Host: Uncle Moon

Guest: Ran Chen

Source: English subtitles. The complete dialogue is grouped into paragraphs, retaining the wording, repetitions, and opening highlights of the supplied subtitles.

Speaker labels are inferred from the Chinese and English subtitles and have not been checked against the audio. A dagger (†) marks uncertain attribution. Timestamps follow the source subtitles for each language; short exchanges within one subtitle may share a timestamp. The English subtitles are consistently 0.2 seconds earlier than the Chinese subtitles, and each edition retains its original timing.

Contents

Opening Highlights

Ran Chen (Guest) (00:00:00)

The risks people face in Silicon Valley are higher than most people can imagine.

Ran Chen (Guest) (00:00:03)

Remember, tomorrow could be even worse. I think the impact of AI will be far more dramatic than people expect.

Ran Chen (Guest) (00:00:11)

Objectively speaking, in the grand sweep of history, jobs haven't been around for very long.

Ran Chen (Guest) (00:00:16)

We don't earn money just to make more money. We earn it to spend where it matters. So I took a year or two off.

Ran Chen (Guest) (00:00:24)

Once someone wins the lottery, it can really mess with their head.

Uncle Moon (Host) (00:00:31)

And Tsinghua's School of Software was a place people didn't want to study!

Ran Chen (Guest) (00:00:34)

When I look for work, I tell people, "I'm here to help you put AI to work."

Introducing Ran Chen

Uncle Moon (Host) (00:00:41)

Interviewing 100 Tsinghua alumni in Silicon Valley. I'm Uncle Moon. Today we're joined by Ran Chen. Tsinghua undergrad, Carnegie Mellon master's. He worked at two startups, both of which were acquired. Then he spent two years as a stay-at-home dad. Now he's the CTO of two startups, while running three zero-person companies on the side, with another ten in the works. Welcome, Ran.

Ran Chen (Guest) (00:01:03)

Thank you, everyone. Thank you, Uncle Moon, for inviting me to this beautiful spot for our conversation. I'm really looking forward to sharing some interesting stories with you.

A Small-Town Test Taker and Standardized Exams

Uncle Moon (Host) (00:01:14)

Let's jump right in. Let's start with your time at Tsinghua.

Ran Chen (Guest) (00:01:18)

Sure.

Uncle Moon (Host) (00:01:18)

On your website, you've always called yourself a small-town test taker.

Ran Chen (Guest) (00:01:22)

Yes.

Uncle Moon (Host) (00:01:22)

But getting into Tsinghua is no small feat. We've interviewed quite a few Tsinghua alumni. Some have photographic memories. Others can read at lightning speed. Some entered college as child prodigies. What do you think got you into Tsinghua?

Ran Chen (Guest) (00:01:33)

First of all, I was lucky. I'm from Yichang, Hubei. Hubei is one of the provinces where the college entrance exam is tougher. As for my own strengths, I'd say there were a few. First, I was pretty good at math. Second, in my last two years of high school, I suddenly realized that the gaokao was a standardized test. There are all kinds of test-taking strategies. I hadn't understood that before. After that, in a way, my grades improved dramatically. Or at least became more consistent.

Ran Chen (Guest) (00:02:06)

Then I got into Tsinghua. That experience really gave me a lot to think about. When I look at something, I want to understand what it really is. I'm quite fixated on that. We need to figure out what we're actually trying to do and how to achieve it. That applies to my career choices, my choices while I was in school, and the jobs I chose afterward. I've always found myself coming back to those questions. I think that started with preparing for the gaokao, or in high school, maybe even middle school. I already had that instinct.

Uncle Moon (Host) (00:02:34)

Could you tell us a bit more about how you came to that realization and what exactly it was?

Ran Chen (Guest) (00:02:40)

Sure. Here's how I see it. When we were studying in high school, in my case, my strengths were pretty uneven. I did well in math. But with subjects like Chinese, I never really understood what I was learning. A big turning point in my life came after talking to some teachers and other people. One teacher explained in detail how Chinese exam papers were graded, especially the essays. Then I watched him grade essays. It was a real eye-opener.

Ran Chen (Guest) (00:03:14)

He actually graded them very quickly. He had certain expectations for an essay. Say you're writing an argumentative essay. To some extent, he expects a certain kind of opening and a certain way of presenting evidence. He might just read the start of an example, then the end, and get a feel for the language in between. What level is the writing at? Is it sophisticated or ordinary? That sort of thing. Then he'd read the conclusion, look at how varied your examples were, and give you a score.

Ran Chen (Guest) (00:03:39)

I hadn't realized that grading itself followed a template. So in my last two years of high school, once that clicked, I adopted a very simple test-taking strategy. I collected about 2,000 Chinese characters of material. The essay requirement was 800 characters. So I collected 2,000 characters' worth and memorized it during morning reading every day. It included some stock openings, some ready-made examples, and other reusable material. From then on, every essay I wrote was built from those 2,000 characters, rearranged into different combinations. And suddenly, my grades stopped fluctuating and became consistent.

Ran Chen (Guest) (00:04:16)

The same idea applied elsewhere. In Chinese class, for example, one thing really stuck with me: reading comprehension. I didn't understand those questions. What exactly was I supposed to comprehend? Then I realized that, in reading comprehension, when they ask you to identify the author's feelings, there might only be about 20 possibilities. Beyond those 20, there aren't any other emotions they expect you to identify. Once I understood that, it suddenly clicked. Every time I did reading comprehension, I could go through the list of 20 answers I'd memorized. Is it the first one? Doesn't seem like it. The second one: homesickness? That doesn't seem right either. Is it a desire to serve the country? Probably not. Go through all 20, and one of them usually fits. That's when I realized that using a few templates could make my grades really consistent.

Ran Chen (Guest) (00:05:09)

That left a deep impression on me. I spent some time understanding what the exam was and what "standardized testing" meant. Then my grades improved and stabilized. Meanwhile, I noticed that many of the students around me would spend a lot of time neatly copying what was on the board and making beautiful notes. But they might not spend much time thinking about what we were actually doing. What was the exam really testing? That was quite a revelation for me. So, yes, when I took the gaokao, with some luck and other factors, I made it into Tsinghua.

Uncle Moon (Host) (00:05:45)

So you went from free-form exploration to figuring out the rules of the game. You turned an open-ended problem into a multiple-choice question. You just had to pick from what you already knew.

Ran Chen (Guest) (00:05:55)

Exactly. And I think at a good school, in a place with better resources, teachers explain standardized testing to students. They give them that guidance. But growing up, we never got that kind of instruction. We had to figure it out ourselves. So when I realized, "Oh, this is a standardized test, and memorizing a few templates can solve these problems," that was a really big shift for me.

Uncle Moon (Host) (00:06:18)

I think that way of thinking runs through everything you've done since. Choosing startups, for instance.

Uncle Moon (Host) (00:06:23)

Being a stay-at-home dad. Then building zero-person companies. We'll dig into all of that later.

Ran Chen (Guest) (00:06:27)

Sounds good.

Tsinghua, Zhihu, and the Early Internet Boom

Uncle Moon (Host) (00:06:28)

My next question is this. After you got into Tsinghua, by your second year, you were posting prolifically on Zhihu.

Ran Chen (Guest) (00:06:34)

Yes.

Uncle Moon (Host) (00:06:35)

I was one of your early followers there. You inspired me to start writing on Zhihu too.

Ran Chen (Guest) (00:06:39)

Wow.

Uncle Moon (Host) (00:06:40)

What made you want to start sharing your ideas there?

Ran Chen (Guest) (00:06:44)

It's pretty simple, really. Tsinghua was so hard. I was having a tough time. I wanted somewhere to express my thoughts. At university, I often felt really discouraged.

Uncle Moon (Host) (00:06:59) †

That's what the Fields Medal winner said.

Ran Chen (Guest) (00:07:00)

Right. Hong Wang called it "a little bit discouraged." I felt that way a lot when I was in college too. At the time, I wasn't just writing on Zhihu. I posted on lots of different sites, about different things. But apparently all the other sites shut down. Only Zhihu is still around. So that's how it happened. You can also see that later, when I was really happy with my life, I actually wrote much less.

Uncle Moon (Host) (00:07:25)

So your original motivation was that you were miserable

Ran Chen (Guest) (00:07:29)

and really anxious? There were so many things I was thinking about, but nobody around me to discuss them with or relate to. So I wanted to go online and put those thoughts out there to see if anyone was on the same wavelength. That's how it felt back then.

Uncle Moon (Host) (00:07:45)

Sounds like you picked the right platform. I also wrote for four years, posting a blog every week.

Ran Chen (Guest) (00:07:51)

Really?

Uncle Moon (Host) (00:07:52)

On my Qzone page.

Ran Chen (Guest) (00:07:53)

Qzone, right?

Uncle Moon (Host) (00:07:54)

But nobody seemed to want to discuss anything. So I couldn't keep it up. Maybe I should've joined Zhihu sooner.

Ran Chen (Guest) (00:07:59)

Yes, exactly.

Uncle Moon (Host) (00:08:01)

Okay. Thinking back to Tsinghua, what stands out the most? How did it help you? Or, as you said, leave you discouraged? What experiences really stuck with you?

Ran Chen (Guest) (00:08:10)

Tsinghua really is an incredible university. For people like us, there was a lot that was interesting. I started college in 2009, for example. Back then, in my case, my exam score wasn't that high. I got into a program with a relatively low cutoff: the School of Software. Of course, now that the internet industry has taken off, the admission scores for those majors have shot up. But in our class back then, many students had transferred in or been assigned there through admissions. They hadn't wanted to study there. The popular majors then were economics, management, architecture, and more traditional engineering fields, like electronics and electrical engineering. Those schools had higher admission cutoffs.

Ran Chen (Guest) (00:08:50)

So my experience at university was this. On the one hand, a lot of my classmates looked down on our major. On the other, the coursework was really hard. Many of us came in with no background at all. I'd competed in programming contests, so I had some experience. But many classmates were far better than me. By a long way. We started major-specific courses early, right in our first year. While many people were still learning for loops, just the basics of loops, our big assignment was to build an application from end to end, using APIs. We were using C# back then, with all the MSDN documentation. I think a lot of people found that really demoralizing.

Ran Chen (Guest) (00:09:37)

Many wanted to transfer to another department. They thought, "This is way too hard. I'll never catch up with these classmates." So the biggest shock for me was how hard the classes were and how talented some of the students were. And yet many people didn't have much confidence in their own major. That was the situation.

Ran Chen (Guest) (00:09:57)

But there was another side to it. During my first two years, around 2010 and 2011, the internet industry was starting to explode. I discovered that world and thought, "This is incredible. It's absolutely going to change the world and the way we live." But finding someone to discuss it with was surprisingly hard. So I felt deeply conflicted and was going through a painful time. That's why I went on Zhihu and wrote so much. It seemed like a place where people cared about the internet. I could write something that others might connect with. So it was a way to find an outlet, or a place where I could feel a sense of belonging.

Uncle Moon (Host) (00:10:36)

Listening to that story, you really feel how much things have changed. The world moves so fast.

Ran Chen (Guest) (00:10:40)

It really does.

Uncle Moon (Host) (00:10:41)

The internet was just taking off.

Ran Chen (Guest) (00:10:43)

Right.

Uncle Moon (Host) (00:10:43)

Hardly anyone used Zhihu.

Ran Chen (Guest) (00:10:44)

Right.

Uncle Moon (Host) (00:10:45)

You couldn't find people to discuss the internet.

Ran Chen (Guest) (00:10:46)

I couldn't.

Uncle Moon (Host) (00:10:47)

And Tsinghua's School of Software was a school people didn't want to attend.

Ran Chen (Guest) (00:10:51)

Exactly.

Uncle Moon (Host) (00:10:51)

In just a few years,

Ran Chen (Guest) (00:10:52)

yes,

Uncle Moon (Host) (00:10:52)

all of that became the default, or even the hottest thing around.

Ran Chen (Guest) (00:10:56)

Yes, the times change incredibly fast.

Uncle Moon (Host) (00:10:59)

The best choice at one point may not be the best choice later. You have to keep adapting to your situation.

Ran Chen (Guest) (00:11:03)

Right. And the smartest people all ended up in tech, didn't they? Whatever they originally studied, they ultimately landed in the internet industry. So when there's a major wave like that, sometimes our individual choices don't make that much difference, in a way. We all ended up working in the same industry.

Choosing Machine Learning Early at CMU

Uncle Moon (Host) (00:11:23)

That's true. Let's talk about what happened after Tsinghua. In 2013, you went to CMU to study machine learning.

Ran Chen (Guest) (00:11:32)

Right.

Uncle Moon (Host) (00:11:33)

Again, back then, machine learning wasn't nearly as popular as it is now.

Ran Chen (Guest) (00:11:37)

Right.

Uncle Moon (Host) (00:11:38)

AlexNet had only just come out in 2012, and hadn't attracted that much attention yet.

Ran Chen (Guest) (00:11:42)

That's right.

Uncle Moon (Host) (00:11:43)

Why did you choose that field?

Ran Chen (Guest) (00:11:46)

I've always been interested in cutting-edge technology, in whatever is at the frontier. But I didn't enjoy doing research myself. I joined a lab quite early, probably in my sophomore year, and started working there. I explored different research areas, and I felt I wasn't cut out for it. I wanted to build applications instead.

Uncle Moon (Host) (00:12:07)

Finally, someone from Tsinghua who doesn't want to do research! Thank you, Ran.

