WEBVTT

00:00:00.170 --> 00:00:04.309
On December 29, 2025, Meta announced that it

00:00:04.309 --> 00:00:06.870
is acquiring a Chinese -founded AI startup called

00:00:06.870 --> 00:00:11.269
Manus for over $2 billion in the hopes to supercharge

00:00:11.269 --> 00:00:14.669
their own AI innovation programs. At announcement,

00:00:14.949 --> 00:00:18.010
Manus had already reached millions of users and

00:00:18.010 --> 00:00:21.870
hit a revenue run rate of more than $125 million.

00:00:22.789 --> 00:00:26.350
Yet the product is only nine months old, making

00:00:26.350 --> 00:00:29.089
it one of the fastest growing products in history.

00:00:30.890 --> 00:00:33.710
When Manus initially launched in the spring of

00:00:33.710 --> 00:00:37.689
2025, it went viral almost instantly, with many

00:00:37.689 --> 00:00:41.409
calling it China's next deep -seek moment. Because

00:00:41.409 --> 00:00:45.189
Manus can execute complex tasks like website

00:00:45.189 --> 00:00:47.570
building, data analysis, and market research,

00:00:47.850 --> 00:00:50.929
it is basically handing technical superpowers

00:00:50.929 --> 00:00:54.490
to the average user and allowing freelancers

00:00:54.490 --> 00:00:56.689
and small businesses to punch above their weight.

00:00:57.009 --> 00:00:59.750
While some are calling this Meta's next Instagram

00:00:59.750 --> 00:01:02.740
moment, the deal has also sparked a geopolitical

00:01:02.740 --> 00:01:05.760
storm. In a bid to bypass U .S. restrictions,

00:01:06.280 --> 00:01:09.599
the company relocated its entire team from China

00:01:09.599 --> 00:01:12.620
to Singapore. As of today, the deal is still

00:01:12.620 --> 00:01:15.159
under active investigation by the Chinese government,

00:01:15.459 --> 00:01:18.439
whose final verdict could rewrite the rules for

00:01:18.439 --> 00:01:22.560
all cross -Pacific AI deals going forward. Which

00:01:22.560 --> 00:01:25.359
is why, for the premiere of Ascent Season 2,

00:01:25.599 --> 00:01:27.900
we want to give you an in -depth look at this

00:01:27.900 --> 00:01:31.280
company. We'll be talking about why we think

00:01:31.280 --> 00:01:33.340
Banner's made it among thousands of other AI

00:01:33.340 --> 00:01:36.219
agents that surfaced last year, how this deal

00:01:36.219 --> 00:01:39.000
affects incumbents like OpenAI and Google, and

00:01:39.000 --> 00:01:41.400
as always, we'll be going back into the histories

00:01:41.400 --> 00:01:44.120
of the founders, who they are and what led them

00:01:44.120 --> 00:01:46.439
to one of the most talked -about AI deals of

00:01:46.439 --> 00:01:49.359
the year. And I'm your host, Linda. And I'm your

00:01:49.359 --> 00:01:52.260
host, Pan. Happy New Year, and welcome to Season

00:01:52.260 --> 00:02:34.009
2 of Ascent. By popular demand, we bring you...

00:02:34.009 --> 00:02:36.169
There's nothing to hide here because usually

00:02:36.169 --> 00:02:39.930
it's in the topic. We bring you Manus AI. This

00:02:39.930 --> 00:02:43.289
company itself is incredibly young, actually.

00:02:43.330 --> 00:02:45.629
Technically, it's only been founded for three

00:02:45.629 --> 00:02:49.150
years. It took about nine months from the first

00:02:49.150 --> 00:02:54.289
product launch to 100 million ARR. And it is

00:02:54.289 --> 00:02:58.620
being acquired by Meta recently for... Over 2

00:02:58.620 --> 00:03:01.199
billion US dollar. Not entirely sure about the

00:03:01.199 --> 00:03:05.879
exact amount. Somewhere between 2 to 2 .5. And

00:03:05.879 --> 00:03:09.439
when I think about it, I always look at this

00:03:09.439 --> 00:03:11.740
company and I feel like, huh, it feels impossible

00:03:11.740 --> 00:03:16.740
for this deal to happen. Because have you ever

00:03:16.740 --> 00:03:19.020
heard of any other US company trying to acquire

00:03:19.020 --> 00:03:23.840
a Chinese company before like this? Not in the

00:03:23.840 --> 00:03:28.000
last... five six years not this whole geopolitics

00:03:28.000 --> 00:03:32.620
is you know there's been cases where u .s investment

00:03:32.620 --> 00:03:36.919
has went to chinese companies but legally there's

00:03:36.919 --> 00:03:40.240
been executive order that prevents this from

00:03:40.240 --> 00:03:43.979
happening right so nope not that i can think

00:03:43.979 --> 00:03:49.060
of also Not to this degree of $2 billion plus

00:03:49.060 --> 00:03:54.060
deal size. Right. And I looked up some deals

00:03:54.060 --> 00:03:58.219
that's similar, like in structure, like a U .S.

