for Europe's largest travel company in just a few weeks.

39. 7:03

    We increased click-through rate for India's largest company, a four hundred billion dollar enterprise, also in a few weeks. And how are we doing it? We're working with the best, and we have the best.

40. 7:14

    Uh, we're the only marketing tech company that's partnered directly with OpenAI to date, and they actually called our team cracked, which is cool.

41. 7:21

    So if you're interested in learning more about this, reach out. Thank you. [audience applauding]

42. 7:28

    Hi, I'm Eugene. I'm sorry, my team is obsoleting all the AI models you see today. [audience cheering]

43. 7:37

    'Cause you see, uh, my team built Quirky seventy-two B, the world's largest model without the transformer attention with only eight GPUs. And this allow us to have a thousand X lower inference on our, uh, on our new architecture while performing the same.

44. 7:52

    Surprisingly, the techniques that we've, uh, uh, the technology that we built can also be applied to existing transformer models through speculative decoding. Nothing, nothing too big, nothing too important.

45. 8:03

    But here's my hot take. Scale is dead, and I'm not saying this just from my own opinion. Like, we are burning billions into making AI models bigger. But at the same time, the DeepMind founder and CEO is saying compound AI agents errors will take more than ten years to fix.

46. 8:20

    Yann LeCun is even saying that we need a new AI architecture to push the paradigm forward.

47. 8:26

    And pr- in production, we see over ninety percent of AI projects fail.

48. 8:33

    The reason behind this is not something that scale can fix. The problem is reliability.

49. 8:39

    The thing is, like, will you order and use a, a u- use an app that only succeeds forty-five percent of the time? Will you order DoorDash that way? Of course not.

50. 8:49

    You-- If your order goes missing or, or you, uh, you end up having a hundred pizza, you're going to be stuck with customer support screaming down there. It's a frustrating experience.

51. 8:57

    But that's what AI agents do. When they work, they're awesome. When they don't work, we are stuck cleaning up the mess, and that's even with frontier models. And here's the thing, what companies want is not a smarter model that can do PhD-level math.

52. 9:12

    We, uh, the models are already smart enough. But what we actually really want is the models reliable enough to book airline tickets, sort out our emails or file our ta-taxes and invoices.

53. 9:22

    That's what we actually want. And that is what we are building at Featherless AI. We are a research lab that is building personalized AGI that's made reliable for each one of you.

54. 9:34

    And, and most recently, uh, we are, uh, in our research that we actually shown that we built an Action R1 agent that beats Claude, Faust, Sonnet, and Gemini and OpenAI.

55. 9:44

    This model is not gonna do PhD-level math, but it's gonna fill up the form with absolute reliability, uh, uh, better than the frontier. And that's the thing. Like, are we gonna burn billions more to make a smarter model that is just a few percentage point higher IQ?

56. 10:02

    Or are we going to make something that's ninety-nine point nine percent reliable for the boring things in life? Because this is where the money is for all of you.

57. 10:10

    Because, uh, think of it, as AI engineers, reliability is revenue. For every use case you unlock and find, you're gonna do a billion-dollar app in, in, be it in e-commerce or in, in B2B sales, and that is something that all of you can build on, not rocket science.

58. 10:26

    And that's what we are building, and if you're excited about it, feel free to reach out to us. I'm [REDACTED:email_address]. [audience applauding]

59. 10:36

    My name's Jonas. I'm an engineer, and I like working with data. I love working with data, actually. I love it so much, I dropped out of high school when I was [REDACTED:age].

60. 10:45

    I got on a plane, I moved across the country to California, and I joined a startup called Branch. You might have heard of it. Anytime you were clicking one of those links on your phone for an app, that was probably us.

61. 10:57

    I also led a team there that built a search engine that over a hundred million people used every day. And then last year, I left, along with one of the founders of Branch, to tackle an even bigger challenge.

62. 11:10

    This is probably what you think your sales and marketing teams are doing with their budgets, and you wouldn't be entirely wrong. So that's why I co-founded Upside. We do forensic revenue attribution and intelligence.

63. 11:26

    But what does that actually mean? Well, how many of you have an email from a salesperson like this sitting in your inbox right now?