Ran Chen (Guest) (00:12:11)

That's right. I wanted to work on applications. I figured that out pretty early. I knew research wasn't for me. So, when I was in college, I did mobile app development, because mobile internet was booming then. I remember I was among the earliest Android developers. Across the whole Android ecosystem in China, I was part of the first wave. I even had the original Xiaomi phone. I went to its launch event, too. At the time, the Xiaomi 1 launch was invitation-only. So, as a college student, I was one of very few who got to go. I watched Lei Jun announce onstage that the Xiaomi 1 would cost 1,999 yuan. It was unbelievable, the atmosphere in that room.

Ran Chen (Guest) (00:12:52)

So, I thought, "Wow." My plan was to graduate and skip grad school. I'd develop mobile apps and start my own company. That was my original plan. Then, in my sophomore and junior years, I talked to some older students who'd gone down the startup path. I realized it wasn't quite right for me. The reason was I didn't have a real competitive edge. With mobile app development, I wasn't sure. I just felt that simply building an app wasn't enough. You needed to build something for a specific industry. But I couldn't find one, or even picture what it might be. So I felt that this wasn't something I should be doing. I couldn't make it work.

Ran Chen (Guest) (00:13:31)

Meanwhile, in the lab, I was exploring different things. What we now call machine learning or AI was called data mining back then. Data mining. I read a lot about it at the time. People were talking about data mining in social networks. That was the hottest topic. I thought, "Wow, this is great. This is what I want to do." So I looked into it quite a bit.

Ran Chen (Guest) (00:13:55)

And when I applied to study abroad, it was also a transitional period. Before I graduated, I was in the class that entered in 2009, the 2006, 2007, and 2008 cohorts mostly pursued PhDs. A coursework-based master's was lower on the prestige ladder. The assumption was that only weak students would do that, people who couldn't get funding for a PhD. By the time I graduated, I went abroad in 2013, and by 2014, 2015, or 2016, people saw coursework-based master's as good options. So I was right at that turning point. Many of my classmates still wanted to get a PhD, rather than a coursework-based master's.

Ran Chen (Guest) (00:14:34)

But I thought, "I'm going to work in industry anyway." And back then, applying for those master's programs was fairly straightforward. You'd make a list of the top 30 or 50 schools, pick the ones in big cities, and apply to all of them. I'd go wherever I got accepted. In the end, I got lucky. CMU has a lot of master's programs, and one of them accepted me. So I went. That's a snapshot of how things were then. That's how I ended up at CMU. My experience there was a bit different, because most of my courses focused on machine learning.

Uncle Moon (Host) (00:15:07)

Right. In your Zhihu posts and blogs, you talked about going all in on AI.

Ran Chen (Guest) (00:15:12)

Right.

Uncle Moon (Host) (00:15:12)

And AI being the fourth industrial revolution.

Ran Chen (Guest) (00:15:15)

Right.

Uncle Moon (Host) (00:15:15)

In 2013!

Ran Chen (Guest) (00:15:16)

At the time, yes, back then, I didn't actually call it "going all in on AI." I just told people that machine learning and data mining were incredibly powerful, and they absolutely needed to learn them. When I went to CMU, I chose different courses from my classmates. Although our program was called the Master of Computational Data Science, at the time, "data science" was the hottest buzzword. Now it's almost disappearing. Right.

Ran Chen (Guest) (00:15:44)

So all my courses revolved around machine learning, data science, and data mining. I remember this vividly. We studied things like large-scale machine learning: running Naive Bayes on a Hadoop cluster. Basically, doing some arithmetic and then some division. This was in 2013, and that's what one of the best schools for machine learning was teaching. Because even in industry, few people really knew how to do it. I also took cloud computing. In 2013 and 2014, they'd give you some credits to use AWS. That was very advanced, too.

Ran Chen (Guest) (00:16:20)

The course I remember most was deep learning. I only took courses related to machine learning. And at a school as big as CMU, for anyone who doesn't know, of the four top computer science schools, CMU has the most students, perhaps as many as the others combined, because it has so many master's programs. Guess how many students were in that deep learning class. Pick a number.

Uncle Moon (Host) (00:16:48)

Back then? Today, probably several thousand. Back then, two or three hundred?

Ran Chen (Guest) (00:16:52)

Well, it was a PhD-level course called Deep Learning. It didn't even fill its 20 seats. That's why I could take it as a coursework-based master's student. If it had filled up, I wouldn't have gotten in. PhD students had priority, then research master's students. Coursework-based master's students, since we weren't doing research, were low on the registration priority list. But it wasn't full, so I signed up. And then, since it was a PhD-level class, we read papers, then presented them to each other.

Uncle Moon (Host) (00:17:23)

Talking to you feels like traveling in a time machine.

Ran Chen (Guest) (00:17:27)

Right. We discussed things like Boltzmann machines, and Alex's paper, sharing those ideas. How do you build a fully connected neural network? Why does it work? Things like that. It was very early days. And I thought, "Wow, this is incredible. This is absolutely the future." Right. Of course, just two or three years later, that class was packed. That's how the times change. It really left an impression on me.

Uncle Moon (Host) (00:17:56)

So you went all in on AI and chose those courses. Even with fewer than 20 people in class, you stuck with it.

Ran Chen (Guest) (00:18:02)

Right.

Uncle Moon (Host) (00:18:03)

It feels like history moves in cycles. At a particular point in time, the earlier you get in, the more you stand to gain later. Is that right?

Ran Chen (Guest) (00:18:11)

Absolutely. But to get in early, you need a strong conviction. Because back then, for example, the work we were doing wasn't at the top of the prestige ladder. Infrastructure was at the top. Building infrastructure,

Uncle Moon (Host) (00:18:29)

cloud computing.

Ran Chen (Guest) (00:18:29)

Cloud computing. And in big data, it was infrastructure, too. The systems that let you compute at scale, including the infrastructure behind advertising. So our most talented people, the smartest ones, took pride in working on infrastructure, in building databases. They felt that was the most important work. And, of course, they had the best job prospects and earned the most. But I just wasn't interested in it. To me, machine learning was what would truly change the world. So, hardly anyone chose the same courses as me.

Ran Chen (Guest) (00:19:02)

Even for my capstone project, although I was in a coursework-based master's, I found a way to adjust my schedule so I could spend three semesters on a real project, applying AI to legal texts. I worked with a PhD student at Pitt. He published a paper, and my name was on it, too. It was such a niche field, yet today that paper has dozens, perhaps a hundred citations. I find that pretty amazing.

Uncle Moon (Host) (00:19:29)

AI in law is huge now.

Ran Chen (Guest) (00:19:32)

It is.

The Business Value of Technology

Uncle Moon (Host) (00:19:33)

But after more than a year of studying, you published some more blog posts,

Ran Chen (Guest) (00:19:38)

right,

Uncle Moon (Host) (00:19:38)

saying technology is definitely not the most important thing in business. What led you to that conclusion?

Ran Chen (Guest) (00:19:47)

Well, here's how I see it. The reason technology, especially machine learning, has always fascinated me is that, deep down, I've always seen the world this way: if you contribute something, that contribution should be measurable, ideally in terms of business value. That's why I find machine learning so compelling. It showed me a possibility: with this technology, you can directly observe how putting it to use creates more value for a company, how much more money it makes, or what new business it makes possible. Right? That's why I found it so appealing.

Ran Chen (Guest) (00:20:29)

Before that, when I worked on mobile or web development, I never had that same clear sense of impact. Someone would tell you, as a programmer, what to build, and you'd build it. But afterward, nobody would say that you, the programmer, generated 100 million in revenue. People wouldn't see it that way. But with data mining or recommendation systems, people would say, "Your algorithm improvements brought in 100 million in revenue." They knew exactly how to measure your contribution. That's why I loved working on algorithms, and why my career has followed that direction ever since. Deep down, I want my work to have an observable business impact.

Ran Chen (Guest) (00:21:10)

So I think what prompted that comment then was seeing people sometimes get too caught up in the technology itself, without thinking about the business value behind it. That's what made me say that. It's also the standard I set for myself, whether at a startup or leading a team at a company. I want everyone's work to be measured directly by its business impact. That makes your job easier, and mine, too.

Uncle Moon (Host) (00:21:38)

A born entrepreneur! A natural in industry. To figure that out so early is really impressive. I think I see what you mean. A lot of technology, like mobile development, is simply a means of automating a business.

Ran Chen (Guest) (00:21:53)

Right.

Uncle Moon (Host) (00:21:53)

But with algorithms, whether you click an ad depends on the algorithm.

Ran Chen (Guest) (00:21:57)

Right.

Uncle Moon (Host) (00:21:58)

And if you click, the company makes millions more.

Ran Chen (Guest) (00:22:01)

Right.

Uncle Moon (Host) (00:22:01)

If you don't, it doesn't. The quality of the algorithm is directly tied to the company's revenue.

Ran Chen (Guest) (00:22:04)

Right.

Uncle Moon (Host) (00:22:06)

Right.

Ran Chen (Guest) (00:22:06)

Exactly. And the key is who gets the credit. Who is responsible? The algorithm engineer gets the credit, rather than a product manager or someone else. You own the impact of your work. When you build an app or a website, the credit doesn't seem to go to you. That's what I thought was so different about it.

Putting Machine Learning into Practice at Trulia

Uncle Moon (Host) (00:22:24)

Your thinking has always been remarkably consistent. Then, in 2014, you graduated from CMU. Why didn't you choose a big company like Google or Meta, instead of going straight to a startup?

Ran Chen (Guest) (00:22:36)

Here's the thing. First, let's look back at what things were like in 2014. Google and Meta were already growing rapidly. But there were some even hotter companies at the time: Pinterest, Airbnb, Uber, and Snapchat. Many of our best classmates chose those companies first. Interestingly, people who graduated before us and after us went back to choosing Google and Meta.

Ran Chen (Guest) (00:23:06)

Here's what I was thinking. I'd always wanted a machine learning role where I could directly influence the business. I thought that would be hard at a big company. I did interview with some big companies, and some rejected me. Probably the best offer I got was from Apple. From Apple. But I felt doing algorithm work there would put me too far from the customer. It didn't align with what I'd always wanted to do.

Ran Chen (Guest) (00:23:44)

So I looked for a smaller company. Trulia was actually publicly listed at the time. It was a real estate search engine. It had already gone public. And its data science team, that's what we called it back then, I remember writing an article about the distinction in data science between applied scientists and data analytics. One focused on analysis, the other on algorithms.

Ran Chen (Guest) (00:24:03)

Our team, the data science team, focused specifically on machine learning algorithms. There were only five or six of us. We worked on lots of applications: image recognition, natural language processing, search, things like that, and recommendations. The main applications of machine learning. What you built could be used directly by end users. I thought, "Wow, this is great. This is it. This is exactly the kind of place where I want to work." The company could directly see the business impact of what I built. That's where I wanted to be.

Ran Chen (Guest) (00:24:37)

So I wasn't particularly interested in joining a bigger company. I wanted somewhere with a concrete application. And since it was real estate search, real estate also interested me, I wanted to work there. Of course, that was another choice that differed from all my classmates'. A lot of my decisions follow what I find interesting and what makes sense to me. They're often different from those around me.

Uncle Moon (Host) (00:25:02)

Right. At CMU, you chose a class with just 20 students. And for your first job after graduation, you chose a company few people knew about, instead of a big tech company.

Ran Chen (Guest) (00:25:11)

That's right.

Uncle Moon (Host) (00:25:12)

Then, Trulia was acquired by Zillow in 2017.

Ran Chen (Guest) (00:25:14)

Right.

Uncle Moon (Host) (00:25:15)

Did you see that coming?

Ran Chen (Guest) (00:25:17)

First of all, the acquisition didn't really have much to do with me. I was just a rank-and-file employee. I wasn't going to make much money from it. As for the company, it was the industry leader buying the runner-up. It seemed like a fairly normal merger. But for me personally, it didn't have much impact. I didn't make any extra money or anything.

Ran Chen (Guest) (00:25:37)

Still, the work at Trulia taught me a lot about the problems you can run into. When I went to my next company, Tubi, and joined the new company, I had a chance to build my own team. I wanted to avoid those problems. So, at Trulia, my first real job, I learned how things actually get put into production, and why most companies' machine learning algorithms never get deployed. Those were some pretty painful lessons. That's what I think I learned most from that company.

Uncle Moon (Host) (00:26:13)

So when the industry leader bought the runner-up, you said you didn't make much extra money. The acquisition didn't give you even a little financial freedom?

Ran Chen (Guest) (00:26:19)

No, none at all.

Uncle Moon (Host) (00:26:21)

You also just mentioned learning why so much work never gets deployed.

Ran Chen (Guest) (00:26:24)

Right.

Uncle Moon (Host) (00:26:25)

Could you share a bit about that?

Ran Chen (Guest) (00:26:27)

Oh, I could talk about that all day. The first and most important point, and I think it's still true in the AI era, back then it was machine learning, is that unless the company supports it from the top down and goes all in on ML or AI, starting with the CTO backing you, your machine learning algorithms will never make it into production. They won't generate much business value for the company.

Uncle Moon (Host) (00:26:53)

You need full support from senior leadership.

Ran Chen (Guest) (00:26:56)

Their full support. Because many of these changes are disruptive. The entire company's priorities need to be aligned, and organizational structures may need to change. The CTO has to step up and take the heat for your team, and tell everyone, "This is what we're going to do." During my time at Trulia, I didn't feel we had much of that support. So, whether it was experiments we wanted to run or algorithms we wanted to launch, it was painful. It was hard to get things on the roadmap, and we couldn't see results. It often felt like we were just amusing ourselves, writing blog posts, without seeing any business value. A bit like patting ourselves on the back. That was really frustrating for me.

