⚡ Behind Latent Space: A Rare Look Inside the Pod | AIEi Shanghai

“Oh, this is a special podcast because it's about AIE Shanghai. Uh, uh, AIE Shanghai is the first AIE in China. Uh, it's brought to us by Standard Partners, which is one of the local partners that we have.”

Claim: AIE Shanghai is the first AIE event in China.

“We have a few AIEs worldwide that my team doesn't run, and this is the first one in Shanghai. I've always wanted to bring AIE to China. I think there's a lot of East meets West, uh, and a lot of progress in China that needs to be shared and distributed among the global community.”

Claim: The speaker has long wanted to bring AIE to China.

“We have a few AIEs worldwide that my team doesn't run, and this is the first one in Shanghai. I've always wanted to bring AIE to China. I think there's a lot of East meets West, uh, and a lot of progress in China that needs to be shared and distributed among the global community.”

Claim: The speaker believes China has substantial East-West exchange and progress that should be shared with the global community.

“China's super interested in us. Um, there's been many Chinese fans of Latent Space, and I'll s- flash a few of them up on screen here on YouTube.”

Claim: The speaker says Latent Space has many fans in China and that China is highly interested in the podcast.

“And, uh, primarily, I want to recognize Meng Shao, who is our primary fan, uh, always translating basically every episode into Chinese, which I, which I am super appreciative of.”

Claim: Meng Shao translates nearly every Latent Space episode into Chinese.

“Um, and in general, I think I wanna also con- congratulate and, and celebrate, uh, the progress that China has made, uh, not just in open source models, but in robotics, uh, and a lot of other, uh, sort of startups.”

Claim: The speaker wants to celebrate China's progress in open-source models, robotics, and startups.

“Uh, you know, who, who can remember Manus that was started, uh, this time last year and then moved to Singapore. Um, and obviously Openclaw had a, a big run.”

Claim: Manus was founded about a year before the recording and later moved to Singapore.

“Uh, you know, who, who can remember Manus that was started, uh, this time last year and then moved to Singapore. Um, and obviously Openclaw had a, a big run.”

Claim: OpenClaw experienced a major period of success or activity.

“And we talk a lot about what that like-- what that's like in terms of the, the overall mood on the ground and how the scale of AI community is so different in China, and yet, uh, the ty- type of things get, that get adopted, as well as the, I guess, the startup culture.”

Claim: The scale of China's AI community differs substantially from that of other places discussed.

“People hear a lot about nine-nine-six. We talk about the reality of that versus, uh, in a place like Cognition. Um, and finally, we also talk a little bit about the origin story of Latent Space, which some of you may not have heard if you're newer to the podcast.”

Claim: The conversation compares the reality of 996 work culture with work culture at Cognition.

“Talk about when it was started and why it was started, and how we think about running the multiple shows that we're running. Um, and a little bit about safety and alignment and all that other stuff.”

Claim: The conversation covers Latent Space's origins, its founding rationale, and the operation of its multiple shows.

“So there's a little bit here for everyone. It's basically for the fans, if you've been following Latent Space for a while. Um, and if you're interested in AIE Shanghai, it's not being live streamed, but we're gonna post it up on the YouTube at some point.”

Claim: AIE Shanghai will not be livestreamed, but its recording is expected to be posted on YouTube later.

“It starts tomorrow. So anyone here in China who's watching this, I'm really looking forward to having it. And thanks to Meng Ying and, um, team for organizing all of this.”

Claim: AIE Shanghai is scheduled to begin the day after the speaker's statement.

“Um, I also have to shout out Han Hua and Leidu, who basically got this thing started. I-- we-- You know, it nearly hap-- did not happen a few times, but it's so, so exciting to have AIE in Shanghai and to bring Latent Space to China in general.”

Claim: Han Hua and Leidu helped initiate AIE Shanghai.

“So thanks for watching and look forward to the interview. The models are very good at clicking through every screen of your software and then replicating it independently.”

Claim: The speaker says models can effectively navigate every screen of a software product and independently reproduce it.

“Right? So the moment I have access to your software, I can clone it. So good luck. There's a lot of talking heads who have never coded a single day in their lives. They're all like, "Oh, my timeline to AGI is, like, two years away.”

Claim: The speaker says some people with no coding experience predict AGI in roughly two years.

“Right? So the moment I have access to your software, I can clone it. So good luck. There's a lot of talking heads who have never coded a single day in their lives. They're all like, "Oh, my timeline to AGI is, like, two years away.”

Claim: The speaker claims that access to a software product enables someone to clone it.

“Uh, all of you are gonna have-- not gonna have jobs in the future. You need to prepare for what's next." Conveniently, whatever is next looks like their job.”

Claim: The speaker criticizes commentators for presenting the future work people should prepare for as work resembling the commentators' own jobs.

“Uh, all of you are gonna have-- not gonna have jobs in the future. You need to prepare for what's next." Conveniently, whatever is next looks like their job.”

Claim: The speaker says some commentators warn people that they will lose their jobs in the future and should prepare for what comes next.

“They live in the wrong country. If they lived in the US- ... they would kill Amazon. They would kill, like, Facebook. The only reason they're not richer is they live in China.”

Claim: The speaker claims that the founders would outcompete Amazon and Facebook if they lived in the United States, and that living in China is the only reason they are not richer.

“AI doom. Uh, in all of human history, when a more advanced civilization comes along and meets a less advanced civilization, the less advanced civilization goes extinct.”

Claim: The speaker asserts that throughout human history, encounters between more and less advanced civilizations have resulted in the less advanced civilization going extinct.

“And this thing is evolving roughly one thousand times faster than us.”

Claim: The speaker says the thing under discussion is evolving at approximately 1,000 times the rate of humans.

“Oh, sorry. Fifth, fifth generation Chinese, right? Uh, most, uh, Singaporeans are m- majority Chinese. Um, I think I am from... People ask me, like, " ."”

Claim: The speaker identifies as a fifth-generation Singaporean of Chinese descent who grew up in Singapore.

“Um, uh, so, um, and I don't know. And, uh, like, that's, that's like how it feels like to be, uh, like Chinese diaspora. Right? Like, we don't know what our origins are because it's just mixed and nobody kept any record. It's like nobody even knows our birthday is, like sometimes, you know?”

Claim: The speaker says Chinese diaspora identity can feel uncertain because family origins are mixed and no records were kept.

“But fifth generation Singapore, uh, grew, grew and raised up in Singapore, and I went to the US for university.”

Claim: The speaker was raised in Singapore and attended university in the United States.

“And, uh, did my first career in finance. I was in, uh, trading and investment banking and, uh, hedge funds for ten years. Uh, sorry, eight years.”

Claim: The speaker spent eight years working in trading, investment banking, and hedge funds.

“And then switched to, um, tech. So I worked in, uh, Amazon and a few startups, and then four years ago became independent, wrote Latent Space, uh, figured out that there was gonna be this AI engineer thing-”

Claim: After a finance career, the speaker moved into technology, worked at Amazon and several startups, and became independent four years before the recording.

“Yeah, so it's very funny because I recently had to go over my taxes ... and like figure out like what money do I have. So Cognition is the biggest one, is I'm an advisor. I'm still an advisor to Cognition.”

Claim: The speaker says Cognition is the largest of their current professional roles and that they are an advisor to Cognition.

“Uh, I was an employee for a long time, for, for, for a while. So basically the idea was, uh, I have three companies. I have AIE, I have Latent Space, and then I have miscellaneous. It used to be small ai, which is my personal sort of startup vehicle.”

Claim: The speaker describes their business activities as three entities: AIE, Latent Space, and miscellaneous work, formerly organized under Small AI as a personal startup vehicle.

“Uh, I was an employee for a long time, for, for, for a while. So basically the idea was, uh, I have three companies. I have AIE, I have Latent Space, and then I have miscellaneous. Worked there for six months, and then stepped back to be an advisor because AIE is now, you know, a global-”

Claim: The speaker says they were an employee at Cognition, worked there for six months, and then stepped back into an advisory role.

“It used to be small ai, which is my personal sort of startup vehicle. Uh, basically got acqui-hired into Cognition.”

Claim: The speaker says their personal startup vehicle was acqui-hired into Cognition.

“Worked there for six months, and then stepped back to be an advisor because AIE is now, you know, a global- ... conference, conference organization, and I, I had to be CEO. I couldn't be part-time employee, part-time here. Cognition is very big. Fireworks is very big. Modal is very big.”

Claim: The speaker says they needed to serve as CEO of AIE after it became a global conference organization, rather than split their time between part-time employment at Cognition and AIE.

“So that, that is my specialty in, which is that I am a technical, but I also write and speak, and that is a good combination of skills to have for people because any startup will want to pay you to like talk about their stuff.”

Claim: The speaker says their specialty combines technical expertise with writing and speaking, and that startups will pay people with this combination to talk about their products.

“And I think, you know, then, then you start working with a bunch of things. So like, uh, a lot of, uh, this is not a normal career. I actually don't even recommend it for most people, but it is my role in life that I, I can advise, uh, a few different companies and, uh, work with them in, in exchange for stock, and that's the, that's the rough idea.”

Claim: The speaker does not recommend their atypical career path for most people; they advise several companies in exchange for stock.

“There's, there's kind of two different things that you're mixing together, right? The first thing is the principle of learning in public, which is something I realized when I left finance and I worked in tech.”

Claim: The speaker says they recognized the value of learning in public after moving from finance to technology.

“Why is tech so much better, going so much better for me? It feels so much more comfortable. Uh, and the very core reason for people who are interested in both fields, both are high-paying jobs, like, uh, prestigious jobs.”

Claim: The speaker characterizes finance and technology as fields that both offer high-paying, prestigious jobs.

“The difference is finance is zero-sum, right? If I know something and I tell you- ... uh, that, that truth lo- loses value.”

Claim: The speaker characterizes finance as zero-sum, where sharing information causes the information to lose value.

“But in tech, it is most stuff is things are positive-sum. If I know something, I tell you, it increases value. Uh, and that's a very different paradigm for people who are crossing over from finance to tech, and it's very liberating.”

Claim: The speaker says most activity in technology is positive-sum: sharing knowledge increases value, a paradigm they find liberating for people transitioning from finance to technology.

“It means that people will get on stage and they open source their work.”

Claim: The speaker says people in technology often present their work publicly and open-source it.

“They share their code. They screw up, they share it that they screwed up. In finance, no. Like you wanna pretend that you never screwed up and everything is top secret.”

Claim: The speaker contrasts technology, where people share code and acknowledge mistakes, with finance, where people conceal mistakes and treat information as top secret.

“And I think that is something that I really took to heart, and I, I think I realized that, oh, the, the other thing that happened to me was my first job was at Two Sigma, which is still in finance, but I was on the software engineering side.”

Claim: The speaker says their first job was a software engineering role at Two Sigma, a finance firm.

“And I realized that I had a bad boss. So then I got really bored, and I started to do a lot of writing and speaking because at work I wasn't able to do the things that I wanted to do 'cause they didn't let me.”

Claim: The speaker says a bad boss and limited opportunities at work left them bored, prompting them to write and speak outside their job.

“But like, obviously, because I was so bored at work, and then I put all my energy into the, the stuff outside of work, then obviously that thing took off because no one else is doing that.”

Claim: The speaker attributes the success of their outside-work activities to investing energy in them while bored at work and to the lack of others doing the same activities.