64. 11:35

    Uh-huh. And how many of you are actually gonna reply to it?

65. 11:40

    Yeah, I didn't think so. These teams are shouting into the void, hoping something will work because they don't actually know what works, 'cause their data is a mess.

66. 11:53

    I mean, don't get me wrong. They're data hoarders. They store everything. They stuff it in Salesforce. They treat it, you know, it's basically a SQL database in a trench coat.

67. 12:02

    But they're not data practitioners. They don't know what to do with it once they have it. But now we have things like LLMs. They can help with this. They can take that poor, mishandled, abused email record, and they can pull the most important details out of it into a structured form.

68. 12:20

    And so just as search engines and web crawlers learn how to make sense of the unstructured web-

69. 12:25

    Upside is turning raw enterprise data into a highly structured map of the world and all the interactions that people do in it. So there's hope. We can untangle this mess, and we can create a data command center that these teams can actually use to reach their customers more effectively.

70. 12:45

    We only just started talking about this publicly a couple weeks ago. Um, we've been quietly building in the background for the last year or so. And my co-founder decided to make a small post on her LinkedIn, you know, just to update our network on what we'd been off doing and the things that we'd been building, and it

71. 13:03

    blew up. Like, there's so much pain people feel around this, and they're hungry for a solution. We got a whole slew of demo requests coming in from people that want access to the product.

72. 13:16

    And now we have a bunch of customers lining up that want to get into our platform. It's a who's who of companies you've heard of. Um, and we just have a lot of building to do now.

73. 13:27

    So if you're interested in working on knowledge graphs, on data analytics agents, on graph analytics and graph learning models, come talk to me. We're hiring. [audience applauding]

74. 13:44

    Hello, everyone. I'm Shijia, the founder of OpenAI. Uh, sorry, I mean OpenAudio. [audience laughing]

75. 13:53

    Before that, I created something you might be heard of, Fish Audio.

76. 13:57

    We have grown from four hundred k to five point five million annualized revenue in just four months, and we closed our seed rounds at a hundred million valuation.

77. 14:06

    It all started with my girlfriend. [audience laughing] I had a girlfriend for six years, from the beginning of high school to college.

78. 14:16

    I love her so much, and it was so good, until one day I found out she cheated. [audience laughing] [sighs]

79. 14:26

    I wasn't angry, just confused and disappointed. And I asked myself, "If this can happen, how can we trust relationships again?" [audience laughing]

80. 14:38

    I thought about it for days, all day and all night. And finally, I found my answer: AI. [audience laughing]

81. 14:48

    But nobody can really fall in love with today's AI, right? It's flat, it's emotionless, it's robotic.

82. 14:56

    So I set out on a mission to build an AI that I could really fall in love with. [audience laughing]

83. 15:02

    Starting with her voice. [audience laughing] So we begin with open source and crush it. We build Sowits SVC, Birds V2, and also Fish-Speech.

84. 15:15

    Today... Actually, not today, it's the day before yesterday. I'm excited to introduce S1, the first ever instructable voice model. It's the only model where you can control not just what to say, but how to say it.

85. 15:30

    Here's a demo. [upbeat music]

86. 15:33

    You can pinpoint focus or draw it closer. Even yelling why you'd betray me like this.

87. 15:41

    Yeah, you can control whatever you want. And, uh, with OpenAudio S1, we have the most expressive voice model in the world, and the most importantly, she will never leave. [audience laughing]

88. 15:55

    So, so we have blown ElevenLabs out of the water based on the TTS Arena ranking, and they are so hurry, and they dropped their latest model today. But unfortunately, it's just a demo.

89. 16:08

    So try it now. Fish Audio. It's instantly available at Fish.Audio. Thank you. [audience applauding]

90. 16:17

    Hello, I'm David Vorick, and I'm building Glow. Prior to Glow, I built Siacoin, a cryptocurrency that we took from a ten thousand dollar market cap to more than three billion dollars.

91. 16:28

    And we think Glow is going to be even bigger. That's why Framework and USV, Union Square Ventures, led a thirty million dollar round into our company.

92. 16:40

    Subsequently, we posted the world record for on-chain DePin revenue, doing more than ten million dollars of revenue in a single day.