Taking Initiative in a Job Search: I Am Here to Solve Your Problems

Uncle Moon (Host) (00:27:35)

You wrote a lot of blog posts then, right? Quite a lot.

Ran Chen (Guest) (00:27:38)

Yes.

Uncle Moon (Host) (00:27:39)

Did those posts help you join your next startup, Tubi?

Ran Chen (Guest) (00:27:44)

It's hard to say. At my previous company, there were a lot of things I wanted to change. And, personally, I really wanted to join a startup. So, for a while, I reached out to lots of startup people to chat over coffee. It was quite simple back then. You could just go online. I remember using Crunchbase, which I think is still around, and searching for startups that had recently raised pre-Series B funding. Then I'd collect the CEOs' email addresses and email them one by one: "I'm looking for a job, and I'm really interested in your startup. Would you be up for a coffee?" I talked to a lot of startups that way.

Ran Chen (Guest) (00:28:25)

One of the companies I spoke to was one I found on Zhihu, called Tubi TV. I came across Tian Chen on Zhihu. He's a really talented engineer. I asked him to meet for coffee, too. And I felt the company was doing a lot of the things I wanted to do. Right. So I interviewed there, and eventually joined. That's how it happened.

Uncle Moon (Host) (00:28:48)

There's a real lesson in that experience. I see two issues now. The first is that a lot of younger alumni keep telling me, "It's so hard to find a job. I've sent out 300 resumes without landing a single job."

Ran Chen (Guest) (00:28:59)

Right.

Uncle Moon (Host) (00:28:59)

But at the same time, I've been talking to lots of startup founders in Silicon Valley, and they say, "We really can't find people."

Ran Chen (Guest) (00:29:04)

They genuinely can't hire anyone.

Uncle Moon (Host) (00:29:05)

So after hearing your story, I think this applies in any era: if you want to join a startup, really do your homework on it and message the CEO directly.

Ran Chen (Guest) (00:29:13)

Exactly. First, reach out directly. Second, a lot of people look for jobs with a student's mindset. I think that's a real problem. What do I mean by that? "Please give me a job. I'm here to learn." I've never approached a job search that way. What I tell people is, "I'm here to put AI to work for you. Trust me. You're not doing AI yet. You're not doing machine learning. I'm that person. I'm the person you need. You should talk to me. I'm here to solve that problem." I'd only been working for a little over a year, a year or two, at that point. But confidence is incredibly important. You have to inspire confidence in others and have confidence in yourself for people to give you that chance. As for how well you'll actually do, I believe most people will do just fine.

Ran Chen (Guest) (00:30:01)

But confidence is key. "I'm the one. I'm the person who'll solve your problem. You have to get coffee with me. If you don't, that's your loss." That's the attitude to bring to the conversation. "I know your company has a lot of machine learning problems you haven't been able to solve. Let's get coffee." That was the mindset I had when I sent those emails. I ended up having lots of conversations. Whether or not I joined the company, or got an offer after interviewing, those were all really good, productive conversations.

Uncle Moon (Host) (00:30:32)

Everyone, memorize what he just said. In this day and age, that's how you need to look for a job.

Ran Chen (Guest) (00:30:36)

That's what winning looks like. Take a page from President Trump. That mindset of his, that focus on winning, really matters. "I'm here to win. I'm not here to learn or just have a look around. I'm here to take responsibility. I'm here to win with you. I'm the winner. I'm the person who's going to win."

Uncle Moon (Host) (00:30:54)

Right.

Ran Chen (Guest) (00:30:55)

I think that mindset matters most.

Uncle Moon (Host) (00:30:57)

All right, we've got the title for this episode: "I'm Here to Win."

Ran Chen (Guest) (00:31:00)

"I'm Here to Win." The same goes for implementing AI. A lot of people today are looking for jobs based on their AI skills. The approach is the same. Tell them, "Struggling to make AI work at your company? I'm the one. I'm the person who'll make it happen." Of course, you have to add value. You need something valuable to offer, and deliver a lot of it. So you really need to research their company, explore the problems with them, and maybe build some prototypes. With AI available now, that's all pretty easy to do. But you have to bring that attitude.

Uncle Moon (Host) (00:31:30)

Everyone, take notes. I need to learn from this too.

Ran Chen (Guest) (00:31:33)

Exactly.

Growth, Career Progression, and Management at Tubi

Uncle Moon (Host) (00:31:34)

Then this company, Tubi TV, got acquired too.

Ran Chen (Guest) (00:31:37)

That's right.

Uncle Moon (Host) (00:31:37)

You joined two companies, and both were acquired. What's your secret?

Ran Chen (Guest) (00:31:41)

I don't think there's really a secret. Anyway, I joined Tubi in 2017. We were in streaming, specifically long-form video. It was really interesting. Tubi was probably one of the few long-form streaming companies that made money. It's an incredibly hard industry to profit in. Hardly anyone makes money. We were probably one of the few. Then, after it was acquired in 2017, now, in 2026, it's probably become one of the biggest streaming companies. After Fox acquired it, its traffic and so on are probably on a similar scale in the U.S. to Netflix and Hulu. So it's actually a very big company. But because it's free and ad-supported, it serves a more budget-conscious U.S. audience. So a lot of people in Silicon Valley don't know it.

Uncle Moon (Host) (00:32:31)

Got it. Did you get a big windfall this time?

Ran Chen (Guest) (00:32:35)

I got something out of it, but certainly not a huge amount. Enough to support me staying home for a year or two, two or three years, doing what I wanted. But, yeah, it wasn't much more than that. It wasn't enough to retire for good. Definitely not.

Uncle Moon (Host) (00:32:49)

Being able to take a year or two off in Silicon Valley is already pretty impressive.

Ran Chen (Guest) (00:32:52)

Yes. So I stayed home for a while. And afterward... My time at Tubi was really my chance to experience a startup growing from almost nothing. It completely changed how I thought about climbing the career ladder and about starting a business. I'd been through the whole cycle. In terms of learning, that was where I learned the most. That's why I think joining a startup is a great opportunity.

Uncle Moon (Host) (00:33:15)

Let's talk more about the career ladder. You eventually became a director.

Ran Chen (Guest) (00:33:20)

Right.

Uncle Moon (Host) (00:33:21)

And you started planning things from a more senior perspective.

Ran Chen (Guest) (00:33:25)

Yes.

Uncle Moon (Host) (00:33:26)

Compared with being a hands-on engineer at a startup, as you worked your way up, how did your perspective change? What did you learn?

Ran Chen (Guest) (00:33:32)

Here's how I see it. A lot of people, I often see people asking, "If I join a company, how can I get promoted fastest? What's actually the fastest way up?" I think big tech and startups are quite different in that respect. Big tech may have its own playbook for promotions. But at a startup, it's often simple. You join early, take ownership of an area, and as the company grows, as long as you can handle the responsibility, your team grows, and you get lifted up with it.

Ran Chen (Guest) (00:34:03) †

So is ability a prerequisite?

Ran Chen (Guest) (00:34:07)

Of course, you need some ability. But I believe a lot of people listening are smart enough that if you gave them a tougher problem, they'd be able to solve it too. That's part of why so many people want to join startups: to take a shot at that opportunity.

Uncle Moon (Host) (00:34:23)

How would you describe your management style once you became a director?

Ran Chen (Guest) (00:34:28)

Wow. That's a great question. Here's the thing. I joined as a hands-on engineer. When I joined Tubi, I was the first ML engineer. The whole data and ML team, across machine learning and data, was just two people, including me. We built all the data systems, the machine learning, and the ML infrastructure from scratch. That included data science, analytics event tracking, all of that. Everything started from zero, including A/B testing.

Ran Chen (Guest) (00:34:55)

One thing really stuck with me. After you've worked for a while... When I joined, I sat with the CTO, because there weren't many people. He cared a lot about this team. So I asked him, "For ML, do we have a roadmap? What's your vision?" He said, "Ran, isn't that exactly why I hired you, to tell me what we should do? Why are you asking me? You're supposed to tell me." I think that really changed my mindset. I thought, "He's right. I'm the one here to do this, not him. That's what he hired me for." And then I thought, "That's great!" So I started thinking about all the frustrations I'd had at Trulia and how I could do things differently here. At the time, that was a huge shift in my mindset.

Ran Chen (Guest) (00:35:49)

So when you ask about my management style, I think it follows the same principle. Basically, when someone joins, it's not about what I want them to do. It's about what they want to do. What do you want to accomplish? What important work do you want to do? I think that's what matters most. I believe people only do their best work when they're doing something they want to do. That's really the key. So as a manager, my most important job is to help everyone establish the broader framework. A framework for iterating, for measuring performance, for example. How do we measure whether you're doing a good job? I put that framework in place. Then I remove obstacles involving other teams. If something's blocking you, slowing down your iterations, I'll help get it out of the way. The rest is up to you. You need to tell me how it's going. That's been a big part of my management style ever since.

Ran Chen (Guest) (00:36:46)

But it has its pros and cons. If you're someone who really trusts me, you'll think, "Wow, this team gets a lot done." But if you don't trust me, you might think, "This guy doesn't seem to do anything." Because it's the people on the team doing the work, and doing it so well. It feels like, "This guy, if we got rid of him, things might be much the same." So there are different trade-offs.

Uncle Moon (Host) (00:37:06)

That's just how it is. The best generals win without fanfare. It's always been that way.

Ran Chen (Guest) (00:37:10)

Exactly. That's why, at a startup, what matters most is the CTO and other leaders trusting you. If they trust you, you can get a lot done. If they don't, life can be miserable for the ML team.

Uncle Moon (Host) (00:37:24)

So another piece of career advice is to put yourself in an environment where people trust you, or work to build that trust.

Ran Chen (Guest) (00:37:33)

That's a whole other topic. Trust can be a very subtle thing. People sometimes complain about differences in race, or differences in language. All kinds of things influence it subtly. But once you have trust, everything gets much easier. That brings us back to a question: when it comes to climbing the career ladder, what determines whether you move up? Is it the work you do, or the person making the decision? Those are two completely different approaches. I think that's something everyone needs to think through. It's something a lot of people haven't really figured out. If what really determines how fast you get promoted is a person, then grinding away at tasks every day is often pointless.

Uncle Moon (Host) (00:38:22)

I'm sure that hits home for a lot of people.

Ran Chen (Guest) (00:38:25)

Right.

Life as a Stay-at-Home Dad and Sharing Family Responsibilities

Uncle Moon (Host) (00:38:25)

Okay, that was your second chapter.

Ran Chen (Guest) (00:38:28)

Right.

Uncle Moon (Host) (00:38:28)

And after that, you became a stay-at-home dad.

Ran Chen (Guest) (00:38:31)

Yes.

Uncle Moon (Host) (00:38:31)

For two years.

Ran Chen (Guest) (00:38:32)

That's right.

Uncle Moon (Host) (00:38:33)

Was it because your kids were just too adorable, or because you'd gained some financial freedom and felt you could take a break?

Ran Chen (Guest) (00:38:39)

I think it was a combination of factors. I left around 2023 or 2024. It had been a while since the acquisition. A lot of the colleagues I'd really enjoyed working with had left to start their own companies and so on. I felt it was a good opportunity to think about what I wanted to do next. And at the time, we had two children. Our older one and our younger one. The younger one was a newborn. Both were little. On top of that, my wife was incredibly busy. She works at a Big Four accounting firm, which is already demanding enough.

Ran Chen (Guest) (00:39:14)

And I really wanted her to be able to advance her career. So she was doing a part-time JD, a Juris Doctor, a professional law degree. It takes four years part-time, or three years full-time. The difficulty of doing that, if you know the industry, is almost unimaginable. Even her colleagues found it hard to imagine someone working at a Big Four firm while also doing a part-time JD.

Ran Chen (Guest) (00:39:44)

From my perspective, for women, having a baby is already incredibly hard. I felt it was worth it. I really wanted to help. I didn't want her career to be interrupted by having children. I even hoped that having children could help her take her career a step further. So I really wanted to do this. I'd made some money, and it was a natural point to take a career break. Why wouldn't I? As I always say, we don't make money just to make more money. We make it so we can use it when it matters. So I took a year or two off.

Ran Chen (Guest) (00:40:17)

And there's another thing these past two years, especially this year, have made me realize. Why was it so important to support her in doing this? AI. AI is setting a brutal cutoff for which jobs survive. I think that's something a lot of people still haven't grasped. Its impact on work, on so many jobs, is enormous. A lot of people still have jobs, but in the face of AI, their work has already become irrelevant. So I really want my wife to stay above that cutoff, to stay afloat and keep moving up. Right.

Ran Chen (Guest) (00:40:51)

So during those two years, in our family, we were two working parents with two kids. Well, at that point, I didn't actually have a job. We were both raising the kids, my wife was working, and I was exploring lots of other things myself. Because I felt that looking after the kids didn't actually keep me that busy, even though I was doing it full-time. So I also spent a lot of time earning licenses. That led to a substantial side business in financial planning. All of that grew out of that period.

Uncle Moon (Host) (00:41:18)

We'll get into the financial planning later.

Ran Chen (Guest) (00:41:21)

One thing I find interesting is, when I say that looking after the kids didn't keep me all that busy, a lot of people just can't understand it. "What do you mean, you weren't that busy?" So it's hard to explain sometimes. But the reality was, my wife and I were raising the kids ourselves, with no hired help, and our parents weren't around. I earned a bunch of licenses, and she took the LSAT, took other exams, and did a lot of studying. We really did find ourselves with more time.

Uncle Moon (Host) (00:41:49)

And that extra time came from you staying home full-time?