“And by comparison, that becomes you being able to choose your own boss, right? Like most people, when they have a job, they join a company, they are assigned a boss, they assign a team. Most of the, most of the time you don't even know if like the team is good or not or like whatever, and then you find out it's not good. You're stuck. When you learn in public, you actually can choose your own boss.”

Claim: The speaker says employees are generally assigned a boss and team when joining a company, often cannot assess the team beforehand, and may be stuck if the team turns out to be poor.

“Most of the, most of the time you don't even know if like the team is good or not or like whatever, and then you find out it's not good. You're stuck. When you learn in public, you actually can choose your own boss. You choose... You, you like work for yourself, choose your own boss, all those things, so it's very, very good. I think the fast-forward to Latent Space, that was created when stable diffusion happened.”

Claim: The speaker says learning in public can enable people to work for themselves and choose their own boss.

“You choose... You, you like work for yourself, choose your own boss, all those things, so it's very, very good. I think the fast-forward to Latent Space, that was created when stable diffusion happened. This is before ChatGPT.”

Claim: The speaker says Latent Space was created around the time Stable Diffusion emerged, before ChatGPT.

“When stable diffusion happened, I was in a startup that was not going very well. Uh, it's called Airbyte. It's still doing okay today, but nobody would call it a rocket ship. That's just, that's objectively true, and I would say that to the founders.”

Claim: The speaker says that when Stable Diffusion emerged, they were at Airbyte, which was doing okay but was not a high-growth rocket ship.

“But because I was like, "Okay, I have-- I, I did like some machine learning courses, but I'm not an AI expert. I am so interested in this field.”

Claim: At the time, the speaker had completed some machine-learning courses, did not consider themselves an AI expert, and was highly interested in AI.

“There's, there's only one way for me to get into the field, which is to learn in public." And so that's what I did. I started a blog, and I started, um, analyzing and learning and, like, writing up my learnings into a blog.”

Claim: The speaker believed learning in public was their only route into AI, and began a blog to analyze topics and document their learning.

“And at, at first nobody will notice, but then eventually if you start talking about people's things, like, uh, literally, Emad Mostaque was the CEO of Stability AI, which-”

Claim: The speaker says their blog initially attracted no attention, but eventually gained notice as they wrote about other people’s work.

“And at, at first nobody will notice, but then eventually if you start talking about people's things, like, uh, literally, Emad Mostaque was the CEO of Stability AI, which- ... created Stable Diffusion, which is the, the reason I started the blog, read the blog post analyzing Stable Diffusion and said, "This is good." And I think, like, that work was basically me learning AI, like the past, like five years of like just grinding and, and, and writing about that.”

Claim: The speaker says Emad Mostaque, who was CEO of Stability AI, read their Stable Diffusion analysis post and praised it as good.

“... created Stable Diffusion, which is the, the reason I started the blog, read the blog post analyzing Stable Diffusion and said, "This is good." And I think, like, that work was basically me learning AI, like the past, like five years of like just grinding and, and, and writing about that.”

Claim: The speaker says they spent roughly the past five years learning AI through sustained effort and writing about the field.

“And I think that is really the, you know, like you are not qualified, and I'm still probably not that qualified today, but I'm a bit more qualified than I used to be f- four years ago.”

Claim: The speaker believes they were probably still not very qualified, but were more qualified than they had been four years earlier.

“So that's good. And I think that's the only way you can learn. Like, I think, um, especially in Asian cultures, there's a lot of emphasis on credentialism.”

Claim: The speaker believes learning by doing is the only way to learn.

“So that's good. And I think that's the only way you can learn. Like, I think, um, especially in Asian cultures, there's a lot of emphasis on credentialism.”

Claim: The speaker says credentialism receives substantial emphasis in many Asian cultures.

“You need that piece of paper. You need like some qualification. Like if it's not-”

Claim: The speaker says credential-focused expectations require people to have a diploma or another qualification.

“... you must have had a job, uh, to do it. And how do you get the job? You must have a degree. Like it's just like a s- infinite loop of things that basically mean that when you're 16, you need to decide what you wanna do in life.”

Claim: The speaker argues that linked requirements for degrees, certifications, and jobs can create an infinite loop that effectively forces people to decide on a career at age 16.

“But like I think life is a lot more flexible than that, uh, and you can choose to pivot if you want to. Uh, you just have to be really persistent and also, uh, realize that Just because you worked very hard, you're not entitled to win.”

Claim: The speaker believes people can change career direction because life is flexible, but doing so requires persistence, and hard work does not guarantee success.

“You have to work very hard and be good. I think a lot of people try, a lot of people fail-”

Claim: The speaker says that success requires working very hard and being good, while many people who try fail.

“... and they are like, "Why didn't it fail?" Well, you're not good. Too bad. But some people try, and they, they worked and, and they had judgments. Uh, they really did, uh, work on substantial things rather than insubstantial things.”

Claim: The speaker says that successful people have worked on substantial rather than insubstantial things.

“Um, so, so yeah. That start- they started LeanSpace, and then, uh, very, very interestingly, LeanSpace was just a blog. It was just- Yes, just a blog, and I had no other, no other idea apart from this is Swyx writing about AI. That's it. But we happened to host some events, like some gatherings, like, uh, oh, like, you know, ChatGPT came out, like obviously the most exciting thing-”

Claim: LeanSpace began as a blog focused on Swyx writing about AI, with no broader plan initially.

“Yes, just a blog, and I had no other, no other idea apart from this is Swyx writing about AI. That's it. But we happened to host some events, like some gatherings, like, uh, oh, like, you know, ChatGPT came out, like obviously the most exciting thing- ... of all time. And then we wanted to host some hackathons, so we hosted a hackathon. And at the hackathon, that's where the idea for AI engineer came about, which is the realization-- It actually was this, uh, woman that came up to me and said, "I'm, I'm all in on AI.”

Claim: LeanSpace hosted gatherings and hackathons, and the AI engineer concept emerged at one of its hackathons.

“... of all time. And then we wanted to host some hackathons, so we hosted a hackathon. And at the hackathon, that's where the idea for AI engineer came about, which is the realization-- It actually was this, uh, woman that came up to me and said, "I'm, I'm all in on AI. I'm non-technical. What can I do?" And I realized that the simple answer is, based-- compared to an engineer, she can do nothing.”

Claim: The speaker concluded that a non-technical person had little to contribute to AI compared with an engineer.

“Engineers are about to be so much more powerful than the non-engineer- ... because engineers can control one hundred AIs at once and program it.”

Claim: The speaker predicted that engineers would become much more powerful than non-engineers because they could program and control 100 AI systems simultaneously.

“Whereas the, the maximum that a non-technical person can do is prompt better. Then I was like, "Oh, like I think I have seen this before." So my, my career-- Throughout my career, I've seen like professionalization of roles that are non-prestigious.”

Claim: The speaker says that, over their career, they have observed non-prestigious roles becoming professionalized.

“Whereas the, the maximum that a non-technical person can do is prompt better. Then I was like, "Oh, like I think I have seen this before." So my, my career-- Throughout my career, I've seen like professionalization of roles that are non-prestigious.”

Claim: The speaker says that, at most, a non-technical person can improve their prompting.

“So what do I mean by non-prestigious? It is not prestigious to be a prompt engineer. The prestigious thing is you train large language models. You do pre-training, you do post-training.”

Claim: The speaker characterizes prompt engineering as non-prestigious and training large language models through pre-training and post-training as prestigious.

“You have to go through a PhD. You make fancy algorithms like GRPO, you know, and, and you, you do manifold hyperconnections, uh, which is what DeepSeek does.”

Claim: The speaker describes PhD-level work on advanced algorithms such as GRPO and manifold hyperconnections as prestigious, and says DeepSeek uses manifold hyperconnections.

“That's prestigious, and that continues to be prestigious. But I think the overcorrection that people have is that looking at everything else that is not that ML research, that is worthless.”

Claim: The speaker believes it is an overcorrection to regard all work outside machine-learning research as worthless.

“Uh, that actually was untrue. You actually do want to build harnesses, and harnesses do not require you to know advanced, advanced math or have a lot of experience in supercomputing.”

Claim: The speaker says building AI harnesses is worthwhile and does not require advanced mathematics or extensive supercomputing experience.

“Like, actually, you just need to be a very good software engineer that is open-minded enough to understand how to put AI into traditional software code, which is a harness.”

Claim: The speaker says that building an AI harness requires strong software-engineering skills and an open-minded approach to integrating AI into conventional software.

“We didn't have the word for harness at the time, but it was very clear that engineers would be able to do things like this that normal people cannot.”

Claim: The speaker says it was clear, even before the term “harness” was used, that engineers could do things ordinary people could not.

“And they would be in demands. And there's not only that. Companies are trying to hire these kind of people, and developers want to become these kind of people.”

Claim: The speaker says companies are seeking to hire AI engineers and developers want to become AI engineers.

“They want, they want a term to call themselves, um, something. And so all you have to do is figure out the right term, which is the simplest one is AI engineer, and make it a industry, make it a discipline, and so that's what I did.”

Claim: The speaker says they chose “AI engineer” as a simple job title people could identify with and worked to establish it as an industry and discipline.

“So there was AI Tinkerers, which is another- ... significant community. I, I know Joe and Alex, uh, well. And then there were people who were saying LLM engineers or applied AI engineer or all these, all these other terminology.”

Claim: The speaker says that alternatives to “AI engineer” included “LLM engineer” and “applied AI engineer,” and that AI Tinkerers was a significant community.

“And I think basically you have to focus on the job title that people want to have.”

Claim: The speaker believes that defining a professional community should focus on the job title people aspire to hold.

“I think so. Um, I, I forget, I forget, I forget the authorship relationship, but I think it's James Clear. Basically, if you want to deeply understand or connect with somebody, you have to connect with them not on the level of their job- ... not on the level of their activities or hobbies or whatever. It is their identity. Like, who are you?”

Claim: The speaker says that deep connection with a person comes from engaging with their identity and aspirations, rather than only their job, activities, or hobbies.

“Uh, what, what do you aspire to be? If you can give them the path towards being the better version of themselves, then you can build a community around that.”

Claim: The speaker believes a community can be built by offering people a path toward becoming a better version of themselves.

“Uh, and so a lot of non-AI people wanted to be in AI.”

Claim: The speaker says many people who were not already in AI wanted to work in AI.

“And what is the path for them? Um, well, you give them a job title that pays money.”

Claim: The speaker says that one way to offer non-AI people a path into AI is to give them a paid job title.

“And so, so yeah, yeah, now there's a lot more AI engineers in the world.”

Claim: The speaker says the number of AI engineers in the world has increased substantially.

“Honestly, it hasn't really changed that much. I think that what's hot within AI engineering has changed, but the job title is pretty much correct. I think there are people-- uh, is it in the early days, it used to be RAG.”

Claim: The speaker says retrieval-augmented generation (RAG) was prominent in the early days of AI engineering.

“Honestly, it hasn't really changed that much. I think that what's hot within AI engineering has changed, but the job title is pretty much correct. I think there are people-- uh, is it in the early days, it used to be RAG.”

Claim: The speaker says the AI engineer job title has changed little, while the areas of AI engineering receiving attention have changed.

“And managing context and all that. Um, and, and then more recently it's been coding agents, it's been harness engineering, it's been, uh, maybe loop engineering or whatever.”