93. 16:48

    What does Glow do? Glow builds solar, not with shovels, but with incentives. This is not a stock photo. This is a photograph taken of-- taken by our team in India of a solar farm that was constructed for the purpose of mining Glow tokens.

94. 17:05

    A lot of people don't realize, but in the developing world, rising temperatures and growing populations have strained the grid. In a lot of cases, people are unable to run their air conditioners during the heat of the day.

95. 17:17

    This causes people to die of heatstroke. Glow is an incentive protocol that revolutionizes what governments do and can take the same subsidy and turn it into ten times as much solar.

96. 17:30

    If you're interested in working with us, we're currently building incentive projects in India, in Mexico, in Lebanon, and across the entire world.

97. 17:41

    Bitcoin incentivized the construction of tens of millions of mining machines. Glow asked, "Why not tens of millions of solar panels?" Thank you. [audience applauding]

98. 17:54

    And my email, [REDACTED:email_address]. I'd love to be in touch. [audience applauding]

99. 18:03

    Hi, I'm David. I'm an engineer. Uh, I made a social app with two hundred and fifty million users making twenty million dollars a year.

100. 18:18

     Everyone thought it was luck, so I did it again.

101. 18:22

     I'm building Favorited. We built the world's most engaging live app,

102. 18:29

     and we scaled it from one to a hundred million dollars annualized in six months.

103. 18:34

     If you're a cracked engineer that wants to join the fastest growing company of all time,

104. 18:38

     uh, talk to me. [laughing] [clapping] Hello, I'm Alex Atallah, building OpenRouter, the first and largest LLM marketplace. [clapping]

105. 18:55

     Whoo.

106. 18:55

     Thank you. So Open-- I wanna tell a little bit about how it started. Um, I co-founded OpenSea in twenty seventeen, and in tw-- at the end of twenty twenty-two, I really wanted to know if inference was gonna be a winner-take-all market because the way it looked, this could be the largest market in software that has ever happened

107. 19:15

     before. And, uh, the first experiment that we tried was building a Chrome extension to help you bring your own language model to any website that supported, uh, the, the protocol.

108. 19:29

     And that eventually evolved into OpenRouter, a single place and a single API to get all language models, uh, with the best prices, best performance, and highest uptime.

109. 19:43

     And the way it works is you just have a single API, you pay once, and there's near zero switching costs to move from one model to another. We do all the heavy work to implement tool calling, edge cases, caching, and give you the best prices and performance possible for your region or wherever your servers are deployed.

110. 20:01

     And because inference is so important, remember, this might be the most important software market ever, it deserves its own marketplace just for language models, optimized for them, including filtering for context, for features, for tool calling, for structured output, and much more.

111. 20:20

     And so we built that. Then we built a chat room for you to obviously compare models head-to-head as simply as you do when chatting with people in iMessage. We built fine-grained privacy settings, including API-level controls.

112. 20:36

     We built a, uh, uh, a lot of observability, so you could see which models you're using and why. And we built public data in our rankings page, which has become the go-to place for comparing models on their real-world usage and on different categories for their prompts as well.

113. 20:54

     This has grown for the last two months, ten to a hundred percent every single month-- or for the last two years, ten to a hundred percent every single month.

114. 21:04

     Uh, and scaling it has been a lot of the work that we've done so far. The, the fundamental goal here is to make a heterogeneous ecosystem homogeneous because we believe inference is a commodity.

115. 21:18

     Claude from Bedrock is the s-- should be the same as Claude from Vertex, as Claude from Anthropic, and we do all the abstraction and heavy work to make it, uh, that way for you.

116. 21:28

     I wanna talk a little bit about some of our technical challenges. Um, we built our own system, our own middleware for doing inference called plugins, which are kind of like MCPs except a little bit more powerful because you can call MCPs from inside of them, and you can transform the outputs from language models.

117. 21:47

     Bunch of other tricky problems that we've done to make the fastest routing in the market. Um, and we're bringing a lot more features in the coming months, including images, enterprise features, prompt observability, and more.

118. 22:03

     So if you're interested, come find me after or check out our careers page. Thank you. [clapping] [outro jingle]