Ran Chen (Guest) (00:41:53)

Right. Yes. But, yeah, even though I was doing it full-time, I still had a lot of time. That's how it felt to me.

Uncle Moon (Host) (00:42:01)

Were you sending the kids to daycare?

Ran Chen (Guest) (00:42:03)

The older one wasn't going at that point. We only started daycare last year, for the younger one. So back then, I looked after them at home.

Uncle Moon (Host) (00:42:11)

So when one parent isn't working, the whole family has a lot more time.

Ran Chen (Guest) (00:42:15)

Right. Of course, I don't know if it also depends on how you parent. But I felt I had quite a bit of time. That was my experience then.

Uncle Moon (Host) (00:42:23)

How did your wife feel about it? She was working at a Big Four firm and preparing for all this.

Ran Chen (Guest) (00:42:30)

Right.

Uncle Moon (Host) (00:42:30)

It must have been exhausting.

Ran Chen (Guest) (00:42:31)

Yes.

Uncle Moon (Host) (00:42:31)

Meanwhile, you were home with the kids.

Ran Chen (Guest) (00:42:33)

Right.

Uncle Moon (Host) (00:42:33)

Had you two fully talked this arrangement through?

Ran Chen (Guest) (00:42:37)

Yes, absolutely. She wanted me to take a break too. And she could push harder in her career. She really wanted to finish the degree. She'd always dreamed of becoming a lawyer. Looking at things today, I think a U.S. law license probably puts you well above that cutoff. AI is having a huge impact on the legal industry, but paralegals are bearing the brunt of it, the people who don't have law licenses. For licensed attorneys, though, I think AI is highly complementary.

Uncle Moon (Host) (00:43:11) †

The barrier to entry is higher.

Ran Chen (Guest) (00:43:13)

It's quite high, and they're not short of clients. They can make a lot of money too.

Uncle Moon (Host) (00:43:20)

So your whole family works together that seamlessly?

Ran Chen (Guest) (00:43:26)

I guess you could say that. But either way, I've always believed that for women, having children is an incredibly difficult experience. So I openly urge other men to do more for their families.

Uncle Moon (Host) (00:43:41)

Thank you. You've just made this podcast a huge hit with the women watching.

Ran Chen (Guest) (00:43:45)

Right.

Parenting: Helping Your Children Enjoy Being with You

Uncle Moon (Host) (00:43:46)

During those two years with the kids, were you with them all day, every day?

Ran Chen (Guest) (00:43:52)

Yes, so they're both very attached to me. I spend more time with them. When something's wrong, they come to me first, rather than their mom.

Uncle Moon (Host) (00:44:00)

Whoever spends more time with them is the one they turn to.

Ran Chen (Guest) (00:44:02)

Exactly.

Uncle Moon (Host) (00:44:03)

During that time, what parenting philosophy did you come to believe in most strongly? And what did you discover that you hadn't before?

Ran Chen (Guest) (00:44:12)

I think kids are remarkable little creatures. They really do love you so much. As a person, I'd never encountered a little being that loved me so constantly. That has given me a tremendous amount of energy. It's also made me think, and keep thinking, about what their future should look like, what kind of life they should have, and how we should approach their education. I used to have certain ideas, but in the AI era, none of those matter anymore. I think people still struggle to imagine just how much AI will change human society. As things stand, our kids have started preschool, and our approach has become very simple. Whether we're choosing a school or anything else, what matters most is being close to home. We've dropped all our other requirements. It just has to be close to home. Because spending more time with the kids, being together more, is a good thing. So pretty much all the activities we do are within a five-minute drive of home.

Ran Chen (Guest) (00:45:13)

And with kids, what matters most, something I hadn't really thought through before, but have felt strongly about ever since, is that the most important thing in parenting is for your kids to like you. I rarely hear people say that. But I don't think it's easy. For my kids, I want being with their parents to give them an instinctive sense of comfort. When they see you, they feel happy. And when they're tired, they feel they can come to you to recharge. That's the relationship I want with my kids.

Ran Chen (Guest) (00:45:46)

As for specific skills, or what knowledge they should acquire, the times have changed. With AI, all of that is so uncertain. I don't even know what my own future looks like. So what I can do is this: if we have sound values, and our kids really like being with us, then we can, in our own way, have a positive influence on them. I think that gives them something better in the long run. That's what I'd rather do. If they dislike you, they'll look to other people for influence. Then it'll be hard for you to guide them in a positive direction. So the more I think about it, the more I feel that's what matters most. As for their grades at school and all that, I think doing reasonably well is enough. That's how I feel.

Uncle Moon (Host) (00:46:29)

I'm clipping this part out and watching it once a month. I need to take this to heart. My kid is two now too.

Ran Chen (Guest) (00:46:35)

How does it feel? What's your experience been like?

Uncle Moon (Host) (00:46:37)

I feel like my child really does love me even more than I love them.

Ran Chen (Guest) (00:46:40)

Yes. It's hard to describe that feeling.

Uncle Moon (Host) (00:46:43)

Very hard to describe. And it makes you feel young again. Seeing the world anew through their eyes, it's like living life all over again. Otherwise, life can feel like it's slowly fading.

Ran Chen (Guest) (00:46:52)

Right. I have this feeling that once you're in your thirties, or your forties, I'm not sure exactly, but at some point, it's hard to find joy anymore. It's difficult to describe that feeling. But, you know, when you've got nothing to do, you scroll Xiaohongshu, right? You just keep scrolling. Are you actually happy? Not really. I used to play Dota 2 sometimes, but I couldn't manage the regular mode anymore. I could only play Turbo, the faster mode. But did it bring me any joy? It didn't really feel like it. Yet somehow, with my kids, I can still feel a genuine sense of joy. That's probably something I never expected.

Uncle Moon (Host) (00:47:33)

So the takeaway is: stay close to home, spend more time together, and give them a safe haven of love and acceptance. Help them trust you more.

Ran Chen (Guest) (00:47:40)

Right. And take them out to play whenever you can. Make plenty of time for fun. Have fun yourself too. You need to be happy. When I talk to people about raising kids, I always say happiness matters most.

Ran Chen (Guest) (00:47:49)

I've said this before: why are the kids unhappy? Because their parents are miserable at work. They get to work and think, "My boss is an idiot. My coworkers are idiots. Zuckerberg is the biggest idiot of them all." All they think about is who's an idiot. Then they come home, and the two of them keep talking about who's the bigger idiot. They go on and on. You're miserable and irritated, so what your kids pick up from you is misery and irritation. Can you really hide that? I don't think you can. Even if you hide it in front of your kids, it'll come out somewhere else. They'll still sense it. So finding happiness in your own life, when it comes to raising children, is the most important thing. I'd rather influence my kids by changing myself.

From Solving His Own Problems to Financial Planning

Uncle Moon (Host) (00:48:45)

Our parenting approach is very close to yours, but we're not at your level yet. I've learned a lot. My next question picks up on what you said earlier. While you were a stay-at-home dad, you also earned a bunch of financial planning qualifications.

Ran Chen (Guest) (00:48:58)

Yes.

Uncle Moon (Host) (00:48:59)

What was the thinking behind that?

Ran Chen (Guest) (00:49:02)

I had a very clear feeling at the time. I'd come into some money, though not a huge amount, and I really wanted to understand how to manage it sensibly. So I talked to a lot of people, financial planners and so on. But I clearly wasn't satisfied. They couldn't understand some of the unique circumstances we face in Silicon Valley and the Bay Area. I'll share some examples in a moment. They're pretty interesting. They just couldn't grasp them. They had a very conventional approach to planning. I didn't think they understood the Bay Area. They were professional, and very nice people.

Ran Chen (Guest) (00:49:37)

Then there were all the free services, at banks, or through other providers. When you talked to them, you could clearly feel this coming: "All of that is fine, but why not buy this wonderful thing? This amazing investment product, this fantastic insurance policy," and so on. You could feel it. The longer you talked, the more it became a sales pitch. It's not that I was unwilling to buy anything, but could we have a real conversation? I still had so many unanswered questions. So after making the rounds, I was pretty annoyed. And I thought, "If you want something done, do it yourself." So, if there was all this stuff I didn't understand, why not study it myself?

Ran Chen (Guest) (00:50:16)

The first license I got was for life insurance, because at the time, that was what I understood least. Then I got a tax credential, the Enrolled Agent designation. I passed those exams pretty quickly too. I think there were three parts. I passed in 40 days, starting from scratch. I remember posting about it on Zhihu. Everyone said it was impossible. "No beginner could pass them that quickly." I said, "Okay, sure. But I've already passed." Then I gradually got securities licenses, Series 6, 63, 65, and so on. I started to feel I was at the testing center so often that the person working there probably recognized me. They kept seeing me come in for exams. Finally, after getting some professional experience, I earned the toughest one: Certified Financial Planner. The CFP designation. I hold that credential too.

Ran Chen (Guest) (00:51:06)

And later on, because I'd had so much experience studying and taking exams, I turned my own materials into a website called Open Exam Preparation. It's a free site where people can study using exam prep materials. But anyway, that experience, for me, was really about trying to solve my own problems. Along the way, I discovered a lot of things I'd never considered, things I couldn't find online either. Studying and earning those qualifications really changed how I thought about a lot of things. Most importantly, I came to understand people's relationship with money. As an individual, how do you think about money itself? That changed a lot of my thinking.

Silicon Valley’s View of Risk and Life Choices

Uncle Moon (Host) (00:51:48)

Let's dig into what you said earlier, about them not understanding you, and not really grasping the situation in Silicon Valley.

Ran Chen (Guest) (00:51:56)

Sure. Let me ask you something. For the typical person in Silicon Valley, looking at their asset allocation, on a risk scale from one to ten, where would you put the average portfolio? One is low risk, ten is high risk. Where would it fall?

Uncle Moon (Host) (00:52:11)

Well, people in Silicon Valley get at least half their compensation in stock.

Ran Chen (Guest) (00:52:14)

Right.

Uncle Moon (Host) (00:52:15)

I'd say around six.

Ran Chen (Guest) (00:52:17)

Okay. Do you think buying the S&P 500 is conservative? Is that low risk or high risk?

Uncle Moon (Host) (00:52:24)

I've been investing in the S&P 500 for seven or eight years. I see that as being extremely risk-averse.

Ran Chen (Guest) (00:52:30)

Extremely risk-averse?

Uncle Moon (Host) (00:52:31)

Yes.

Ran Chen (Guest) (00:52:31)

So you'd put it at, say, two or three?

Uncle Moon (Host) (00:52:34)

Right.

Ran Chen (Guest) (00:52:34)

Exactly. And that shows how different Silicon Valley is. When I talked to those planners, I said, "Let's talk about Silicon Valley. A lot of people treat a single stock, their company's RSUs, like a savings account. They never sell the shares they receive, except for a little when they buy a house." I asked, "What do you make of that? Almost everyone around me does this." Do you know what they said? "That's impossible. Nobody would do that."

Ran Chen (Guest) (00:53:07)

Why? Because from a conventional risk-management perspective, your job depends on tech. Your biggest asset, your home, depends on tech. All your investments, your individual stock and RSUs, are in your own tech company. In terms of risk, you're already off the charts. Forget a scale of one to ten. You're at a hundred. Nobody would take on that much risk.

Uncle Moon (Host) (00:53:32)

But everyone in Silicon Valley does exactly that.

Ran Chen (Guest) (00:53:34)

Right. That's the problem. If you look at it, this is what everyone in Silicon Valley does. But when you talk to them, they say, "Nobody would do that." I say, "Everyone in Silicon Valley does." "Impossible."

Uncle Moon (Host) (00:53:43)

Then you clearly haven't met enough clients.

Ran Chen (Guest) (00:53:45)

See? We understand risk differently. I just asked you, and you put the S&P 500 at one or two. But if you look at the conventional approach to assessing portfolio risk, holding mostly S&P 500 stocks puts you on the high-risk side. Because those are equities, and that category is high risk, around seven or eight. But what do we think seven or eight means? We tend to think holding Nvidia is a relatively low-risk investment. Speculating on small stocks or buying options, that's high risk. Our understanding of risk is totally different.

Uncle Moon (Host) (00:54:21)

To us, options and futures, that's where the financial risk is. If it's just stocks, there's hardly any risk.

Ran Chen (Guest) (00:54:25)

Right. That's how we perceive risk. So our risk appetite, the reasons we think this way, and our ideas about asset allocation are completely different from Wall Street or conventional financial planning. In their eyes, these people are a bunch of crazy risk-takers. "Nobody would actually do that." But from inside our world, it all seems pretty normal. Of course, the question is, who's right? Strictly speaking, we don't know. With this kind of thing, all we can do is look in the rearview mirror of history. But your asset allocation is a reflection of the life choices you make. That's what I learned. The portfolio you choose is part of your plan for your life. It's not an objective judgment. There's no objectively better or worse choice. It's about what kind of life you want to live.

Ran Chen (Guest) (00:55:18)

You need to understand the different assets and their potential impact on your life. Then you choose your asset allocation. That's your decision. It's your own plan, based on figuring out what you want from life. So that was when I realized how different Silicon Valley really is.

Ran Chen (Guest) (00:55:38)

Of course, there's one thing I don't think people realize. Everything may be going well today, but what if one day, even if the chance is just one percent, the bubble bursts or something? If your holdings today are that heavily concentrated, you could lose your house, because you can't pay the mortgage. You could lose your job, and all your RSUs could go to zero. You might have worked for thirty or forty years, and lose everything. And your kids, because you can't afford to live in Silicon Valley, can't stay in their schools. If you and your wife both live in Silicon Valley, you could both be out of work. Then you'd have to pack up and slink back to China. Your whole life here would be gone. It's a low-probability event, but if it happens, that's what life could look like for a lot of people. The level of risk people take in Silicon Valley is higher than most people can imagine.