Claim: The speaker says coding agents and harness engineering, and perhaps loop engineering, have been prominent areas of AI engineering more recently.

“There's also other disciplines on memory and all these things. So for those who are interested in all the fields of AI engineering, in AIE World's Fair, which is the, the, the flagship, uh, biggest conference, we have 30 different tracks of what it means to be an AI engineer.”

Claim: AIE World's Fair offers 30 tracks about the work and disciplines involved in being an AI engineer.

“Um, and eval is part of that, multimodal is part of it, you know, all that stuff. So I think it has shifted over time. Um, I think agent engineer maybe is, is one of them, or-”

Claim: The speaker says evaluation and multimodal work are included in AI engineering, whose areas of focus have shifted over time.

“... AI architect. Uh, you know, so AI engineer is like the individual contributor version of this, but who is like the company-level, company-wide boss. And, uh, AI architect is, uh, seems to be the emerging term for, for that, that person that decides h- this is how everybody in the company does AI. Yeah, that's AI architect. So and then, like, there's, there's all the, all the other, like, AI designer, AI PM, all these things.”

Claim: The speaker distinguishes an AI engineer as an individual contributor from an AI architect, who decides how AI is done across a company.

“And, uh, AI architect is, uh, seems to be the emerging term for, for that, that person that decides h- this is how everybody in the company does AI. Yeah, that's AI architect. So and then, like, there's, there's all the, all the other, like, AI designer, AI PM, all these things.”

Claim: The speaker says “AI architect” seems to be emerging as the title for the person who determines company-wide AI practices.

“So, uh, it's just like who does that work and who writes code and who is responsible for checking that code. If, if something fails, are you responsible? No, you're not an AI engineer.”

Claim: The speaker defines an AI engineer partly by responsibility for writing and checking code, including accountability when it fails.

“You know, if the, the person who wakes up in like four AM in the morning to go and fix it, that's the engineer.”

Claim: The speaker characterizes an engineer as the person who takes responsibility for fixing problems, even at four in the morning.

“Right? And so we want to serve those people, and that's the real core of it. With-- the other problem with AI is that there's a lot of bullshit. There's a lot of talking heads who have never coded a single day in their lives, but they are very big picture. They're very, very senior. They're like, "Oh, I'm CTO of blah, blah, blah." And, but they haven't-- they don't touch code.”

Claim: The speaker criticizes AI discourse as containing many senior, big-picture commentators who have never coded or do not work directly with code.

“And so a lot of AI conferences and communities focus on those guys because they have big, fancy titles, but then they all sound the same. They're all like, "Oh, my timeline to AGI is like two years away. You know, uh, all of you are gonna have-- not gonna have jobs in the future. You need to prepare for what's next." Conveniently, whatever is next looks like their job, so it's very self-serving.”

Claim: The speaker says many AI conferences and communities center senior figures who repeat predictions that AGI is about two years away and that people will lose their jobs and need to prepare for what comes next.

“You know, uh, all of you are gonna have-- not gonna have jobs in the future. You need to prepare for what's next." Conveniently, whatever is next looks like their job, so it's very self-serving.”

Claim: The speaker argues that some senior commentators’ warnings about the future are self-serving because their proposed future roles resemble their current jobs.

“And I think that that's one of the reasons that AI Engineer Works is that we try to focus on the lower-level engineer that usually doesn't get the spotlight, that actually did the work and is able to talk about all the war stories and share with their fellow engineers about what they're doing.”

Claim: AI Engineer World's Fair aims to spotlight hands-on engineers who can share their work and practical experiences with peers.

“And so I borrowed this idea from other parts, other conferences I've seen in my career, like GDC, um, and KubeCon and JSConf and PyCon and all these other conferences that I've spoken at, where basically, uh, it is engineer to engineer, and they are the decision-makers for the tech stack, and they are the people who create and innovate the new things.”

Claim: The speaker says AI Engineer World's Fair draws on conferences such as GDC, KubeCon, JSConf, and PyCon, which facilitate engineer-to-engineer exchange.

“And so I borrowed this idea from other parts, other conferences I've seen in my career, like GDC, um, and KubeCon and JSConf and PyCon and all these other conferences that I've spoken at, where basically, uh, it is engineer to engineer, and they are the decision-makers for the tech stack, and they are the people who create and innovate the new things.”

Claim: The speaker says engineers make technology-stack decisions and create and innovate new technologies.

“And they're not necessarily the most famous person, but they are definitely correct when it comes to the engineering. Uh, there's no bullshit, right?”

Claim: The speaker says conference speakers may not be famous but are reliable on engineering matters.

“And I think there's a lot of bullshit in the other things.”

Claim: The speaker believes other forms of AI events and discourse contain substantial amounts of misleading or low-quality talk.

“AIE World's Fair is basically a, a similar industry flagship conference similar to the ones that every other industry al-already has.”

Claim: The speaker describes AIE World's Fair as a flagship industry conference analogous to major conferences in other fields.

“So gaming has GDC, thirty thousand people in San Francisco. Security has two. Security has RSA, Security has Black Hat. If you're in the cybersecurity field, you go to one of those things.”

Claim: The speaker cites GDC, with 30,000 attendees in San Francisco, and RSA and Black Hat as major conferences for gaming and cybersecurity.

“And that's, that's where you learn. I, I really like the gaming one because gaming is very similar to labs. You have AAA studios, they work on the hardest games in the world, and it is extreme engineering, and they're all top secret.”

Claim: The speaker says AAA game studios resemble AI labs in undertaking difficult, technically demanding projects while keeping their work secret.

“They don't do anything else. But one time a year they come to GDC and they will share, like, "This is how I did Game of the Year." And that's the only way, if you're trying to be a game developer and you want to work at a AAA game studio, that's the only way you learn.”

Claim: The speaker says GDC provides an annual opportunity for studios to share how they made a Game of the Year, and characterizes it as the only way aspiring AAA-studio developers can learn that work.

“So I think that now the model labs are similar to game studios. They have a lot of money. They're working on very, very big major projects. There's a lot of smaller companies that are, you know, trying to compete with them, but the labs are the important ones.”

Claim: The speaker compares model labs to game studios, saying model labs have substantial funding, pursue major projects, and are important despite competition from smaller companies.

“... talking to other engineers. And then there's like fifty other labs that are not OpenAI and Anthropic that also-”

Claim: The speaker says roughly fifty labs other than OpenAI and Anthropic have interesting work or offerings.

“... have something interesting, right? So they-- usually, every industry needs one big, like everyone comes here, and then you just walk around the room and just talk to everybody. And then they can talk to each other.”

Claim: The speaker argues that each industry needs a large shared event where participants can meet and talk with one another.

“Which is very interesting because especially for Chinese labs- ... some of whom are on the entity list. Is this something that you guys talk about?”

Claim: The speaker says some Chinese AI labs are on the entity list.

“They cannot officially take meetings with you. A lot of American companies cannot officially take meetings with you. They're not allowed. They don't even respond to you. It's not because they don't want to, it's not because they're assholes.”

Claim: The speaker says companies on the entity list cannot officially meet with his group, and many American companies are not allowed to officially meet with those companies.

“They're not allowed. They don't even respond to you. It's not because they don't want to, it's not because they're assholes.”

Claim: The speaker says American companies’ failure to meet or respond is due to restrictions, not unwillingness or personal hostility.

“And I think that's, that's kind of interesting, as for me to create opportunities for, for magic to happen. And it's like people get hired, people find a co-founder. I have one couple that started, uh, dating and now they, you know, they're, they're, they're basically married.”

Claim: The speaker says the community creates opportunities for people to meet, leading to hires, co-founder relationships, and at least one couple who began dating and are now basically married.

“... uh, yeah, that, that is my number one KPI- ... number of, number of marriages and babies that I've created. Uh, and so, so I, I, I really like that. I, I think that, uh, we can create that kind of high-quality community.”

Claim: The speaker's number-one KPI is the number of marriages and babies he has helped bring about, and he believes the community can be high-quality.

“And, you know, we a- we're aiming for six thousand people this year. We managed to hit seven thousand plus. One point five million online watched our talks. So there are people who cannot afford to come, don't live in San Francisco, don't live in the US, they can still watch-”

Claim: The speaker says the event aimed for 6,000 attendees that year, exceeded 7,000, and had 1.5 million online viewers for its talks.

“... our videos and catch up on all the content. We publish every single talk, every single workshop. And like I said, it's like thirty different tracks. Um, there's too many tracks to name , but, uh, you can see on the website.”

Claim: The speaker says the organization publishes every talk and workshop and offers roughly 30 tracks.

“But I, I basically think that I really try to focus on things that are working, that you can make money. Uh, there are clear role models for you.”

Claim: The speaker says he focuses on approaches that are working, can generate revenue, and have clear role models.

“Like, if you're trying to learn this field or you're a startup trying to build something, there's somebody who has probably talked about your thing at AI.”

Claim: The speaker says that someone has probably spoken at AI about the subject of interest to a person learning the field or building a startup.

“That becomes a very valuable resource for anyone looking up anything. So another thing I've actually recently done that I haven't really been super public about is I transcribe everything and I catalog everything.”

Claim: The speaker says he transcribes and catalogs all the organization's content, making it a valuable reference resource.

“So you don't even have to watch the talks. You can just take all of our website and feed it into an LLM and then it can teach you whatever.”

Claim: The speaker says users can feed the organization's website content into an LLM and use it to learn about a subject.

“Every-everything is, uh, is MCP.”

Claim: The speaker says the organization's materials or system are available through MCP.

“Um, everything has been transcribed. Uh, it's available as a data set. If you, if you wanna train- ... your LLM on it, yeah, go ahead.”

Claim: The speaker says all the content has been transcribed and is available as a dataset that people may use to train an LLM.

“... ai.engineer/data.”

Claim: The speaker says the dataset is published at ai.engineer/data.

“... which also, yeah, shows the L- shows the MCP and the data sets that you can download. It just de-depends how you wanna consume it. You know, we transcribe everything. We also deep research every speaker and every organization.”

Claim: The speaker says the organization transcribes all its content and conducts in-depth research on every speaker and organization.

“... which also, yeah, shows the L- shows the MCP and the data sets that you can download. It just de-depends how you wanna consume it. You know, we transcribe everything. We also deep research every speaker and every organization.”

Claim: The speaker says the site provides access to MCP and downloadable datasets, allowing users to choose how to consume the material.

“So whoever has ever spoken at AI, we'll-- you'll have a profile and we'll basically try to make the Wikipedia of AI.”

Claim: The speaker says the organization aims to create a profile for every person who has spoken at AI and to build a Wikipedia-like resource for AI.

“Mostly that is basically exactly following the JSConf model. JS-JSConf is the number one JS conference .”

Claim: The speaker says the organization's international expansion follows the JSConf model and describes JSConf as the leading JavaScript conference.

“And, uh, it's a very simple model, right? Like I will never reach every country in the world, but every non-country in the world will probably wanna do some AI. So obviously their country deserves to have a leading conference.”

Claim: The speaker says he cannot personally reach every country, but believes every country will probably want to engage in AI and therefore deserves a leading AI conference.

“They can either make their own or they can borrow our brand- ... and our YouTube channel and our social media. So basically we just set up a partnership program. Paris was the first to apply.”

Claim: The speaker says local organizers can create their own conference or partner using the organization's brand and media channels; Paris was the first applicant to its partnership program.