Financial Freedom, Compound Growth, and the Barbell Strategy

Uncle Moon (Host) (00:56:25)

After you earned those credentials, you must have helped a lot of people with planning.

Ran Chen (Guest) (00:56:29)

Yes.

Uncle Moon (Host) (00:56:30)

Here in Silicon Valley, what do the least rational people look like to you? And what about the most rational?

Ran Chen (Guest) (00:56:37)

I find this really interesting. If you look at conventional financial planning advice, it'll always tell you, if Google gives you shares, or Apple, if you work there, as soon as you receive the shares, you should sell them. Right? But in reality, looking back over the last thirty years, the people who did that are probably still working.

Uncle Moon (Host) (00:56:58)

Right.

Ran Chen (Guest) (00:56:58)

The ones who didn't sell achieved financial freedom long ago. That's a fact you can't get around. In other words, if you followed the textbook, you're still working. Meanwhile, the people who completely ignored the textbook are living the good life, posting happy updates on WeChat every day. What do you make of that?

Uncle Moon (Host) (00:57:17)

Right.

Ran Chen (Guest) (00:57:17)

So financial planning, or anything like it, is, a lot of the time, just a tool, in my view. It helps us make informed decisions. You need to understand what you're dealing with. Silicon Valley is an extraordinary place, because all the money is here. Almost all the rapid growth comes from here. So people constantly hear countless stories about someone they know who went all in on something, or made a fortune at some company. Right? I think there are two things here. The way I see it, there's no right or wrong, but you really need to think them through. The first question is: If you win the lottery, do you spend it on another lottery ticket, or go and enjoy your life? You really need to think that through. Because a lot of people think, "I've made this much money already. I can make even more. I'll use this money to make more money."

Ran Chen (Guest) (00:58:10)

If that's your decision, go ahead. But you need to be clear about it. Because that money could... You could lose the money you invest. But if you think, "I've made enough," you can set a goal to live well, enjoy your life, spend more time with your kids, and be there for your spouse. I think that's great too. You need to think it through.

Ran Chen (Guest) (00:58:27)

The second thing is, once someone wins the lottery, it can go to their head. You need to be aware of that too. Did we make our money because of the times? Was it a gift of the era we lived in, or did it really come from our own ability? A lot of lottery winners think they're the chosen one. "I'll definitely win again." And with people like that, you see many of them end up living miserable lives. So when you're faced with a huge windfall, I think you need to see things clearly.

Ran Chen (Guest) (00:59:02)

And then, think about casino stories. Stories of winning rarely get passed around. You don't hear many stories of losing either. Among the many clients I've worked with, there have actually been quite a few losses. People lose money for all sorts of reasons. They trust someone else's recommendation, or stick with a particular investment strategy. And the money is just gone. So when you make these decisions, you need to think through the pros and cons of each one. Personally, I recommend two approaches. These two approaches are quite different. One is more of a textbook approach: compound growth. You put some money away every year and watch it gradually build up. A lot of people believe in that. I can see the logic. It's the textbook approach. But these days, I'm not so sure young people buy into it.

Ran Chen (Guest) (00:59:57)

So there's another approach, called the barbell strategy. You keep part of your assets very conservative, and go all in with the rest. Now, compared with compound growth, there's a difference here. When you pursue compound growth, as your assets increase, your spending, including your home, car, and vacations, gradually increases too. You have more assets and more cash flow. So everything slowly gets better, little by little. You get that steady, lasting happiness.

Ran Chen (Guest) (01:00:29)

But with a barbell strategy, you need to be prepared. The money you go all in with could disappear. So you have to be ready to keep living the way you do today, possibly for the rest of your life. The same house, the same car. Don't upgrade your lifestyle. You need a stable cash flow. Build a very reliable income stream and a stable asset base, so even if everything else disappears, you can still live a good life. A perfectly decent life. Then, with the money that's left, you can do some really crazy things, some wild things. Go all in, use leverage, do whatever you want. But you have to make sure it doesn't jeopardize your everyday life.

Ran Chen (Guest) (01:01:06)

What's the upside? If the thing you go all in on really works out for you, you could move up a whole wealth bracket. You might make your first ten million. And if you're willing to stop there, your life moves into another bracket, with a completely different lifestyle. I think that sort of thing may be more appealing to younger people, or to a lot of people, for that matter. So these are two different paths. People often mix them up, but they really are different. I think each has its pros and cons. You just need to decide which path you want.

Uncle Moon (Host) (01:01:40)

Nothing in this show constitutes advice. Please use your own judgment. I've certainly seen both approaches among plenty of people in Silicon Valley.

Ran Chen (Guest) (01:01:48)

Of course, there are other approaches too.

Uncle Moon (Host) (01:01:50)

But after listening, just give it some thought yourself. We shouldn't go any deeper. We'll get into trouble if we do.

Ran Chen (Guest) (01:01:55)

Exactly.

Preparing Careers and Finances for AI Disruption

Uncle Moon (Host) (01:01:56)

As you were saying, for many people, suddenly having a lot of money can mess with their heads. But there's another situation. Many people in Silicon Valley earn a lot, like employees at big tech companies, and people in all sorts of other roles. Yet when you see them on social media, they're always complaining. They don't feel that free in their lives. Where do you think things have gone wrong?

Ran Chen (Guest) (01:02:19)

When work was easier, I don't think so many people complained. It's probably just these last two years. AI has had an enormous impact recently. A few years ago, people had a steady path up the career ladder. There was something to look forward to. So whatever problems you had in life, you'd complain a little: "This person's an idiot. So is that one." And then you'd move on. But competition in Silicon Valley has intensified, and AI is disrupting every industry. I don't think there's an easy answer to that. So today, once you're done complaining, remember: tomorrow could be even worse. I think the impact of AI will be much more intense than people expect, in how companies are structured and how industries change.

Ran Chen (Guest) (01:03:05)

So if you prepare yourself for the possibility that, in two years, your company could be gone, and approach your job with that mindset, you might feel a little better. Whether it's the career ladder or something else, so much is changing so dramatically that it's hard to predict what will happen. So when you say people have money but still complain, I'm sure a lot of them are struggling with cash flow or other things. But how can I put it? This is the era we're living in. And there will be even more turmoil ahead. So I really hope people can take a more cautious view of the future and prepare better now.

Uncle Moon (Host) (01:03:47)

Be more cautious about their career prospects, to put it more precisely?

Ran Chen (Guest) (01:03:51)

Yes, whether it's your career, your spending today, or your financial planning, all of it may be affected. Society could be entering a period of enormous upheaval.

Uncle Moon (Host) (01:04:01)

But society's productivity is also exploding exponentially, right?

Ran Chen (Guest) (01:04:06)

Yes. You'll see some people race ahead, whether they're building a side business or finding other ways to make money. We're seeing lots of people use new tools to do really interesting things and make money. But traditional jobs in big tech, and that whole internet-industry career story, along with other industries, are being hit hard.

Uncle Moon (Host) (01:04:28)

Ran has worked at so many different companies, managed people, and been an individual contributor himself. Using AI tools these days, you probably burn through billions of tokens a day, right? With all that experience, and being on the front lines in Silicon Valley, he's bound to feel these changes sooner.

Ran Chen (Guest) (01:04:46)

That's right.

A Free Exam-Prep Platform and Zero-Person Side Businesses

Uncle Moon (Host) (01:04:46)

So it's worth hearing what he has to say. Let's move on to our next topic: your zero-person side businesses.

Ran Chen (Guest) (01:04:52)

Sure.

Uncle Moon (Host) (01:04:53)

Everyone's talking about one-person companies, and you've gone straight to zero-person businesses.

Ran Chen (Guest) (01:04:57)

Yes.

Uncle Moon (Host) (01:04:58)

The first one is the exam-prep company you mentioned.

Ran Chen (Guest) (01:05:01)

Right.

Uncle Moon (Host) (01:05:02)

How did you build that side business, step by step?

Ran Chen (Guest) (01:05:05)

Here's how it happened. When I was taking these exams myself, I spent a lot of time preparing and a lot of money on study materials. But honestly, I didn't think the materials were particularly well written. And I realized that, in the end, for these exams, I mostly just asked AI questions and had it help me find lots of resources to study. That may also explain why I learned so quickly and passed the exams so fast. It was probably thanks to AI. So I thought, given all that, why don't I take this idea of free study materials and AI-assisted learning and turn it into a website? The idea was that simple. The great thing about AI is that you can keep adding more exams. So at the beginning, we covered a dozen or so exams. Then hundreds, thousands, and tens of thousands. It just keeps growing, getting bigger and bigger.

Uncle Moon (Host) (01:05:58)

You've built a tool to help people prepare for exams, and it uses AI.

Ran Chen (Guest) (01:06:02)

Yes.

Uncle Moon (Host) (01:06:03)

So going forward, this is sort of a transitional stage. In the future, do you think people will still need these exams?

Ran Chen (Guest) (01:06:08)

I don't know. Nobody knows what the future holds. But I think, because people are facing more and more competition, if someone wants to enter a particular profession, how do you decide who gets in? The traditional approach has been professional qualifications, across different industries. So I don't know whether, in the AI era, there will be even more certifications, with people trying to find a job by earning credentials to improve their prospects. But how should I put it? For my business, will that broader trend affect me? I don't think it matters that much. The approach I'm taking is very similar to what we did at Tubi. It's free, and we mainly make money through ads. Tubi offered free movies and TV shows, with advertising revenue supporting it. So we're approaching this product from a slightly different angle.

Uncle Moon (Host) (01:07:06)

You even publish the project's monthly revenue on your website.

Ran Chen (Guest) (01:07:09)

Yes.

Uncle Moon (Host) (01:07:09)

About $10,000 a month.

Ran Chen (Guest) (01:07:11)

Right.

Uncle Moon (Host) (01:07:11)

And it's all passive income? You don't have to run it yourself at all?

Ran Chen (Guest) (01:07:15)

I check on it every day. But it doesn't really need much attention.

Uncle Moon (Host) (01:07:21)

Does this project follow your advertising-only model?

Ran Chen (Guest) (01:07:24)

Yes, it's just advertising. There are also some third-party links, like buying books on Amazon or going to paid exam-prep platforms. We get some revenue from those too.

Uncle Moon (Host) (01:07:35)

What if someone sees your website and copies it in three days with AI coding tools? What do you do then?

Ran Chen (Guest) (01:07:40)

Right. That's a very good question. I think everyone in business faces this question. Take a bottle of water, for example. Everyone's water is the same. There's nothing special about it. So why can different brands charge different prices? And ultimately, where does the competition between them lie? I think there are a few factors. First, whoever gets there earliest has an advantage. Second, it eventually comes down to supply-chain management. Can I use fewer tokens to give you better results? That tests my own understanding of AI tools.

Ran Chen (Guest) (01:08:19)

Someone else might burn ten times as many tokens, or pay ten or a hundred times as much for those tokens. Or they might have to hire a full-time employee to do it. Their business wouldn't be viable. But in my case, I can find cheaper tokens, try different models, and figure out their limits. Then I can provide the service at a lower cost.

Ran Chen (Guest) (01:08:41)

It's the same idea with something like Tubi TV. Why hadn't anyone done it before? Or at least, done it at Tubi's scale? I think, for Tubi TV, several factors came together: the cost of engineers, the cost of content delivery networks, and also the quality of our engineering talent, the maturity of the tools, and the inventory of free movies and TV shows. The amount of content you could acquire for free. They all converged at just the right moment, making a business that wasn't viable before suddenly viable. And once that happened, the position you secured was yours to keep.

Uncle Moon (Host) (01:09:19)

So, having been involved in so many startups, you never really worry about competition? Any profitable business is bound to attract it.

Ran Chen (Guest) (01:09:26)

Of course it is. Just compete. If you do it better than everyone else, all the better. If you can't win, move on. That's always my philosophy: can't win, move on. Then find the next project. After all, there are plenty of things you can do.

AI Music, Finding Projects, and Underserved Opportunities

Uncle Moon (Host) (01:09:41)

That's right. Then you started another side business: a zero-person AI music company. What exactly is that?

Ran Chen (Guest) (01:09:48)

That wasn't actually my second one. It was when AI... When AI music first appeared, there was Suno, that model. You could use it to make music. At the time, I thought, "There are so many songs I could make." So I made a bunch of songs and uploaded them to Spotify, Apple Music, and so on. After that, though, I didn't really do much with it. And around the end of last year, I stopped working on it. The feedback loop in the music industry, the time it takes to see positive feedback, is too long. You put in all that time and don't see results. So I left it alone. I don't think it's worth a person's time. You can't make money from it. But if you can get it to run automatically, that's fine. It just won't make much money. Right. I think that project currently brings in a little over $100 a month.

Uncle Moon (Host) (01:10:36)

Enough to buy yourself a coffee.

Ran Chen (Guest) (01:10:38)

Exactly. But the funny thing is, it's growing every month. I have no idea how long that'll continue. I just don't know if this is the kind of business that keeps growing on its own.

Uncle Moon (Host) (01:10:50)

Then you built another zero-person company, with AI automatically finding underserved opportunities for worthwhile zero-person businesses.

Ran Chen (Guest) (01:10:57)

Yes.

Uncle Moon (Host) (01:10:57)

Where does that stand now?

Ran Chen (Guest) (01:10:59)

I wouldn't even call it a zero-person company. It's just looking for projects. That's really all it is. I might set different criteria, based on what I think could make money, and have AI track those opportunities. It searches through APIs for different projects, constantly scanning to see if there's an opportunity. If there is, we build it first, put it out there, and see whether it makes money. If it doesn't, that's that. Then we feed that result back to the AI: did this work or not? If it does make money, we tell the AI that it worked. With that real-world feedback, it gets better at identifying projects.

Uncle Moon (Host) (01:11:37)

So you've built a workflow for collaborating with AI to create zero-person companies.

Ran Chen (Guest) (01:11:44)

Yes. It's really just a way, in line with my philosophy, to look for possible projects in underserved markets. But we haven't actually found that many, maybe ten or twenty. We're building them one by one. Most probably won't work out. That's how I see it. If it works, it works. If it doesn't, it doesn't. After all, these days, tokens are about the cheapest thing around. The point is to use those tokens and learn something about the world. That's how I think about it.

Uncle Moon (Host) (01:12:11)

You're taking the cheapest resource and learning how to make money in the real world.

Ran Chen (Guest) (01:12:15)

Yes. I think that's great. There's one project I remember that's still running. It's quite interesting. It's called Second Chance Info. What is it? When I was looking for opportunities, it struck me that in America, there are a lot of people who are homeless, or have a history of drug use, for example. They need help finding work and all sorts of other things. They need a second chance, a fresh start. They might need a website that brings together resources for health care, housing, and so on. So I wrote code to scrape a lot of information online and built the website.

Ran Chen (Guest) (01:12:53)

Of course, I don't really promote it. It mainly relies on organic traffic. You can see people searching there every day. Some people who use it even email to thank you. They say they used my website and found really useful information. They're very grateful. That makes me happy too. And with something like this, once you've built a website, by yourself, you can just leave it running at virtually no cost. So I leave it there and contribute something to the world. If one day it gets a lot of traffic, I might consider putting ads on it to make some money.

Uncle Moon (Host) (01:13:26)

You make it all sound so effortless. What would you say is the biggest challenge?

Ran Chen (Guest) (01:13:31)

Here's how I see it. The biggest challenge, I think, is that our understanding of the world is limited. I don't really understand each industry's pain points. So a lot of the products we build aren't good enough, I think, because I don't truly understand those problems. I can only build based on my own assumptions. Why do I think Open Exam Preparation has done well? Because I felt that pain myself when taking exams. I took so many exams, and it was frustrating. I couldn't find the material I needed, or get answers to my questions. Neither the forums nor the instructors could explain things as well as AI. So why, when you're building something and there's a long stretch with no feedback, do you keep going? Because I truly understand that pain point. I believe there's real value in building this, so I'm willing to put in the time.

Ran Chen (Guest) (01:14:18)

But in lots of other industries, even when I build something, it's hard to get positive feedback, because I don't know their real pain points. So I don't really know, not fully, how to improve the product. Sometimes it's just a seed. You scatter it and see how it grows. If it grows well, you harvest it. If not, you let it go.

Doing Business: Find the Demand and Do the Math

Uncle Moon (Host) (01:14:36)

You've worked on so many zero-person projects. It feels to me like you've grasped the essence of doing business. Could you share some of your business secrets?

Ran Chen (Guest) (01:14:47)

Okay. I think that's a great question. It reminds me of something. For a while, recently, I was obsessed with watching Yong Ge Talks Restaurants. Have you seen it?

Uncle Moon (Host) (01:14:56)

No.

Ran Chen (Guest) (01:14:57)

You haven't? It's huge on short-video apps. Those video platforms. There's this guy called Yong Ge. People join his livestream and ask him to look at how their restaurants are doing. Then he gives his assessment. "Turn around and show me a full 360," right? What I find interesting is that he gave me a framework for thinking about what actually makes money, and the mistakes people commonly make.

Ran Chen (Guest) (01:15:21)

I think in Silicon Valley, or really anywhere, it's much the same. Why have those people calling Yong Ge failed? What's the most common reason? "I like bread, so I opened a bakery." Or, "I wanted to franchise Mixue, but I couldn't find their official website. So I made a phone call and ended up talking to some other company." Or someone makes them a bunch of grand promises: "Open this and you're guaranteed to make money." And they wire over 200,000 yuan without thinking. Or, "I don't know where the customers are, but if I do a good job marketing online, post videos on Xiaohongshu, the customers will come." I think that's very similar to how a lot of us approach startups.

Ran Chen (Guest) (01:16:08)

So what does Yong Ge do? First, he has them show him the surroundings. Then he asks, "Where's the foot traffic?" You have to find a place that already has customers. First of all, there has to be a real, existing demand. You can't start a business assuming your personal charm will bring in customers. You can't count on becoming an influencer. You need somewhere with actual traffic. So with many of my businesses, I look for that too. For instance, I use Google search APIs, third-party APIs, to look at Google search keywords and their rankings. What are people searching for that doesn't have good results? Instead of just dreaming up what I think might work, I look for where the people are.

Ran Chen (Guest) (01:16:53)

Once you've found that traffic, you have to decide what to sell. That doesn't mean liking cake and opening a cake shop, liking bread and opening a bakery, or liking bubble tea and opening a bubble tea shop. No. What do those people want? That's what you offer. I remember Yong Ge often saying, "Those two people in your shop, are they your employees? Fire them first." Why? Because making money means doing the math. You can't just go and hire... Just because you're doing something doesn't mean you need all those people. You, the owner, have to get your hands dirty. You have to work out what it actually costs and how you're going to make money.

Ran Chen (Guest) (01:17:33)

So, to me, a real business comes down to this: find a place with real demand, give those people what they want, and do the math. How much revenue do you bring in? How much do you spend? Then you need to optimize a lot of things, like your supply chain and so on. A lot of the time, there's a kind of startup where you build something nobody else has. And that's great. You raise funding. You don't even have to make money. But if you actually want to run a profitable business, that's a completely different path. You still have to come back to the numbers. What's your competitive advantage? How do you optimize your supply chain? How do you find what real people need? Those are, to me, when you're building something that actually makes money, the most important questions. So watching those videos was really eye-opening for me.

Uncle Moon (Host) (01:18:21)

And then you put it into practice.

Ran Chen (Guest) (01:18:23)

Yes, so much of what he says matches my thinking. It made me realize, "Yes, that's exactly it."

Uncle Moon (Host) (01:18:29)

You independently reached the same ideas.

Ran Chen (Guest) (01:18:29)

Exactly, the same ideas. And when you see the mistakes those callers make, you recognize them in other people. People come to me and ask, "I'm thinking about making some money. I'm not trying to raise funding or launch a startup. I want to build something profitable. Would this approach work?" It feels a lot like those calls with Yong Ge. Objectively, it won't work, because they haven't properly researched where the demand is. They've just imagined a need. They simply assume, "If it tastes good, people will buy it." That's often not how it works.

Two Paths: Running a Business and Launching a Startup

Uncle Moon (Host) (01:18:59)

You also mentioned that this differs from a lot of startups in Silicon Valley. Running a business and launching a startup can actually be two different things.

Ran Chen (Guest) (01:19:06)

Yes.

Uncle Moon (Host) (01:19:07)

With a startup, you might be creating demand that doesn't yet exist, or, in the short term, have no revenue at all.

Ran Chen (Guest) (01:19:13)

Right. Exactly. That's why I think Silicon Valley has a bit of a status hierarchy. People think, "I want to build something nobody has ever built," or, "I want to raise a huge amount of funding," and do something really fancy, something nobody else has. And that's great. If that's the path you choose, you need to commit to it. On the other hand, if you want to run a profitable business, that's a completely different path.

Ran Chen (Guest) (01:19:44)

If that's what you want, you have to respect the basic principles of running a profitable business. Do it well, and you can make money, possibly a lot of money. So I think, given how smart people are, once you know what you want and keep improving, you can do very well. Just don't mix those two paths up. If you do, you're in trouble. You can't... You can't have it both ways. They're different things.

Uncle Moon (Host) (01:20:07) †

When you've been in Silicon Valley a long time, you see all kinds of examples of people confusing the two. Some things need long-term investment. Others should start making money right away. But when you blur the two, you end up achieving neither. Neither works.

Ran Chen (Guest) (01:20:20) †

Exactly.

Returning to Work in Health Care and Life Sciences

Uncle Moon (Host) (01:20:21)

Then, from 2025 onward, you returned to work after being a stay-at-home dad and helped found two companies.

Ran Chen (Guest) (01:20:28)

Yes.

Uncle Moon (Host) (01:20:28)

Could you tell us that story?

Ran Chen (Guest) (01:20:30)

In 2025, a good friend and his team happened to come to San Francisco for the J.P. Morgan conference. Their company is called Pure Global. It handles global registration for medical devices. And we... He was talking with us about advances in AI, and his hopes for transforming the company with it. I thought it was a great opportunity. The company was founded in 2020 or 2022, I think. It had been around a few years. I joined as CTO, to help the whole company make that AI transition.

Ran Chen (Guest) (01:21:03)

I think we made some pretty big changes. After those changes, the company's business operations and ability to deliver improved significantly. Then we hit some new bottlenecks. I think, once delivering the work is no longer a problem, customer acquisition becomes a major challenge. So we incubated another company called VayoMed. Its main focus is helping life sciences companies grow. On the one hand, we obviously need to solve our own customer acquisition problem. On the other, we want to help more companies like ours in the life sciences industry solve the same problem.

Ran Chen (Guest) (01:21:40)

Because we see that, with AI advancing at this incredible pace, once we reach a point where producing something is no longer a problem, finding customers becomes very difficult. So there are definitely things where the earlier you start, the better. That's also why we started this. So I'm currently the CTO of both companies.

Uncle Moon (Host) (01:22:01)

You started as a machine learning engineer, and now you've moved into health care.

Ran Chen (Guest) (01:22:06)

Yes.

Uncle Moon (Host) (01:22:07)

Into that industry.

Ran Chen (Guest) (01:22:08)

Right.

Uncle Moon (Host) (01:22:08)

What led you to make that choice?

Ran Chen (Guest) (01:22:11)

I've always worked on applied machine learning. That has never changed. It's what I truly love doing: taking AI, machine learning, and cutting-edge algorithms into the real world. At CMU, I worked on a legal project. My first job, at Trulia, was in real estate. Then at Tubi, it was online streaming, including recommendations, search, and so on. Now it's health-care regulatory compliance, mostly working with text, and customer acquisition too. To me, they're all practical applications of AI.

Ran Chen (Guest) (01:22:47)

The only real difference is that in an industry like life sciences, the regulatory burden is heavier. So there's some specialized know-how and methodology that you need to learn and master. But here's one view I have. AI's impact is so enormous that competition in many industries will be brutal. So industries with higher compliance costs, like finance, health care, and law, might offer more durable opportunities. That was part of my thinking.

How AI Transforms a Company

Uncle Moon (Host) (01:23:22)

As you've been building this company, you've really felt AI boosting your productivity.

Ran Chen (Guest) (01:23:28)

Yes.

Uncle Moon (Host) (01:23:28)

Could you share a few examples from your perspective as CTO?

Ran Chen (Guest) (01:23:32)

Sure. We saw some very obvious changes. As I mentioned, I joined in 2025. In 2025, Anthropic released Sonnet 3.5, two versions, I think, and then 3.7. That was around the middle of 2025. They also released Claude Code, the tool that runs in the command line. Before, people generally needed a user interface. This ran in the command line. When we tried it, we thought, "That's it." From now on, work would definitely revolve around this.

Ran Chen (Guest) (01:24:04)

Before that, everyone used ChatGPT. You had to copy and paste. It didn't feel like part of a workflow. It was a kind of assistant, or not even an assistant, because it couldn't actually do things for you. It was just something to spark ideas. But once it really became part of the workflow, we could feel it: "That's it." So many things were definitely going to change. So in 2025, we made some big changes within the company. There are a few I can share.

Ran Chen (Guest) (01:24:35)

First, we had some SaaS products with annual subscription fees. We felt that SaaS no longer made sense. We just gave the source code to some of those customers. "We're not doing this anymore. Here's all the source code. It's yours." And we made some of our resources publicly available online, like databases people could use for free. We used to charge for them. Now they're a customer acquisition channel. I thought the times had changed. Charging for them no longer made sense.

Ran Chen (Guest) (01:25:02)

Then, when I came in as CTO, I took our internal systems and rewrote them all, using AI. I dropped a lot of features that seemed pointless, spent a lot of time talking to customers, and kept only the few things they actually needed. It turned out that what customers wanted was much simpler than what we'd been offering. A lot of features existed because a team thought, "We should build this." They built them to justify their own existence. But when you can rewrite everything, you find there are very few things that truly matter. And with AI helping as well, once we'd rewritten everything, it all fit together much better. Whether it was AI using the system or our own employees, it was easy to work with.

Ran Chen (Guest) (01:25:48)

Then we spent a long time building a knowledge base, an internal one. That completely changed how the entire company works. Today, everything revolves around that AI knowledge base. If anyone has a question, we say, "Stop talking. Don't ask each other. Whatever you need, just ask the AI." If the AI gives a bad answer, we fix the knowledge base. If it answers well, great.

Ran Chen (Guest) (01:26:13)

The benefit is that the company now has something that accumulates knowledge and grows. We didn't have that before. All that knowledge used to live in different people's heads. Now we have this knowledge base that keeps growing and improving. And because it exists, every new hire can quickly get up to speed and get things done in the company. Of course, how you build that knowledge base is a difficult question in itself.