“A very fun fact, the organizers of AI Paris got acquired into Mistral. ... you know, France's and Europe's leading AI lab is- ... uh, runs AI Paris, which is very cool. Like, I want that for every single, you know, country to like... Yeah, you, you have AI India, you have AI Japan.”

Claim: The speaker says the organizers of AI Paris were acquired by Mistral and that Mistral now runs AI Paris.

“... uh, runs AI Paris, which is very cool. Like, I want that for every single, you know, country to like... Yeah, you, you have AI India, you have AI Japan.”

Claim: The speaker wants the partnership conference model to operate in every country, including examples such as AI India and AI Japan.

“Yeah. So I'm, I'm working with all those, those companies now and it's very cool. Like, I will never know that region as much as them, but like they of-- do they deserve a good conference?”

Claim: The speaker says he is working with local companies and will never know their regions as well as those companies do.

“Yeah. The, the world is very big. So, um, we, we follow that model. So basically it's the idea that you have-- there's, there's three things, right? Like one, you agree to not abuse the brand name and like, we'll, you know, um, we'll have some sort of like, uh, revenue share thing. Two, you're publishing all, everything that you're, you're, you're doing. And then three, you have a code of conduct where like if, you know, if, um, someone is, uh, being abused or mistreated, there's some way of, of resolving that.”

Claim: The speaker says conference partners must respect the brand and agree to revenue sharing, publish their activities, and maintain a code of conduct with a process for addressing abuse or mistreatment.

“I think that's mostly an, an ethical thing rather than anything else. So we take a lot of risk by doing this because we're sharing our brand with people that we don't work with.”

Claim: The speaker views the code of conduct primarily as an ethical matter and says sharing the brand with unaffiliated partners involves substantial risk.

“So far it hasn't happened”

Claim: The speaker says that, so far, no partner has held a bad AI event.

“Yeah. Just I feel like, uh, mostly we're, we're careful about who we wanna work with, but then also, uh, the, the kind of people that are drawn to us anyway are pretty enthusiastic and organically interested in like...”

Claim: The speaker says the organization is selective about its partners and that the people attracted to it tend to be enthusiastic and organically interested.

“Yeah. First of all, I'm Singaporean, but that actually did not matter at all. The current top conference in Singapore that is an AI conference is actually a crypto conference, right?”

Claim: The speaker says Singapore's leading conference that is an AI conference is currently a crypto conference.

“Yeah. First of all, I'm Singaporean, but that actually did not matter at all. The current top conference in Singapore that is an AI conference is actually a crypto conference, right?”

Claim: The speaker says his being Singaporean did not matter to the relevant decision or outcome.

“A lot of crypto people want to be, want to be AI.”

Claim: The speaker says many people from the crypto community want to enter or participate in AI.

“"We need something better." And so yeah, uh, now I think AI Singapore is, is up there. It's at least competitive- ... and maybe, maybe better than Super AI. And, uh, I'm very proud of the team for doing that. I-- and I didn't organize any of it. So I think that is the important thing for every country in the world to have it, and we can support that.”

Claim: The speaker thinks AI Singapore is now among the leading AI conferences in Singapore, at least competitive with and possibly better than Super AI.

“... and maybe, maybe better than Super AI. And, uh, I'm very proud of the team for doing that. I-- and I didn't organize any of it. So I think that is the important thing for every country in the world to have it, and we can support that.”

Claim: The speaker believes every country should have an AI conference and says his organization can support that goal.

“... and maybe, maybe better than Super AI. And, uh, I'm very proud of the team for doing that. I-- and I didn't organize any of it. So I think that is the important thing for every country in the world to have it, and we can support that.”

Claim: The speaker says he is proud of the team responsible for AI Singapore and did not organize the event himself.

“Yeah. So China is the tricky one, right? Because I'm not a-- I'm Singapore citizen, I'm a Chinese. And obviously there's a lot of restrictions between the US and China. So that is the main tricky part that-- which is why I needed Lei and his team to really own the whole thing.”

Claim: US-China restrictions make operating in China difficult for the speaker, so he needed Lei and Lei's team to own the entire operation.

“Because I, like, I cannot even set up a local entity in China, right? There's a lot of controls. I don't want to get in trouble because my main business is US, right?”

Claim: The speaker says he cannot establish a local entity in China because of controls and wants to avoid trouble because his main business is in the United States.

“So I cannot get into trouble with the White House. But if you want to share the brand and run it as your own thing and it's a franchise, whatever, okay. Like, there's lots of franchises.”

Claim: The speaker wants to avoid trouble with the White House but is comfortable with another party using the brand and operating independently as a franchise.

“As far as we're concerned, I'm McDonald's and you're McDonald's China.”

Claim: The speaker characterizes the arrangement as analogous to McDonald's and McDonald's China, with the parties operating as separate franchise-like entities.

“Okay. I'm wondering how many of your listeners will care about this. Okay. You will only care if you own a business.”

Claim: The speaker expects that listeners will care about the topic only if they own a business.

“Because most developers do not own businesses. They're like, not that interested. But if you aspire to own a business someday- ... you will be better care .”

Claim: The speaker says most developers do not own businesses and are less interested in business ownership topics, while aspiring business owners may be more interested.

“Oh, yeah. SaaS is dying.”

Claim: The speaker asserts that software as a service (SaaS) is dying.

“My budget for running a conference is somewhere between three to ten million, depending on the size of the conference. Uh, roughly about 300k of that per conference is software.”

Claim: The speaker's budget for running a conference ranges from $3 million to $10 million, depending on its size.

“My budget for running a conference is somewhere between three to ten million, depending on the size of the conference. Uh, roughly about 300k of that per conference is software.”

Claim: The speaker spends roughly $300,000 on software for each conference.

“I'm paying s- like, that's like 10% of my budget. That's like 10% of my margin that I could be- ... getting back. That is to just somebody else's software that is shitty, slow, uh, ugly, not exactly what we want. And, you know, so I'm paying three hundred thousand, m-maybe I call it paying one million a year in just like SaaS.”

Claim: The speaker estimates that conference software costs represent about 10% of his budget and about 10% of his margin that he could otherwise recover.

“... getting back. That is to just somebody else's software that is shitty, slow, uh, ugly, not exactly what we want. And, you know, so I'm paying three hundred thousand, m-maybe I call it paying one million a year in just like SaaS.”

Claim: The speaker says his annual SaaS spending may be as much as $1 million, though he also cites $300,000 in the same account.

“... getting back. That is to just somebody else's software that is shitty, slow, uh, ugly, not exactly what we want. And, you know, so I'm paying three hundred thousand, m-maybe I call it paying one million a year in just like SaaS.”

Claim: The speaker considers the SaaS his company pays for to be poor-quality, slow, unattractive, and not well matched to its needs.

“... okay, okay, okay. Like, as the boss, like, my employee comes to me and wants to use this thing, and they say it's important for their work. Uh, yeah, I'm gonna say yes. But, like, when it all adds up, like, it's a lot of money.”

Claim: The speaker says he would approve an employee's request to use a work tool they consider important, although the accumulated costs are substantial.

“And then, like-- And then I use it, and they use it. They're like: "Yeah, it doesn't give me what we want." It's like it has-- It's slow, like they request. Like, as a customer of a SaaS, what happens is you are in a queue for support, right? You want this feature, and they're like, "Okay-"”

Claim: The speaker says SaaS customers who request support or features may have to wait in a queue.

“"... six months from now, you know, we'll do-- we'll put you on the Q3 roadmap."”

Claim: The speaker describes SaaS providers as potentially delaying a requested feature for six months and placing it on a later quarterly roadmap.

“"You are very important to us." Like it's, you know, we are AI conference company, and this is twenty thirty-six, and there's coding agents, right? Like, we can do this. So, uh, the idea was I took a forty thousand dollar SaaS that my employee was proposing to pay, and I said, "I will pay somebody ten thousand dollars to clone this whole thing."”

Claim: The speaker says that, in the scenario he describes as 2036, coding agents enable his AI conference company to build software itself.

“Like, we can do this. So, uh, the idea was I took a forty thousand dollar SaaS that my employee was proposing to pay, and I said, "I will pay somebody ten thousand dollars to clone this whole thing."”

Claim: The speaker offered $10,000 to have someone clone a SaaS product that an employee had proposed purchasing for $40,000.

“It's open source, and it's exactly what we want. And from here we can use whatever coding agent, Codex, Claude, Coding, uh, Devin, to just customize it, whatever, right?”

Claim: The speaker says the open-source clone matched his company's needs and could be customized using coding agents such as Codex, Claude, or Devin.

“And we did it. We, you know, announced the competition. We had sixty-nine submissions. When you have a competition like that and you offer free money-”

Claim: The speaker says the competition received 69 submissions.

“... uh, I, I've offered to cover everybody's c-coding agent subscription, right? So you have a lot of bad submissions.”

Claim: The speaker says that offering to cover all participants' coding-agent subscriptions resulted in many poor submissions.

“And so I just like, this is obviously not, like, I didn't, I didn't give money for that. But, um, we have sixty-nine good ones, and most of them were, like, usable.”

Claim: The speaker says 69 submissions were good and that most of them were usable.

“About five, I would say that we would pay for. Like, it was just, like, good. Like, it was done- ... in one week.”

Claim: The speaker estimates that about five submissions were good enough to pay for and were completed in one week.

“As a owner of a SaaS business, you have maybe like fifty customers, all of them paying you forty K. Imagine if all of them realize this, and then they just- They just, they just turn away. In one year, your, your business is gone. You went from earning twenty million dollars a year to nothing.”

Claim: The speaker warns that if customers of a SaaS business replicate its product themselves, the business could lose its $20 million annual revenue within one year.

“Yeah. That's, uh... It's very scary, I think, uh, for a, for a business owner, and I think that's the tricky thing of, of owning a SaaS, right?”

Claim: The speaker considers the risks of owning a SaaS business very frightening for business owners.

“Like, like I own a physical business, so that's not going away anytime soon. Robots are cool, but they're not- ... they're not organizing conference.”

Claim: The speaker expects his physical conference business to remain viable for the foreseeable future because robots do not organize conferences.

“But, like, people who do software B2B, who own this like mid-tier SaaS, Salesforce is fine, like nobody's worried about Salesforce. But like the mid-tier SaaS where like, you know, a lot of small, mid s- SMB, do you guys call them SMBs? SMBs, they're gonna really struggle.”

Claim: The speaker believes Salesforce is not at risk in the way that mid-tier SaaS providers serving small and medium-sized businesses are, and predicts those mid-tier providers will struggle.

“Uh, or they are gonna pivot to something that's better. And I think they, they really better figure it out now because all of us business owners who are their customers- ... we also can code.”

Claim: The speaker says mid-tier SaaS providers should quickly find a better offering or pivot because their business-owner customers can now code as well.

“And it doesn't take that much long time, you know?”

Claim: The speaker says business owners can code without it taking very long.

“And by the way, the models are very good at clicking through every screen of your software and then replicating it in the i-independently.”

Claim: The speaker says models can effectively navigate a software product's screens and independently reproduce the product.

“Right? So the moment I have access to your software, I can clone it. So good luck. Um, and so I think that is the, the reality. I would say that if you're a SaaS owner, nobody knows the, the right answer.”

Claim: The speaker says there is no known right answer for SaaS owners facing the current situation.

“Right? So the moment I have access to your software, I can clone it. So good luck. Um, and so I think that is the, the reality. I would say that if you're a SaaS owner, nobody knows the, the right answer.”

Claim: The speaker says access to a software product enables him to clone it.

“I think that you basically e-end up having to pivot to being an agent builder. So you have to become like the US equivalent is like Harvey or, um- ... Open Evidence or... Yeah, yeah. You, you own a vertical, and you're the top agent in that vertical. Or you own some specialized data, and you're basically a data vendor.”

Claim: The speaker thinks SaaS owners will generally need to pivot into agent building, for example by owning a vertical and providing its leading agent.

“... Open Evidence or... Yeah, yeah. You, you own a vertical, and you're the top agent in that vertical. Or you own some specialized data, and you're basically a data vendor.”

Claim: The speaker says SaaS owners could pivot to owning specialized data and becoming data vendors.

“... Open Evidence or... Yeah, yeah. You, you own a vertical, and you're the top agent in that vertical. Or you own some specialized data, and you're basically a data vendor. Anything in between, someone else is gonna just code their own thing.”

Claim: The speaker predicts that businesses outside the positions of owning a vertical-leading agent or specialized data will have their offerings replicated by someone else.

“I'm more interested, like, so I'm not, I'm not so sympathetic to the, the, the SaaS owners, right? SaaS owners, they, they're big, big boys. They can figure it out. Uh, actually, I'm more, I'm more keen on advising the business owners.”

Claim: The speaker is more interested in advising business owners than SaaS owners, whom he believes can figure things out themselves.

“Yeah. It's-- I, I run a small business. I'm an SMB.”

Claim: The speaker runs a small business and identifies as a small or medium-sized business (SMB).

“There's a million people like me.”

Claim: The speaker estimates that there are a million people like him.

“They're all trying to just... I, I, you know, I make an honest living. I pay employees, I pay my taxes, and it's a good life.”

Claim: The speaker says he earns an honest living, pays his employees and taxes, and considers his life good.

“It can be better if you vibe code your own SaaS.”

Claim: The speaker believes business owners can improve their situation by vibe-coding their own SaaS.

“And, like, my customers are happy. My employees are happy. I'm happy. It's a cool story to tell. Uh, yeah, you know, SMBs, I think it's a golden age for us.”

Claim: The speaker says his customers, employees, and he are happy, and he believes small and medium-sized businesses are in a golden age.

“If you do anything in the real world that is you're just trying to do your job, and there's a lot of shitty software between you and your, your, your intended outcome, you can suddenly shorten that timeline.”

Claim: The speaker says that people doing real-world work can suddenly shorten the time to their intended outcome when bad software stands between them and that outcome.

“Yeah, yeah. I mean, like, you know, forty thousand dollars worth of Claude is a lot of Claude.”

Claim: The speaker considers $40,000 worth of Claude usage to be a large amount.

“I can spend- ... ten thousand and, like, have a lot of money to like- ... do whatever I want.”

Claim: The speaker says that spending $10,000 would still leave him with substantial money and freedom to do what he wants.

“Uh, so I think, uh, uh, that's why I, I s- I call this is a business owner problem, not a developer problem. A developer problem would say like: "Ugh, two hundred dollars is a lot. Ugh."”

Claim: The speaker characterizes the issue as a business-owner problem rather than a developer problem, suggesting a developer might consider $200 expensive.

“You know, like, like I'm a business owner. I understand the productivity benefits of having ten of my employees work together well.”

Claim: The speaker believes that having ten of his employees work together effectively produces productivity benefits.

“Like ten of my employees, I'm paying one point five million for them, right?”

Claim: The speaker says that employing ten people costs him $1.5 million.

“Like, if, if they work together 10% better, that's 150K. The potentially... It's not, it's not 10% better. It's like five times better. Right? That's like seven point five million.”

Claim: The speaker estimates that a 10% productivity improvement among his ten employees, whose labor costs $1.5 million, would be worth $150,000; he then asserts that the improvement is fivefold, worth $7.5 million.

“Great. And that is, uh, all the agent work that is being done on, on my behalf that people used to, uh, say like, "Well, okay, I'll, you know, I'll come back next week with a proposal." No, I want it this afternoon.”

Claim: The speaker says agents perform work on his behalf that previously would have involved waiting until the next week for a proposal, whereas he now wants the proposal that afternoon.

“Yeah, it's, it's real, right? It's not zero. It's not zero, but we have coding agents and-”

Claim: The speaker says maintaining custom software requires some effort, but coding agents can help with that maintenance.

“Yeah. And a lot of it is... This is why I'm saying like mid-tier SaaS. You are not Salesforce. You are not- Google. You're not AWS. You're not this, like, mission-critical ninety-nine point nine nine nine- ... uptime thing, right? You're just like a database where people type in their thing and they get something out.”

Claim: The speaker distinguishes mid-tier SaaS from Salesforce, Google, and AWS, saying it may not require mission-critical 99.999% uptime and may instead be a simple input-output database.

“So, uh, yeah, okay, it goes down for a day, like, whatever, and you can fix it in a day.”

Claim: The speaker says that a simple mid-tier SaaS product can tolerate a day of downtime and be fixed within a day.

“And there's a lot of coding agents where... Like, I, I think I, I would argue I would take that downtime for in exchange for, for- I can customize.”

Claim: The speaker would accept downtime in a software product in exchange for the ability to customize it.

“I can customize anything. It's amazing, you know. And, and so my team is so much more productive. You know, uh, uh, I actually put this in the episode in Kill My SaaS.”

Claim: The speaker says he can customize anything and that this has made his team much more productive.

“One of the favorite features, let's say I have, uh... So for example, for AIU Code- ... my, my next conference next month, I have one thousand submissions, and I have twenty-five seats. Okay? I have to go through one thousand people to get twenty-five, and in the current SaaS that we use for that piece of software, we have to click one by one by one by one by one.”

Claim: For a conference scheduled for the following month, the speaker has 1,000 submissions for 25 seats, and the SaaS he currently uses requires reviewing each submission individually.

“Every screen, click, click, click, click, click, click, click. And so in the Kill My SaaS competition, one of our contestants just changed everything so that you can sort of reorder it in one table. This view doesn't exist in that SaaS.”

Claim: A contestant in the Kill My SaaS competition created a single-table view for reordering submissions that the existing SaaS did not provide.

“To... In order to get this view, we have to request for six months. Well, but-”

Claim: The speaker says that obtaining the single-table view through the existing SaaS would have required a six-month request.

“... But you just, you just request whatever you want. And like, "Ah, now I can do it and I can see everything in one go."”

Claim: The speaker says users can request custom functionality and then view all the relevant information at once.

“I, I am, uh, for listen-- You know, if you're thinking that this is a stupid example, it is. Uh, it, I'm- ... simplifying for audio.”

Claim: The speaker describes his example as deliberately simplified for audio and acknowledges that it is a stupid example.

“Yeah, and it's good, it's good UX. They are testing it- ... uh, from the point of, point of view of humans. And I think that there is a lot of difference between the average submission and the top submission.”

Claim: The speaker says the winning submission had good user experience and was being tested from a human-centered perspective.

“... uh, from the point of, point of view of humans. And I think that there is a lot of difference between the average submission and the top submission.”

Claim: The speaker believes that the quality of an average submission differs substantially from that of a top submission.

“So, uh, that is true that today, in twenty twenty-six, if you're a business owner and you decide to hire the youngest kid with, like, you know, that looks vaguely technical to, "Hey, vibe code this SaaS for me, please," maybe they're not gonna do so well because they don't have any training.”

Claim: The speaker says that in 2026, a business owner might get poor results by hiring a vaguely technical young person without training to vibe-code a SaaS.

“But just give it one year. Watch a lot of AIU videos.”

Claim: The speaker advises allowing one year and watching many AIU videos.

“There's a skill level, right?”

Claim: The speaker says that producing good results with this approach requires skill.

“Like, uh, a, a lot of it is slop, and you don't want to use slop, right? Your, your employees deserve better. You deserve better. And actually, sometimes buying from a tried and tested SaaS is going to be better.”

Claim: The speaker says much AI-generated software is low quality and should not be used, and that established SaaS is sometimes the better choice because employees and business owners deserve better.

“It is going to be better.”

Claim: The speaker asserts that established, tested SaaS can be the better option.

“But for us, uh, we want to be AI forward as a company. We will be more cutting edge.”

Claim: The speaker says his company intends to adopt AI proactively and become more cutting edge.

“Oh, so that was, uh, partially on purpose and partially not the goal of the judge. The judge was purely, did you submit a valid submission, right? Did you qualify?”

Claim: The speaker says the AI judge's purpose was to determine whether a submission was valid and qualified.

“But the judge was not aimed for, is this the best among the whole cohort?”

Claim: The speaker says the AI judge was not designed to select the best submission among all entries in the cohort.

“Because we didn't know what we wanted.”

Claim: The speaker says his team did not know what it wanted from the judging process.

“So I cannot make an AI judge for, for things-”

Claim: The speaker says he cannot make an AI judge for unspecified kinds of things.

“Ninety-Nine Designs, you put up a contest, and I got four hundred different ideas- ... for a logo.”

Claim: The speaker says he used Ninety-Nine Designs to run a logo contest that produced 400 different logo ideas.

“And then I award the prize to the, the logo that we got, which is, um- ... it's a very good logo.”

Claim: The speaker considers the logo selected through the Ninety-Nine Designs contest to be very good.

“It's so much better, right? You want diversity in ideas. So actually, this idea of like, do contests and-”

Claim: The speaker believes design contests produce better results by providing a diversity of ideas.

“Most people cannot do because they don't have the audience.”

Claim: The speaker says most people cannot run design contests because they lack an audience.

“But I think that there is such a thing... But Ninety-Nine Designs is a very good business because there's many designers. It actually doesn't take that much to create a design, right? You just need-- It's more about the quality of the idea than the amount of work. So you just need high-quality ideas, and it, and, like, you just need, uh, to be able to-”

Claim: The speaker believes design creation requires relatively little work and that idea quality matters more than the amount of work.

“But I think that there is such a thing... But Ninety-Nine Designs is a very good business because there's many designers. It actually doesn't take that much to create a design, right?”

Claim: The speaker considers Ninety-Nine Designs a very good business because it brings together many designers.

“... over to a lot of different contests. And so Ninety-Nine Designs is like that, and I think there could be a lot of Ninety-Nine Designs for SaaS, which is something I'm thinking about.”

Claim: The speaker thinks there could be many contest-based services modeled on Ninety-Nine Designs for SaaS products.

“You know what I mean? Like, basically, I had to create a Ninety-Nine Designs for myself for saying, "Here's the forty thousand dollars SaaS. Here's the specs. Here's the judging criteria. Here's the blah, blah, blah," even right down to like, "Here's how we pay you money when you win." Everything had to be worked out by me. But would I pay some other person to do all this for me so that I just tell them what I want-”

Claim: The speaker says he personally organized a $40,000 SaaS contest, defining its specifications, judging criteria, and how the winner would be paid.

“It's a lot of work. We took, like, two months to eval everything, partially because I'm busy, but, uh, also it's just, it's a lot of eval.”

Claim: The speaker says evaluating the SaaS project took about two months, partly because he was busy and partly because the evaluation itself required substantial work.