Ran Chen (Guest) (01:26:37)

The last thing is, because we help clients register medical devices around the world, we have nearly 20 offices globally, including Brazil, Mexico, Southeast Asia, the Americas, and Europe. We have offices in all those places with local staff. We help clients register products locally, in Brazil, for example. So the project teams have to work locally. And there's a lot of documentation. Once we introduced AI tools internally, our document-writing speed shot up.

Ran Chen (Guest) (01:27:07)

We used to have a backlog of maybe 50 or 60 projects waiting to be done. Now we've reached the point where there's no work left in the queue. That's why we say our biggest problem is not having enough customers. Because our delivery capacity—our ability to produce—now far exceeds our ability to acquire customers. So pretty much everyone on the team is focused on how to bring in more customers. That's our one big priority: getting more customers.

Ran Chen (Guest) (01:27:33)

And our pricing has changed dramatically, too. In our industry, if you want an analogy, it's a bit like moving abroad and applying for a green card. There's a mountain of paperwork. Traditionally, this industry billed by the hour, with two components to the fees: a large registration fee in year one, then an annual fee after that. That's how it worked. We took all our prices—and pricing used to be totally opaque—and became the first company to publish prices for every service online. We also stopped charging registration fees. There's just an annual fee.

Ran Chen (Guest) (01:28:14)

And that's because our own cost of handling registration is now extremely low. So the result is, people in our industry sometimes struggle to understand what we're actually offering. "What do you mean, registration is free?" Free means free, right? Other companies, for example, might charge 5,000 for registration in year one, plus an annual fee of 3,000. We just charge 2,000 a year. It's 2,000 in year one, and 2,000 in year two. And people ask—people in the industry, a lot of them ask—"What does that mean? You don't charge for registration anymore?" That's right. It's free now. That's all there is to it. Just 2,000 a year.

Ran Chen (Guest) (01:28:50)

So I really think AI has completely changed our business model. And, if you search within our industry for the price of these registration services, we're still pretty much all you'll find. We're almost the only company that publishes its prices online. And it's a flat fee: a fixed price, not one that varies. So I think AI has completely transformed every part of our company. It's also improved how efficiently we operate, and everyone's happier, to some extent. And of course, we've also changed some of our incentives—our incentive compensation. We've changed how we pay people, too. So AI has had an enormous impact.

Uncle Moon (Host) (01:29:31)

Listening to all that, it sounds like AI has completely rebuilt your company. Some people might focus on how you've driven prices down. You've made them so low, isn't that a race to the bottom? But then you also said everyone at the company is happy.

Ran Chen (Guest) (01:29:44)

Right.

Uncle Moon (Host) (01:29:44)

And their working hours haven't really increased much. You were just early adopters of AI,

Uncle Moon (Host) (01:29:50)

and captured the benefits of this wave.

Ran Chen (Guest) (01:29:53)

Mm-hmm.

Uncle Moon (Host) (01:29:54)

And you got started early, back in 2025, last year, using Sonnet 3.5 and 3.7. So well before the people using OpenClaw, you'd already realized that Claude Code was a core productivity tool.

Ran Chen (Guest) (01:30:08)

Exactly. At the time, we were still hands-on, writing code every day. So we really understood just how powerful this was. Once it moved into a CLI, into a terminal, what that meant for workflows—the implications of that—were immediately obvious.

Uncle Moon (Host) (01:30:26)

I think a lot of developers in Silicon Valley didn't need OpenClaw to realize this. It probably hit them with 3.5 or 3.6. "Oh my God, is building an app already this easy?"

Ran Chen (Guest) (01:30:37)

Exactly. And with this whole AI-based approach to workflows, the challenge today is putting it into practice. It's not the models' capabilities. You might need help implementing it, but in many cases, the models are already more than capable.

Ran Chen (Guest) (01:30:54)

In other words, the work most of us do really isn't that difficult. It just isn't. The difficulty of what we're doing is below what these models can handle. So once we plug a model in, a lot of work can be replaced.

A Technological Wave, Organizations, and Competition

Uncle Moon (Host) (01:31:11)

So whether you're looking at it as a hands-on engineer or as a CTO, you've experienced firsthand this enormous shift in our times. And at your own company, you've led the way through the whole process.

Ran Chen (Guest) (01:31:23)

Yes.

Uncle Moon (Host) (01:31:23)

Which makes you quite cautious about people's career prospects going forward.

Ran Chen (Guest) (01:31:26)

Yes. I always tell people: we're living in the age of horse-drawn carriages, but the car has already been invented. Sure, maybe the car isn't fully developed yet, or it wobbles around when you drive it. But plenty of people are already driving cars all over the world. Meanwhile, the carriage crowd is still inside a castle. A carriage can only take you so far, so everyone lives within that area. But cars already exist. So let's stop debating carriages, or whether cars are dangerous, and all that. There's no point. The people who've taken off are already a long way ahead. And carriages will eventually disappear.

Ran Chen (Guest) (01:31:58)

The question is how we envision that world today, and find better sources of information. That's why I really like your interviews. You're out there on the ground, interviewing people in Silicon Valley. It's not secondhand information. It's firsthand. That's often incredibly important. People need that firsthand information to understand the changes actually happening in the world.

Uncle Moon (Host) (01:32:19)

Follow Uncle Moon for firsthand insights from Silicon Valley.

Ran Chen (Guest) (01:32:22)

Exactly.

Uncle Moon (Host) (01:32:23)

Thank you, everyone. And thank you, Ran. Your whole story also brought home another big point. Anthropic's CEO said something recently: once writing code is easy, reviewing and shipping it start to become the bottlenecks instead. You might code five times faster—

Ran Chen (Guest) (01:32:39)

Right.

Uncle Moon (Host) (01:32:39)

—but overall productivity doesn't go up fivefold.

Ran Chen (Guest) (01:32:41)

Right.

Uncle Moon (Host) (01:32:41)

Because coding accounts for only 20% of everything from product to acquisition, distribution, and even after-sales support. So if coding gets five times faster, the overall improvement is 27%, I think.

Ran Chen (Guest) (01:32:50)

Right.

Uncle Moon (Host) (01:32:50)

Something like that. So you quickly recognized that, once those productivity constraints are no longer the problem, you need to focus on acquiring customers and getting other things done.

Ran Chen (Guest) (01:32:57)

Right. In the situation you described, it's often an organizational issue. The organizational structure hasn't fully adapted to that level of productivity. For example, there are all these layers in the hierarchy, and all these roles, like managers and directors. Those positions used to be necessary. But organizations may get flatter and flatter. That's inevitable. Some jobs exist just to pass information along or get everyone into meetings. They may not be necessary anymore, because AI can gather that information itself. So organizations will also undergo major structural changes. All of this will happen together.

Ran Chen (Guest) (01:33:35)

And I think there's another really important point. A lot of the time, people complain about this, about how their organizations are changing. But sometimes I feel that those complaints miss the point. AI-driven productivity is advancing so quickly and on such a huge scale, these companies will soon face a new reality: whether they transform or not, a wave of startups will emerge that can match what they do at 10% of their cost, or even 1%, and build exactly the same thing.

Ran Chen (Guest) (01:34:19)

That's why I keep saying this shift will be bigger than people expect. I'm seeing countless AI-native products springing up. These people really are doing big things at very low cost. At some point, they're bound to go head-to-head with those companies.

Uncle Moon (Host) (01:34:39) †

Right.

Uncle Moon (Host) (01:34:39)

I completely agree. I've seen these companies, too. A lot of startups in San Francisco now

Ran Chen (Guest) (01:34:44)

—right—

Uncle Moon (Host) (01:34:45)

pull up the businesses of the top 500 Nasdaq-listed companies and target them one by one.

Ran Chen (Guest) (01:34:50)

Exactly.

Uncle Moon (Host) (01:34:50)

You handle payments and billing?

Ran Chen (Guest) (01:34:55)

Right.

Uncle Moon (Host) (01:34:55)

Then I'll build AI-native billing.

Ran Chen (Guest) (01:34:57)

Right.

Uncle Moon (Host) (01:34:57)

You do customer acquisition? I'll do AI-native customer acquisition.

Ran Chen (Guest) (01:34:59)

Exactly.

Uncle Moon (Host) (01:34:59)

You work in legal services? I'll build an AI-native legal assistant. They're going after them like that.

Ran Chen (Guest) (01:35:04)

Exactly.

Uncle Moon (Host) (01:35:04)

And on the one hand, these products really are AI-native. They really reflect this era's productivity and ways of organizing work. On the other hand, today's 17- and 18-year-olds, 18- and 19-year-olds, are much more inclined to use these tools than those stuffy, outdated products from the previous era.

Ran Chen (Guest) (01:35:17)

Exactly.

Uncle Moon (Host) (01:35:18)

So building more AI-native products at a lower cost is already an unstoppable trend. A lot of people haven't felt it yet, perhaps because, for one, this wave from Silicon Valley hasn't hit them.

Ran Chen (Guest) (01:35:33)

Right.

Uncle Moon (Host) (01:35:34)

Or the wave from Shenzhen and Shanghai. But sooner or later, my hometown of Qingdao, and all kinds of other places—

Ran Chen (Guest) (01:35:40)

Yes.

Uncle Moon (Host) (01:35:40)

It's only a matter of time. It's one big wave, spreading out bit by bit.

Ran Chen (Guest) (01:35:46)

Exactly.

Uncle Moon (Host) (01:35:46)

People here in Silicon Valley, right at the center of this wave, are already stunned by it.

Ran Chen (Guest) (01:35:52)

Exactly. Take us, for example. We've finished building our internal tools. Our next step is to go to our vendors, our service providers, and ask them one by one: "Can you lower your price? If not, I'll just build it myself." So those providers will be under enormous pressure, too. Say they used to charge 5,000 a month. If I build my own version, I might finish it pretty quickly. If they're not adding much else, they'll lose me as a customer. And they'll start losing lots of customers. So every company is facing a huge disruption.

Uncle Moon (Host) (01:36:24)

This isn't a race to the bottom. Some viewers may say it is, but personally, I think humanity's relentless pursuit of productivity will continue, whatever anyone thinks.

Ran Chen (Guest) (01:36:34)

Exactly. Think about it: so many things will get cheaper. Take my Open Exam Preparation platform. You used to have to pay for that. Now I can offer it to everyone for free. That's productivity improving.

Founder Brands and Investing in AI Tools

Uncle Moon (Host) (01:36:46)

I'm sure Ran's thoughts on building businesses in the AI era have given everyone food for thought. So let's move on to our final chapter: looking ahead, or reflecting on life.

Ran Chen (Guest) (01:36:58)

Sure.

Uncle Moon (Host) (01:36:59)

You've written more than 1,000 blog posts. You're already quite a prominent online personality. But you've repeatedly said on your blog: "If you want to make money, don't build a founder brand."

Ran Chen (Guest) (01:37:11)

Mm-hmm.

Uncle Moon (Host) (01:37:11)

Why is that?

Ran Chen (Guest) (01:37:14)

That post, yes. I think I wrote it after watching Yong Ge. That really left an impression on me. What I mainly wanted to push back on was this idea—I wanted to distinguish startups from making money. Building a venture-backed startup and doing something to earn money are two different things. Building a founder's personal brand, putting yourself in front of the camera, is often better suited to raising venture capital and pursuing other goals. But if you simply want to build something profitable, you usually need to spend more time finding demand and meeting that demand.

Ran Chen (Guest) (01:37:48)

Now, does putting yourself out there as the founder help at all? It can help with acquiring customers. But it's not the most fundamental thing. So what I'm really saying is, let's not spend so much time on that. Let's turn our attention back to customers, solve their problems, and think about how we can add more value for them. I think that's more important.

Uncle Moon (Host) (01:38:11)

So that's a clear distinction: a founder brand helps you raise venture capital.

Ran Chen (Guest) (01:38:15)

Right.

Uncle Moon (Host) (01:38:15)

But to build a profitable business, understanding demand and where customers come from is what matters most.

Ran Chen (Guest) (01:38:20)

And people often come across this concept of "building in public." But the people you see building in public and making money from it are doing that because the things they build are for developers in the first place. So building in public attracts their customers.

Ran Chen (Guest) (01:38:41)

But for many businesses, if the people you're trying to reach, your customers, aren't developers, then building in public isn't actually that helpful. It doesn't attract your real customers, in many cases. Right. So that's another way to look at it: what you're doing, and how you approach it.

Uncle Moon (Host) (01:39:00)

These days, you use tens of billions, even 30 or 40 billion tokens a month. How much do you spend on these AI tools each month?

Ran Chen (Guest) (01:39:08)

I think it's only around $1,000, maybe. Very little. Why is that? Because there's a huge opportunity right now with coding plans—what they call coding subscriptions. A lot of these companies want to attract high-value developers and use their activity as training data, I suppose. So they make it really cheap. For example, OpenAI's $200 plan can give you 10 to 20 billion tokens a month. That's 10 to 20 billion tokens available to you. That would be very expensive through the API. But on their subscription, it's very cheap.

Ran Chen (Guest) (01:39:46)

And lots of companies offer plans like that. So I subscribe to a bunch of them: Anthropic, OpenAI, Google's Gemini, and also Ollama, a third-party cloud platform. I subscribe to that, too. I basically get the top tier for all of them, because they're so heavily subsidized. So for me, I see it as like China's food-delivery wars back then. They burned cash to subsidize your takeout. In America, it's a token war. They're burning cash to subsidize your token usage.

Ran Chen (Guest) (01:40:16)

So the more I use, the more value I get. That's why, as long as a model meets my minimum standard, I'll subscribe if I can. Right away. I'm always one of the first to sign up, then I share how well it works.

Uncle Moon (Host) (01:40:32)

I was going to ask you for advice for people hesitant to spend that much, but you've already given it: the more you subscribe to, the more value you get.