“Like, uh, just there's five roles for this particular SaaS because it's about, like, organizer, speaker, attendee, sponsor- ... vendor.”

Claim: The speaker says the SaaS supports five roles: organizer, speaker, attendee, sponsor, and vendor.

“They all have to agree on what the conference organization looks like.”

Claim: The speaker says the organizer, speaker, attendee, sponsor, and vendor roles must agree on the conference's organization.

“So you have to create an account for each of the-- In order to test, you have to create account for each of the things, and you have to do some data entry here, and you change accounts and do some-- and you read and data entry here, and you change, change, change, change, change. And so that's why it's, like, complicated.”

Claim: The speaker says testing the multi-role SaaS is complicated because it requires separate accounts, data entry, and repeated switching between accounts.

“That's why I pay forty thousand and fifty dollars for this- ... 'cause I don't wanna do it. But, uh, now that you can, it's very fun job and problem to do.”

Claim: The speaker says he paid $40,050 for the SaaS because he did not want to do the work himself.

“But you know what I mean. Like, like the booking link, like that's how we, that's how we got to this meeting. Yeah, Calendly is like a thirteen dol- thirteen billion dollar business.”

Claim: The speaker describes Calendly as a $13 billion business.

“I, I have, I have my own. I have, I have two versions. Yes. Yes, yes, yeah. I, I have my own Calendly. I have my own Google Drive. I have my own GitHub. I'm serious. I have my own Dropbox. I, I have my own Slack. The-- All of this is because I have money and I can... I, I have people and I can, I can assign them to do things. But I do think that you can build software for yourself for audience of one or audience of one company-”

Claim: The speaker says he has built or maintains personal versions of Calendly, Google Drive, GitHub, Dropbox, and Slack.

“I, I have my own Slack. The-- All of this is because I have money and I can... I, I have people and I can, I can assign them to do things. But I do think that you can build software for yourself for audience of one or audience of one company-”

Claim: The speaker says his ability to create personal software versions is enabled by having money and people he can assign to the work.

“I, I have my own Slack. The-- All of this is because I have money and I can... I, I have people and I can, I can assign them to do things. But I do think that you can build software for yourself for audience of one or audience of one company- ... and that's okay.”

Claim: The speaker believes it is acceptable to build software for an individual user or a single company.

“And the, the, the fact that you can do that now is something that has never existed before for any business owner- ... which is amazing.”

Claim: The speaker claims that business owners have never before had the ability to build software for themselves or for a single company.

“Ten billion dollar company. Or, uh, maybe seven now.”

Claim: The speaker estimates DocuSign's value at $10 billion, then qualifies that it may be about $7 billion now.

“They call it en- You know. And like, yeah, I, I... There's an open source DocuSign and I host it for myself. And you know what's the difference? Like because I have my own drive and I have my own DocuSign, the moment that the DocuSign is signed, it goes straight into the correct folder on the drive.”

Claim: The speaker says he self-hosts an open-source DocuSign alternative that automatically places signed documents in the correct folder in his personal drive.

“Yes. And that does not exist in the DocuSign.”

Claim: The speaker says DocuSign does not provide the integration that automatically files signed documents into the correct drive folder.

“No. Electronic signatures are legally enforceable everywhere.”

Claim: The speaker asserts that electronic signatures are legally enforceable everywhere.

“... it, it is so stupid. And, like, I can keep going about why I made each of, each of those things. It doesn't matter. It's not like I'm encouraging everyone to do this. I think that whatever problem you face in your business or your life, you really can find open source version of it- ... host it for yourself, customize it, integrate. Okay, I want this plus this. Boom. They just work 'cause I fucking use them all the time. I copy here, paste there, copy here, paste there.”

Claim: The speaker believes that people can find an open-source solution for almost any business or personal problem, then host, customize, and integrate it themselves.

“Inside of every one million copy-paste, I think there is a fantastic startup and product to be built.”

Claim: The speaker believes that repeated copy-and-paste workflows can reveal opportunities to build excellent startups and products.

“Yeah. So, uh, for the those who didn't listen to the podcast, we were the first podcast with Jeff.”

Claim: The speaker says his podcast was the first to feature Jeff.

“Because, uh, I, I forced him because I'm- ... I've friends with him for four years, and I say, "If you don't-- If you don't give me the first podcast, I'm gonna kill you." No, he's, he's, he's, he's obviously like very, very, uh, accomplished.”

Claim: The speaker says he had been friends with Jeff for four years and pressured Jeff to make his podcast the first to interview him.

“He was the co-author on, on, uh, the InstructGPT, which became ChatGPT. And I think the interesting thing about Jeff is that it is the kind of the original vision of AI engineering for me, which is that instead of right now, AI engineering is, is what we talked about, is a big harness-”

Claim: Jeff co-authored InstructGPT, which later became ChatGPT.

“So if you think about what is the failure mode of, of what the LM, current LM paradigm is, is that you have the big traditional SaaS and then your LM is this window that is a chatbot.”

Claim: The speaker describes a failure mode of the current language-model paradigm as adding a chatbot window to traditional SaaS software.

“... menu here. That's it. For what Jeff represents and what system one models represent, this is why he doesn't call it decision model, he calls it system one models. It is that they are in the path of code. You write it as casually as you write an if statement inside of a normal programming language. And I think that is really what he means by smarter software, which I really, really endorse, which is like your software should just be at every point smarter instead of at one point smart and then everything else is dumb.”

Claim: The speaker says system-one models can be embedded in code paths and invoked as casually as ordinary if statements.

“It is that they are in the path of code. You write it as casually as you write an if statement inside of a normal programming language. And I think that is really what he means by smarter software, which I really, really endorse, which is like your software should just be at every point smarter instead of at one point smart and then everything else is dumb.”

Claim: The speaker endorses making software intelligent throughout its workflow instead of concentrating intelligence in a single feature while leaving the rest unintelligent.

“Yeah. I think the, the speed stuff is almost secondary to the reliability, which is what he really is trying to say. That's why the company is called TypeSafe.”

Claim: The speaker believes reliability matters more than speed in Jeff's approach and says this priority explains the company's name, TypeSafe.

“He has a non-traditional background himself. He didn't go to, like, Stanford PhD and, you know, the kind-- those kind of things. So he has that sympathy with the regular developer, but he also worked on ChatGPT.”

Claim: The speaker says Jeff's non-traditional background gives him sympathy for ordinary developers, while his work on ChatGPT gives him experience with a major AI product.

“And I think that a lot of people who maybe they went through the traditional ML researcher path, you go to a top university, and then you work on some interesting paper, and then you get hired into a lab, and you work on a lab. They're in the big models cult. Everything will be solved- ... by making our model bigger, by having more ch- more flops in our training run and more parameters and more, more, more, more, more.”

Claim: The speaker says many traditionally trained machine-learning researchers believe that scaling models through more compute and parameters will solve everything.

“Right? Um, and so it takes someone who doesn't come from that mold to really say, "Well, actually, there's another way to do this," which is smaller stuff that is sprinkled everywhere instead of one big one.”

Claim: The speaker presents an alternative to one large model: distribute smaller intelligent components throughout software.

“And I think that is-- By the way, Diego was not the first to do that. There's been thousands of people who've tried to say, "Smaller model's good."”

Claim: The speaker says thousands of people had advocated for smaller models before Diego.

“One, because he has the credibility of OpenAI. Two, he made it into a system one versus system two thing, which I think is-- makes the communication very clear in-- to the point where you can play Doom or whatever it was.”

Claim: The speaker attributes part of Diego's success in communicating his ideas to his credibility from OpenAI.

“One, because he has the credibility of OpenAI. Two, he made it into a system one versus system two thing, which I think is-- makes the communication very clear in-- to the point where you can play Doom or whatever it was.”

Claim: The speaker believes Diego's system-one versus system-two framing made his ideas clearer and easier to communicate.

“What-- There's some game demo that he, that he showed in the launch post that was very clear, like, there's no way you'll ever use LLM to do this.”

Claim: The speaker says a game demo in Diego's launch post clearly illustrated a task for which an LLM would not be usable.

“And I think that showmanship, that he wears the pink jacket, he, he's so passionate, and he's talking. That showmanship plus the technical capability is, again, like, similar to the Learn in Public.”

Claim: The speaker sees Diego's combination of showmanship and technical ability as similar to the practice of learning in public.

“Like, I'm a small version of that, right? Like, I'm technically capable, and I can talk. But for him, it-- he's like ten X on both dimensions, and I think that really comes through in when you talk to him.”

Claim: The speaker says he is technically capable and communicative, while Diego is ten times stronger than him in both technical ability and communication.

“Oh, yeah. So we are actually a community that started three years ago that we found out a lot of, uh, Chinese-based founders want to go global.”

Claim: The speaker's community began three years ago after its organizers found that many China-based founders wanted to expand globally.

“And then, uh, I think w- uh, they were pretty good at building, uh, or shipping products quickly, but they found a huge gap between, uh, you know, go-to-market marketing and understanding- ... the market.”

Claim: The speaker says China-based founders seeking to go global were good at building and shipping products quickly but often lacked go-to-market marketing and market understanding.

“Yeah. So that's basically we hold a lot of meetups and then working with some advisors here- ... and then try to provide different kinds of workshops and advisories to those founders really start to find a little bit their, you know, on the right track of PMF but still lack some talents and, uh, you know, help from, from the go-to-market side.”

Claim: The speaker's community holds meetups and works with local advisors to offer workshops and advisory support to founders who are approaching product-market fit but lack go-to-market expertise.

“Uh, we hold this, like, sort of, um, writers meetup thing where people can show their work, and I will give commentary on it, on it. Um, and so the, the sort of typesafe manifesto came out of that.”

Claim: The speaker holds a writers' meetup where participants present their work for his commentary; the TypeSafe manifesto emerged from the meetup.

“But I, I think it's very similar to, to that, except that I would-- for some pe- some of them, I'll do it for equity. Most of it I do it for free, and then some of them, they want to have a, like a recurring call, um-”

Claim: The speaker provides most of his startup advising for free and accepts equity for some engagements.

“... or they or something, they, then, then they offer some equity. It's just easier to do that, and it turns out it's worth a lot of money if it-- if the company does very well.”

Claim: Equity received for advising can become very valuable if the company performs very well.

“So I think that is interesting in terms of, like, a job because it's equ-- you cannot eat equity. And like, maybe they never sell.”

Claim: Equity compensation cannot directly pay for living expenses, and the company may never be sold.

“Like, Databricks is like, whatever, fifteen years old and worth two hundred billion dollars, and it's like, I have my equity. It's like, what is it? But, um, I think it's, it's very useful to be able to test your theories on a number of different companies.”

Claim: Advising multiple companies is useful to the speaker because it lets him test his theories across different organizations.

“... advice only works in one company. But now that I have done thirty something, thirty-five, thirty-six different companies- ... I can really see the universal truths among all of them, and I'm, I'm better as an advisor to all of them because I've tried everything.”

Claim: After advising approximately 35–36 companies and trying many approaches, the speaker believes he can identify patterns common across companies and advise them better.

“Yeah. Talked to everyone. I, like-- And I, I understand the different scenarios of like, okay, my advice is not gonna work for you here because I tried it at, at this other company and, and, and, you know, your situation is, like, clearly very different.”

Claim: The speaker says advising across companies helps him recognize when a recommendation from one company is unsuitable for another company's different circumstances.

“And like, I, I, I feel like actually almost all domain subject matter experts, like I happen to own developer marketing, right?”

Claim: The speaker identifies developer marketing as his area of expertise.

“That's my thing. But if you work in, I don't know, you're strategic finance, or you're working in like GPU compute acquisition or whatever, and you n- and you are trying to be a global number one in that, you kind of have to go through a tour of duty of just advising everyone So you can just take-- fucking test everything.”

Claim: The speaker argues that domain experts who aim to become global leaders in their field need to advise many organizations and test their approaches across them.

“Some of it doesn't work, but some, but sometimes it... I mean some of them is clearly going to work.”

Claim: The speaker says that when testing approaches across companies, some approaches fail while others are clearly effective.

“And then you will learn, and that's the only... Like, your sample size has to be more than one, and most people's sample size is, like, five. But, like, you know, if you can get it up to, like, fifty then, like, you probably are one of the top in the field anyway.”

Claim: The speaker argues that expertise requires learning from more than one case; most people have a sample size of about five, while someone who reaches about fifty cases is probably among the top people in the field.

“And I can share with you some, you know, some-something that I find, uh, quite similar in, in those, you know, go global Chinese founders because right now more and more came from engineering background, even research background.”

Claim: The speaker observes that an increasing number of Chinese founders seeking to expand globally have engineering or research backgrounds.

“... they, they think that they're building the best products because they have- ... like, best harness skill or, you know, technical skills and knowledges, et cetera. But actually there's still sometimes a huge gap between, you know, what the customer actually wants that the skills- ... and the product you offer and sometimes, like, different local markets have even have different, you know, demands and have you seen a similar thing-”

Claim: The speaker says technically skilled founders may believe their products are the best, yet their technical capabilities and products can remain misaligned with customer needs.

“... and the product you offer and sometimes, like, different local markets have even have different, you know, demands and have you seen a similar thing-”

Claim: Customer demands can differ across local markets.

“There's a bunch of ways to, to survey this or to tackle this. Uh, the, the most important thing to recognize is that if you are... You, you have something smart and, like, you think it's valuable and- ... global number one, you know, really, like- ... you, you are world best at one thing. Like, that's the most important thing. So everything else is, uh, I always call it 说 话 是 龙 ， 做 事 是 鹰 .”

Claim: The speaker argues that the most important thing is to be world-best in one area.

“Um, I think that there are some percentage of people who don't even have the long, and they think they have the long and, like, then this like, okay, it's like a arrogance equal thing, right?”

Claim: The speaker believes some people mistakenly think they have an exceptional strength they do not actually possess, and associates this with arrogance.

“But let's assume that you are generally have a very good product.”

Claim: For the following discussion, the speaker asks listeners to assume that the product is generally very good.

“And I think the, the important thing actually here is that you should not treat marketing as the last mile of, like, "Hey, I did everything. It is the last night before the blog post. Can you review my blog post and change it?" There's nothing I can do. Like, just you already did everything. But if you involve marketing earlier in the process, right down, down to, like, before it's really built or specced out, like what, what is the way we're gonna pitch?”

Claim: Marketing should be involved before a product is built or fully specified, rather than only at the final blog-post stage.

“Can you review my blog post and change it?" There's nothing I can do. Like, just you already did everything. But if you involve marketing earlier in the process, right down, down to, like, before it's really built or specced out, like what, what is the way we're gonna pitch? What part of the segment of the market we're gonna go? Then you have more GTM built into the product in the first place.”

Claim: Involving marketing before product development helps incorporate go-to-market considerations into the product from the outset.

“And I think that a lot of engineer or, or researcher-defined audiences don't have that GTM brain that early.”

Claim: Many engineer- or researcher-led teams do not bring go-to-market thinking into the process early.

“They usually put it all the way at the end. They're-”

Claim: Engineer- or researcher-led teams usually defer go-to-market work until the end of the process.

“Uh, and so, uh, very simple things most en-researcher engineer founders have not thought about this, but if they have a GTM partner, I think that that is often very useful.”

Claim: Having a go-to-market partner is often useful for engineer- or researcher-founders.

“Um, and robotics is also a very interesting space because it is much longer time horizon and still very early days, right?”

Claim: Robotics is an interesting field with a long time horizon that remains at an early stage.

“... I would like them to see things that you never expected Unitree to ever do. But right now, you know, the, the, the thing that they just launched this week, they're doing kung fu moves.”

Claim: The speaker wants Unitree to demonstrate unexpected capabilities beyond familiar demonstrations such as kung fu moves.

“But like, okay, I've seen that before. You know, like, show me something that I haven't seen and then, then I'm really going to pay attention.”

Claim: The speaker is more likely to pay attention to a robotics demonstration that shows something novel rather than a familiar demonstration.

“You know, I think that's, that's something that, uh, is hard to show in robotics, but, um, I think with models you can do that still 'cause it's still so early days.”

Claim: Although novel demonstrations are hard to produce in robotics, the speaker thinks models can still enable them because the field is at an early stage.

“And with, uh, innovative, like, uh, product ideas or for, like, for consumer experiences, you can do that. So I think to show the demo, I think is very important, which is why, again, obviously Typesafe is, like, very recent-”

Claim: Innovative product ideas and consumer experiences can be presented through novel demonstrations.

“And with, uh, innovative, like, uh, product ideas or for, like, for consumer experiences, you can do that. So I think to show the demo, I think is very important, which is why, again, obviously Typesafe is, like, very recent-”

Claim: Demonstrations are very important for communicating a product or idea.

“... but also very good. It is not the... Diogo did not come out with the Doom demo, but the Doom demo is what everyone talks about.”

Claim: The Doom demo is the demonstration that receives broad attention in this example.

“... but also very good. It is not the... Diogo did not come out with the Doom demo, but the Doom demo is what everyone talks about. Doom demo was done by DevRel, like GTM, and that, that is, that is something that just is a core principle in marketing, which is show don't tell, right?”

Claim: The Doom demo was created by developer relations, which the speaker describes as part of go-to-market.

“Doom demo was done by DevRel, like GTM, and that, that is, that is something that just is a core principle in marketing, which is show don't tell, right? Like, show me this thing having the benefits- ... but don't tell me in bullet points what this is.”

Claim: A core marketing principle is to demonstrate a product rather than merely describe it.

“Okay. I can define taste in a way that is not helpful. Taste is something where you are able to... O-out of a large set of possibilities, you're able to narrow it down quickly, cheaply- ... to the set of possibilities that when you look back, it's immediately obvious it's good. So-”

Claim: Taste is the ability to quickly and cheaply narrow many possibilities to a choice that appears obviously good in retrospect.

“The brute for- you can brute force. You can try one thousand of them, and then you find a good one. But a tasteful person can say, "No, obviously these are bad. I, I don't... I cannot explain it. I cannot explain it, but obviously these are bad. This one looks in-- This, this tool looks interesting."”

Claim: A person with taste can intuitively reject many bad options and recognize an interesting one, even without being able to explain the judgment.

“... yeah. What do you mean we're considering the other nine hundred and ninety-nine? Obviously, this was the correct one. There's many versions, applications of this, but, um, the one that's coming to mind right now is the iPhone, where, uh, for a long time the s- the, the smartphone had a keyboard. And it took Steve Jobs to say, "No We will do whatever it takes to have touch screen and virtualized screen because it look-- needs to look like a screen that's a flat slab.”

Claim: The speaker presents Steve Jobs’s choice of a touchscreen, virtualized display, and flat-slab design for the iPhone instead of a keyboard as an example of taste.

“And it, that took ten years to R&D 'cause there was a company called General Magic that tried to do this and failed. And but, like, obviously now every phone in the world looks like this.”

Claim: The speaker says that phones worldwide now share the flat touchscreen design being discussed.

“And it, that took ten years to R&D 'cause there was a company called General Magic that tried to do this and failed. And but, like, obviously now every phone in the world looks like this.”

Claim: The touchscreen smartphone approach took ten years of research and development, after General Magic had attempted it and failed.

“That is, that is taste. In 1990, when people were starting to have smartphones, you could do a number-- a huge number of things.”

Claim: The speaker says that around 1990, as smartphones were emerging, there were many possible directions for their design and use.

“But it takes the, the person with taste to say like, "Yeah, this is what we want, and I'm gonna say no to everything else." And, uh, my first phone I shipped was gonna be this.”

Claim: Taste involves selecting a direction from many possibilities and rejecting the alternatives.

“I actually think that cur-- everything that Cursor does has a lot of taste because they were first to the VS Code thing, they first becau- and now they're, they're doing GroqBot-”

Claim: The speaker considers Cursor’s work to demonstrate strong taste.

“... which is, like, a g-good tasting. Be-because, like, now, now, like, O-OpenAI has to, like, copy them and what have you. And, uh, Claude has to copy them.”

Claim: The speaker says OpenAI and Claude have to imitate Cursor’s work.

“Which, like, you know you have taste when other people copy the thing that you just shipped.”

Claim: Being copied by others after shipping a product is an indicator of taste, according to the speaker.

“The interesting way, you know, to go below the labs, I think it's interesting to think about, like, the more boring kind of problems that people solve.”

Claim: The speaker finds it worthwhile to examine less prominent, practical problems solved outside AI labs.

“Uh, Lagora, you know, in, in the legal field- ... they, uh, they, they have this sort of, like, document review process that's, like, massively parallel.”

Claim: Lagora operates in the legal field and has a highly parallel document-review process.

“And, like, again, like, anyone could have come up with it. Anyone. Y-Ideas are cheap.”

Claim: The speaker says the idea behind Lagora was not unique and that anyone could have come up with it.

“Lagora did it. And so someone at Lagora has that taste of, like, we will define this billion possible things we can do. We will just define these products that, that becomes the way that people do their work.”

Claim: The speaker says Lagora created the product and that someone at Lagora had the taste to define products that could become how people do their work.

“I It's tough because it's more of a differential for me. I want to see how very different Shanghai is from, uh, San Francisco because I think San Francisco has a B2B SaaS bubble.”

Claim: The speaker wants to compare Shanghai with San Francisco and believes San Francisco has a B2B SaaS bubble.

“But Shanghai, uh, you, you guys use Comfy UI? We don't use it here. We like our Canva. We like our Fal.ai.”

Claim: The speaker says their community uses Canva and Fal.ai but does not use ComfyUI.

“Uh, don't use Comf- don't use Comfy. Like, Comfy is more of like a open source tool. And it's same, same for OpenClau. OpenClau had more traction here than Comfy. But OpenClau is really traction”

Claim: The speaker characterizes Comfy as an open-source tool and says OpenClau has gained more traction than Comfy in their setting.

“Yeah, yeah, of course I know. But, like, I haven't seen that kind of energy myself.”

Claim: Although the speaker knows OpenClau has traction in China, they have not personally witnessed that level of energy.

“And I think it's very different. Like, the, you know, for you, a medium-sized community is one million people. That's huge in here.”

Claim: The speaker says that a community of one million people is considered medium-sized by the other person but is huge in the speaker's setting.

“No, I-- like, I've seen those kind of like, you know, Chinese, like, influencer farms. Like, it's, like, all these Chinese girls, like, doing this and the, the phone. I-It's-- like, the scale in China is just so different and it's so competitive, right?”

Claim: The speaker says China's scale is very different and its environment is highly competitive.