Ran Chen (Guest) (01:40:40)

Exactly.

Ran Chen (Guest) (01:40:40) †

Just take the savings, right?

Uncle Moon (Host) (01:40:42) †

That's all there is to it.

Ran Chen (Guest) (01:40:43)

And when these companies go public, they'll face financial pressure. They might cut back on those subsidies. So this is a very short-lived window of opportunity.

Agency, Money, and Choosing Your Life

Uncle Moon (Host) (01:40:51)

Looking back on your life today, from small-town exam ace to Tsinghua, then CMU, then two startups that were acquired, then being a stay-at-home dad, and now working hands-on as CTO at startups and running zero-person companies, why do you think you've consistently achieved such good results? What's behind that?

Ran Chen (Guest) (01:41:11)

Well, whether the results are good is really an outside judgment. If I look up the ladder, some classmates reached financial freedom earlier, or built more successful startups. But for me personally, I think I'm someone who has a fairly clear sense of what I want. And I ask myself whether I've reached the goals I set for myself. If I have, that makes me happy. The key is to do things I find interesting, and enjoy doing them.

Ran Chen (Guest) (01:41:40)

I get bored easily. But when I'm bored, I go and do something interesting. That's a big part of my philosophy of life.

Uncle Moon (Host) (01:41:48)

Let boredom and curiosity drive you to keep trying things.

Ran Chen (Guest) (01:41:51)

Exactly. I didn't know this term before, but through Chinese podcasts, I learned the word "agency." I think it's a great word. It means we look at life from our own perspective. Take work, for example. It doesn't define your life. It's not some objective reality, right? It's just there to serve the life we want. So what matters more is what kind of life we want to live. Then, based on that vision, you choose where to work, how to relate to money, and how to raise your children. All of those choices start with your own perspective on your life.

Ran Chen (Guest) (01:42:32)

Don't assume that just because you have a particular job, you're a certain person. Your job doesn't define you. Objectively speaking, if you look at history, jobs haven't been around that long. A few decades ago—maybe a century or two ago—the concept of a job didn't even exist. And a few decades from now, it might be gone. So the things we think are important today aren't important in the sweep of history. The only things that really matter are getting your 2,000 calories a day, and getting your sleep. Those might be the only essentials for staying alive. Everything else is extra. So explore things, give them your best, and do things that make you happy.

Uncle Moon (Host) (01:43:11)

That's true. Maybe jobs are just a fleeting moment in human history.

Ran Chen (Guest) (01:43:16)

Right. Working for a big tech company—in a few decades, people might say, "What the hell was that? Why were all those people calling each other idiots?" What kind of phenomenon was that, back in the internet age? Who knows?

Uncle Moon (Host) (01:43:28)

After all these experiences, when you were making each decision, did you ever worry, "If I fail this time and run out of money, what happens to my family? What happens to me?" Did you ever feel that fear?

Ran Chen (Guest) (01:43:39)

I'd be lying if I said I didn't. How should I put it? For example, part of why I studied financial planning was to really explore our relationship with money. Questions like, how much money, how much income is enough?

Ran Chen (Guest) (01:43:50)

And something really stuck with me. In financial planning, one of the most important areas is retirement planning. The reason you don't see many people around you getting financial advice is that most people seeking it are planning for retirement. Very few planners focus on young people, because young people don't buy into it. They think, "This person doesn't matter to me. What they say doesn't matter."

Ran Chen (Guest) (01:44:14)

So I met a lot of older people. And talking about money with them radically changed how I thought about and understood money. Right. So seeing more of the world, seeing ways you might live in the future, how people might live when they're older—imagining life with the end in mind can really change the way you think about a lot of things.

Uncle Moon (Host) (01:44:38)

Could you tell us more about starting with the end in mind? After meeting so many older people, how do you think we should live when we're young?

Ran Chen (Guest) (01:44:44)

When you talk to a lot of older people, you discover something. The most important thing, I think, is this: the things that seemed so important when you were young don't matter at all.

Uncle Moon (Host) (01:44:57) †

Not at all.

Ran Chen (Guest) (01:44:58)

When you ask them what they did when they were young, what they regret, or what they want to do with their money, you realize that the things you're anxious about now, that feel so important, really aren't.

Ran Chen (Guest) (01:45:10)

And that pushes you to ask yourself what truly matters. It's doing what you want to do, living a life you've actively chosen, and living it on your own terms. I think that's what matters most.

Full-Time Work, Pay, and New Ways to Organize

Uncle Moon (Host) (01:45:25)

Four words really stand out: "Make your own choices." Don't just passively accept things. We talked about agency, and how older people view the work that young people do. Even the way large companies, or any companies, are organized will keep changing. You've also kept emphasizing that future companies may not need employees. They may only need to assign contractors or agents to complete tasks. Could you explain that in more detail?

Ran Chen (Guest) (01:45:52)

Sure. Here's how I see it. For a lot of people, their fear of AI comes down to worrying about layoffs, and whether full-time employees will be affected. But I look at this from a different angle. I think full-time employment today actually puts limits on people's productivity and the way they work. It's an unhealthy way to organize labor, in some respects.

Ran Chen (Guest) (01:46:23)

Let me give you an example. If you work directly with a customer and provide a service, and that customer pays you 10,000, you're delighted, right? You want to serve them well. But if you work for a boss, and your boss pays you 10,000, you feel like they're exploiting you. You think you're worth 20,000. And once you think that way, you're certainly not thinking, "I want to do more work." What you're thinking is, "I want to slack off. How do I make this pay worth my while?" You're not thinking, "How can I make sure I do a great job?"

Ran Chen (Guest) (01:46:54)

So what creates that mindset? I think it's the compensation, the way pay is structured. Because we don't get to share in the extra income that greater productivity generates, we feel no desire to work more. But why does AI change this so much? Because AI can genuinely make someone who wants to work 10 or even 100 times more productive.

Ran Chen (Guest) (01:47:21)

That's why I think pay structures and employment arrangements have to change, too. It's why I'd like to see more arrangements like partnerships. However much you earn, if you earn more, everyone shares in it. Only then will people think, "I want to use better tools. I want to earn more money. Because if I do, I get a bigger share, too." That gives me an incentive, a stronger incentive to earn more.

Ran Chen (Guest) (01:47:47)

Instead of thinking, "I use more AI tools, and save more time, but it seems like the bosses, the shareholders, get all the productivity gains. I'm just an increasingly miserable workhorse, and eventually I'll get replaced anyway." So what I'd really like is for us to find a better way to organize work, so that everyone feels more empowered, and everyone can share in the gains, the extra income from greater productivity. I think that's the most important thing.

Uncle Moon (Host) (01:48:16)

Other programmers think they'll be replaced.

Ran Chen (Guest) (01:48:19)

Right.

Uncle Moon (Host) (01:48:19)

They feel afraid. You see a new era of reform and opportunity for programmers.

Ran Chen (Guest) (01:48:24)

Right. As long as working arrangements change for the better, I think everyone will have more incentive to earn more money, and to provide more value. I think that's the most important change.

Uncle Moon (Host) (01:48:34)

I'll have to listen to this part again. It really shows how everything has both positive and negative effects.

Ran Chen (Guest) (01:48:41)

I'd really like to see this transformation happen, so people have more incentive to do more.

Uncle Moon (Host) (01:48:46) †

A bit like the big law firms in America. You start out taking on cases, and work your way up to senior partner. But on every case, you share revenue with the firm.

Ran Chen (Guest) (01:48:55) †

Exactly.

Uncle Moon (Host) (01:48:56) †

I think in the future, this really might be the direction employment and company structures—

Ran Chen (Guest) (01:49:01) †

Yes.

Uncle Moon (Host) (01:49:02) †

—will move toward.

Advice for Different Stages of Life

Uncle Moon (Host) (01:49:04)

So, given that vision of future work and organizational structures, what should today's college students, aged 18 to 22, be doing to prepare?

Ran Chen (Guest) (01:49:14)

Well, first of all, this really is an era of enormous change. People will feel lost, frustrated, and uncertain. But if you look at history, at every stage, people in every era have faced those same problems. Take my generation. If you tell people that computers and software were considered sunset industries, nobody believes you. There was no consensus back then, either. People had no idea why the internet would become such a huge thing. But if you're in the industry, if you really believe in it, and put in the time to learn, there will absolutely be opportunities for you. I think you need to hold on to that belief.

Ran Chen (Guest) (01:49:56)

Second, as I keep saying, your sources of information really matter. If you're a young person with ambition, with ideals and a desire to do something in this field, you have to find the right sources of information. Let me give you an example. I often say we're in the age of horse-drawn carriages, living inside a castle, because carriages only allow us to live within that area. Then, suddenly, cars appear. Some people have already raced ahead. They drive around every day, planting flags out in the world, putting up a flag and saying, "This is my land." They're making lots of money.

Ran Chen (Guest) (01:50:31)

So the question is, where are those people? Can we find them and hear what they have to say? The answer is: it's hard. Because those people are busy. They're busy doing those things. They don't want to talk about it. They're not interested in that. So if, if Uncle Moon gets the chance to find people like that and help tell their stories, that's an excellent source of firsthand information. Right?

Ran Chen (Guest) (01:50:51)

So the people driving cars are out exploring. Then there are people who talk a lot online. They've also driven a little way outside. I call what they do "demoing in public." They haven't necessarily built a viable business, but they can use AI to make a tool, a slide deck, or something else. And they have more time to share what they've made. Of course, that's helpful. But there's still a big gap to a working business model. They'll get plenty of attention online, though. All right.

Ran Chen (Guest) (01:51:21)

So, on the internet, who actually has the loudest voice? The people building the cars.

Uncle Moon (Host) (01:51:27) †

The carmakers.

Ran Chen (Guest) (01:51:28)

The model builders. They're the ones in this industry making the most money, for example. Those companies, the people who work there, the people building infrastructure, the serving layer, and all sorts of other things—they're the carmakers. They make more money, and attract more attention. But for an ordinary person, knowing how to build a car has nothing to do with your life. Whether you understand it or not, the car will get built. Whether you know how it's made doesn't affect your ability to drive it.

Ran Chen (Guest) (01:51:57)

But, for various reasons, the vast majority of the reporting, the interviews, and everything else revolve around how to build cars. Today, we're discussing M3. Tomorrow, an Anthropic model. Then it's OpenAI, and models with trillions of parameters. Does that matter to ordinary people? No, it doesn't.

Ran Chen (Guest) (01:52:19)

So what I'm saying is, you need to find the people already driving those cars around. What are they doing? And once you see what they're doing, can you dig into exactly how they're doing it? Digging into their work will, I think, help you the most. After that, the people demoing in public are also very helpful. They may offer a more accessible, clearer way to help you understand what these models can do. But spend more of your energy finding people who are really putting AI to work. What are they doing? What path have they taken? I think that's what young people most need to pay attention to.

Ran Chen (Guest) (01:53:05)

Where do you find those people? First, you can look at startups: the earliest, newest AI-native startups. What are they doing? How are they doing it? Second, look for things like podcasts, where people share firsthand experiences. I think those can be very helpful. It has to be firsthand. By the time it's retold, the meaning often gets distorted. It's no longer quite the same. So if you ask me, for young people, information sources matter most. If you get the right information, I believe you're all smart enough to find something that suits you. I don't think that will be a problem.

Uncle Moon (Host) (01:53:39)

Taking that a step further, what about people in their 30s, 31, 32, or 33, who've spent five or six years building a career? With AI bringing such radical change, it can actually be harder for them to pivot. What would you advise them?

Ran Chen (Guest) (01:53:51)

I agree. A lot of people have talked to me about this. On the one hand, we still have real advantages. People like us have all built up some industry know-how, learned how things work, and what the real pain points are. But on the other hand, our sense of identity, our past achievements, and our family and financial situations all place major constraints on us.

Ran Chen (Guest) (01:54:15)

I think this is a good opportunity to stop and take stock again, whether it's your finances or your family. What actually matters? What doesn't matter so much? Then rethink what value you can actually provide. If everything disappeared today, and I had to start again from scratch, making money by offering something truly valuable, what should I do? I think this is a great opportunity to pause and really think that through.

Ran Chen (Guest) (01:54:43)

I know many people come to talk to me because, in this AI era, they're miserable at work. Everything feels chaotic, with all the layoffs and so on. But whatever happens, you still have a job right now. You still have an income coming in. That's a good thing. So don't spend all your time feeling miserable and dwelling on things, getting stuck in that. Spend more time exploring what really matters: how you can actually create value. Make something yourself. Build something. Create a small product, using AI tools. Start with that $200 subscription, give yourself the full token allowance, and try to use it all up. See what you can create. I think that's what you should be doing now.

Uncle Moon (Host) (01:55:28)

What about someone who's 40, living in the U.S. with a mortgage, a family mortgage to pay, and kids who are in school, and they and their family depend on salaried jobs?

Ran Chen (Guest) (01:55:39)

I think the same advice applies. People in their 30s and 40s face pretty similar problems. It's the same approach: review your family's finances, and take a hard look at what value you can provide. Then, building on that value, see whether you can use AI tools to create something that actually makes money. I think right now is the best time to do that.

Uncle Moon (Host) (01:56:02)

Okay, great. That's all for today's conversation. Thank you, Ran, for being such an unconventional Tsinghua graduate, and sharing your remarkable life.

Ran Chen (Guest) (01:56:10)

Thank you, Uncle Moon. It's been a pleasure sharing with everyone today. I'm off to pick up the kids.

Uncle Moon (Host) (01:56:14)

All right.