“Like, people say they do nine-nine-six in, in San Francisco. They don't really. Like, Saturday, Sunday, they're still kinda chilling. But in China, they are really nine-nine-six.”

Claim: The speaker contrasts San Francisco with China, saying that San Francisco workers who claim to work 996 still tend to relax on weekends, whereas workers in China genuinely work 996.

“Yeah, but I think that, I think that's really intense. People don't, like, Toutiao and, like, Meituan and- Intense.”

Claim: The speaker considers China's work culture very intense.

“Right? And, like, people here don't work that hard.”

Claim: The speaker says people in their setting work less hard than people in China.

“I work that hard 'cause I have the most equity. But, like, employees- ... legally you cannot ask people to work that hard here. It's illegal. Okay?”

Claim: The speaker says they work very hard because they hold the most equity, while employers in their setting legally cannot require employees to work equally hard.

“Yeah, I've seen some of the most competitive top-tier- ... US startups, and I'm sure they're not working as hard as China. There's BAT, there's, like, the first generation, then there's, like, second-”

Claim: The speaker believes top-tier US startups do not work as hard as their Chinese counterparts.

“They're such good founders.”

Claim: The speaker considers Chinese founders exceptionally good.

“They're such good f- like, if they, if they live in the wrong country, if they lived in the US, they would kill, they'll kill Amazon, they'll kill, like, Facebook, right?”

Claim: The speaker claims that if the founders they are discussing lived in the United States, they would be capable of defeating Amazon and Facebook competitively.

“Yeah. I mean, even, even, like... Okay, so when I did my study abroad in Beida in 2007, BYD was still like, "Oh, we're, we're up and coming," like, you know. "We're, we're Chinese-based. We'll always be Chinese," right? At least-”

Claim: The speaker recalls that during their 2007 study abroad at Beida, BYD was described as an up-and-coming company that would remain based in China.

“People love BYD here. BYD is taking over the US.”

Claim: The speaker says people in their setting are enthusiastic about BYD and asserts that BYD is taking over the United States.

“I think that the Chinese energy and spirit and creativity and grit is something that people don't have exposure to.”

Claim: The speaker believes people outside China have limited exposure to Chinese energy, spirit, creativity, and grit.

“And then don't-- let's not even talking about the model labs, m-which obviously, like, DeepSeek and, um, all those, all those labs. So I think there's many reasons to be excited about exposing that more.”

Claim: The speaker sees many reasons to increase exposure to Chinese model labs such as DeepSeek.

“Uh, especially, you know, I think Chinese events, they're very... There's regulations against live streaming, but even they're presented differently than the way that Western audiences like to consume them, so they don't have the same reach.”

Claim: The speaker says Chinese events face live-streaming regulations and are presented in formats less suited to Western audiences, limiting their reach.

“Right? Ali did, like, Ali Yun did, like, this thing where they presented the whole Alibaba roadmap with the QAN and everything. Nobody saw it here. Why? Because it was in the middle of some corporate presentation presented by other people, so I had to go and translate everything- ... for, for the US audience.”

Claim: The speaker says Alibaba's roadmap presentation received no attention in their setting because it was buried in a corporate presentation delivered by others, requiring the speaker to translate it for US audiences.

“Yeah. So I mean, uh, we, we hope to have representation from everybody. Um, and I, I think that there's a lot of people interested in, in that cross-cultural exchange that are not interested in the politics.”

Claim: The speaker hopes the event will represent everyone and believes many people want cross-cultural exchange without engaging in politics.

“Like, we will not do politics. We'll only do tech.”

Claim: The speaker says the event will exclude politics and focus only on technology.

“AI developments that most excites me, I think system one models is like a very clear number one answer, so it's boring, so I'm not gonna say it. So I think the thing that people are under the radar about is interaction models.”

Claim: The speaker says system-one models are the AI development that excites them most, while declining to discuss them because they consider that answer boring.

“AI developments that most excites me, I think system one models is like a very clear number one answer, so it's boring, so I'm not gonna say it. So I think the thing that people are under the radar about is interaction models.”

Claim: The speaker believes interaction models are an underappreciated AI development.

“Uh, so this is something that Thinking Machine is doing- ... for people who haven't, who missed the announcement, look it up. It is where you are full duplex back and forth communication at all times.”

Claim: The speaker says Thinking Machine is developing interaction models for continuous full-duplex communication.

“You don't start and then you take a turn, I take a turn, you take a turn. No, at, at like, at all times, like you are listening while you're speaking. And I'm listening to you while I'm speaking, and I can interrupt you, and you will immediately respond and adjust.”

Claim: The speaker describes full-duplex AI interaction as simultaneous listening and speaking, with interruptions prompting immediate responses and adjustments.

“That should be the way of AI, right? There's a corollary of that, which is continual learning AI, where you don't have pre-trained, post-trained done. You are learn, learn, learn, learn, learn, learn as I work with you, and that's just very... It's a very interesting and important field.”

Claim: The speaker describes continual-learning AI as a system that continues learning during use rather than ending its learning after pre-training and post-training.

“That should be the way of AI, right? There's a corollary of that, which is continual learning AI, where you don't have pre-trained, post-trained done.”

Claim: The speaker believes continuous, full-duplex interaction should be the standard way people interact with AI.

“You are learn, learn, learn, learn, learn, learn as I work with you, and that's just very... It's a very interesting and important field.”

Claim: The speaker considers continual learning a highly interesting and important field.

“And then the last part is world models- ... which are obviously doing very well.”

Claim: The speaker says world models are performing very well.

“This is probably the longest AI summer we've ever had in human history, but the history of AI has been, is about fifty years old. There's always summers. Everything's gonna work.”

Claim: The speaker believes the current AI summer is probably the longest in human history, while noting that AI's history is only about fifty years long.

“This is probably the longest AI summer we've ever had in human history, but the history of AI has been, is about fifty years old. There's always summers. Everything's gonna work. Winter, summer, winter. If we ever have a winter, when is it gonna happen, and am I gonna go bankrupt? Which is, is a real concern-”

Claim: The speaker characterizes AI's history as alternating periods of optimism (summers) and downturns (winters).

“Winter, summer, winter. If we ever have a winter, when is it gonna happen, and am I gonna go bankrupt? Which is, is a real concern- ... for a conference company that relies on hype. You know, more, more broadly, I think that there is a real concern about safety that all my friends at all the labs are really, really wo- very worried about, and that society cannot handle and adjust quickly enough.”

Claim: The speaker worries that an AI winter could harm a hype-dependent conference company and potentially bankrupt him.

“... for a conference company that relies on hype. You know, more, more broadly, I think that there is a real concern about safety that all my friends at all the labs are really, really wo- very worried about, and that society cannot handle and adjust quickly enough.”

Claim: The speaker says his friends at AI labs are very worried about AI safety and that society may be unable to adapt quickly enough.

“And it is not because of fake news. It is not because of losing your jobs. It is something worse. Uh, you know, right now the, the way that we express it is cybersecurity.”

Claim: The speaker says the AI-related safety threat is worse than fake news or job loss, and that cybersecurity is currently the way people describe it.

“It's like, oh, I get hacked all the time every- ... every day I just hack, hack, hack, hack, hack. I feel like there's something even worse on the horizon that we don't know about. The most extreme version of this is bioweapons, right? Like oops, like I have invented COVID two.”

Claim: The speaker suspects an unknown threat worse than routine cyberattacks may be approaching, and identifies bioweapons as its most extreme possible form.

“So, so, um, the way that we phrase it in, on the podcast that we do is that the default assumption f-for most people should be AI doom, that, uh, in all of human history when a more advanced civilization comes along and meets a less advanced civilization, the less advanced civilization goes extinct.”

Claim: The speaker argues that most people should treat AI doom as the default assumption.

“So, so, um, the way that we phrase it in, on the podcast that we do is that the default assumption f-for most people should be AI doom, that, uh, in all of human history when a more advanced civilization comes along and meets a less advanced civilization, the less advanced civilization goes extinct.”

Claim: The speaker claims that historically, encounters between more and less advanced civilizations have resulted in the less advanced civilization going extinct.

“And this thing is evolving roughly one thousand times faster than us. And so at some point it's gonna be more advanced, and then we have to figure out what to do there.”

Claim: The speaker says people will need to determine how to respond once AI becomes more advanced than humans.

“And this thing is evolving roughly one thousand times faster than us. And so at some point it's gonna be more advanced, and then we have to figure out what to do there.”

Claim: The speaker claims that AI is evolving roughly 1,000 times faster than humans and will eventually surpass human advancement.

“I have a list. Uh, it's called the Good AI Newsletters and Podcast List. ... yeah, yeah. It, it's on, it's on my GitHub. Uh, but actually the, for the more recent one you can check, uh, aeo.latent.space. So basically what I did was I asked every model what are the best AI podcasts for engineers-”

Claim: The speaker maintains a list of AI newsletters and podcasts and says its recent version is available at aeo.latent.space.

“... yeah, yeah. It, it's on, it's on my GitHub. Uh, but actually the, for the more recent one you can check, uh, aeo.latent.space. So basically what I did was I asked every model what are the best AI podcasts for engineers- ... and it gives me a list, and I transcribe it and put it in a ranking.”

Claim: The speaker queries AI models for the best podcasts for engineers, transcribes their answers, and compiles them into a ranking.

“I ask two hundred of these questions. What is the best payment startup? What is the best bank? What is the best financial advisor? What is the best database? What is... Everything. Every, every, every question that I can-”

Claim: The speaker says he asks AI models 200 questions across categories such as payment startups, banks, financial advisors, and databases.

“But it's probably a pretty decent proxy for what the people who choose the data sets in the, in the labs choose.”

Claim: The speaker believes the models' answers are probably a reasonable proxy for the preferences of people who select datasets at AI labs.

“And they choose Latent Space. It, it does very well. Uh, it's either number one or number... between number one and number three. And, and yeah, so that's, that's the answer.”

Claim: The speaker says Latent Space typically ranks somewhere between first and third in the model-generated results.

“Yeah, of course. I, I hope to listen to more on topics about, like you said, some more infrastructure stuff. For example, like, like the safety, the payments, or the things that come in to unlock more, uh, applications in the world that have come in from different like infrastructure, Asian infrastructure.”

Claim: The speaker hopes AI Shanghai will cover infrastructure topics such as safety, payments, and infrastructure that can unlock more applications.

“I, I, I'm just really curious about those topic because, uh, it's kind of a bit less discussed about in China, but I found out that especially spending few da- few weeks here, I'm finding more people talking about, and also probably RSI or those are frontline researchers.”

Claim: The speaker says AI infrastructure and safety topics are less discussed in China, though after spending a few weeks there he has encountered more people, including frontline researchers, discussing them.

“So, and I hope that it could be a great conversations between the East and the, and the Western, like the researchers here and then the, the, the people in China as well.”

Claim: The speaker hopes AI Shanghai will facilitate productive conversations between Eastern and Western researchers, including researchers in China.

“And I hope it can be a successful one, and then we can carry on for continue to host it on next year.”

Claim: The speaker hopes AI Shanghai will succeed and continue in the following year.

“Uh, you know, I, I wish, I wish the, the team a lot of luck. Uh- ... doing the first one of anything- ... the first one is the hardest.”

Claim: The speaker believes the first iteration of any undertaking is the hardest.

“Yeah. I'm, I'm so impressed that they're doing it.”

Claim: The speaker is impressed that the team is organizing the inaugural event.