00:03:58.219 --> 00:04:00.159
company acquiring a Chinese one. I just couldn't

00:04:00.159 --> 00:04:05.400
find any predecessor at all. The only other examples

00:04:05.400 --> 00:04:09.810
that I could find. That is similar is the reverse,

00:04:09.969 --> 00:04:13.169
like some Chinese companies buying like U .S.

00:04:13.169 --> 00:04:15.210
companies in total. There was Lenovo buying Motorola

00:04:15.210 --> 00:04:17.889
and Geely or Geely buying Volvo. But that was

00:04:17.889 --> 00:04:23.649
a long, long time ago. And Anta, of course. And

00:04:23.649 --> 00:04:25.910
also, how many years do you think it takes for

00:04:25.910 --> 00:04:28.910
you to become a Magnificent Sevens VP and report

00:04:28.910 --> 00:04:31.550
directly to Mark Zuckerberg when you start your

00:04:31.550 --> 00:04:33.490
own startup? How many years do you think it should

00:04:33.490 --> 00:04:36.470
take? Never, never, ever. And then the answer

00:04:36.470 --> 00:04:40.149
is three years. So with all that, I think it's

00:04:40.149 --> 00:04:46.189
time for us to dive into Manus AI. The AI agent

00:04:46.189 --> 00:04:49.319
that kind of shocked the world. I'm going to

00:04:49.319 --> 00:04:51.279
spend maybe one minute just to give a background

00:04:51.279 --> 00:04:55.060
on LLMs and this whole AI. If you've been hiding

00:04:55.060 --> 00:05:00.980
under a rock, listen to this minute. So when

00:05:00.980 --> 00:05:05.680
most people think of AI, right? Nowadays, we're

00:05:05.680 --> 00:05:09.819
mostly talking about ChatGPT or Cloud or Gemini,

00:05:09.980 --> 00:05:15.399
maybe DeepSeek for a point in time. So these

00:05:15.399 --> 00:05:19.279
are large language models. And they're like your

00:05:19.279 --> 00:05:24.699
giant brain in a jar. So they're immensely intelligent

00:05:24.699 --> 00:05:29.860
because they have processed essentially all the

00:05:29.860 --> 00:05:33.920
available data available from the Internet or

00:05:33.920 --> 00:05:38.139
from books or written materials. So I would say

00:05:38.139 --> 00:05:40.259
I use these models day to day. And that's where

00:05:40.259 --> 00:05:43.300
most people talk about. when they talk about

00:05:43.300 --> 00:05:45.980
like AI models, right? Or when they say they

00:05:45.980 --> 00:05:49.920
use AI. But there is, after two years of using

00:05:49.920 --> 00:05:53.579
them, people are hitting this fundamental limitation,

00:05:53.699 --> 00:05:58.040
which is while they are super smart, for the

00:05:58.040 --> 00:06:00.759
most part, they're restricted to answering questions.

00:06:01.899 --> 00:06:04.819
They cannot perform the actual task for you.

00:06:04.920 --> 00:06:10.259
So for instance, you can ask ChatGPT to build

00:06:10.259 --> 00:06:13.360
you a website. What it will do is it provides

00:06:13.360 --> 00:06:16.579
this conversational guidance on how to code,

00:06:16.699 --> 00:06:19.620
or it can even give you the code, but it cannot

00:06:19.620 --> 00:06:24.939
execute the task itself. I think it... I don't

00:06:24.939 --> 00:06:28.199
know if it like Gemini executes a bit, but it's

00:06:28.199 --> 00:06:30.560
not really like self -sufficient. The user, for

00:06:30.560 --> 00:06:33.699
the most part, would take that code, copy it

00:06:33.699 --> 00:06:39.730
into their own IDE, run it. highly likely you'll

00:06:39.730 --> 00:06:43.230
run into an error and then you'll have to copy

00:06:43.230 --> 00:06:46.790
the error, go back to chat GPT, troubleshoot,

00:06:46.970 --> 00:06:50.389
seek further assistance, go back, and then the

00:06:50.389 --> 00:06:52.730
LLM starts hallucinating. That's where, you know,

00:06:52.750 --> 00:06:55.949
after two years of use, a lot of users have been

00:06:55.949 --> 00:07:00.269
hitting a wall with these LLMs. By nature, it's

00:07:00.269 --> 00:07:02.490
actually kind of reactive, right? So you ask

00:07:02.490 --> 00:07:06.629
a question and it'll try to give you an answer

00:07:06.629 --> 00:07:10.160
passively. after you ask the question, and it'll

00:07:10.160 --> 00:07:13.639
try to formulate its answer in one goal. And

00:07:13.639 --> 00:07:15.319
then you ask the next question, it'll try to

00:07:15.319 --> 00:07:19.860
formulate the answer in one goal. So that's the,

00:07:20.019 --> 00:07:24.860
I guess, the OG AI, right? So big shift. The

00:07:24.860 --> 00:07:28.319
OG AI, wow. The OG AI from 2023. This day and

00:07:28.319 --> 00:07:31.939
age, this OG AI, all right, let's do it. And

00:07:31.939 --> 00:07:37.329
so the big shift from 2025. We're in January

00:07:37.329 --> 00:07:41.810
2026. So the big shift last year is the proliferation

00:07:41.810 --> 00:07:45.430
of something called agents or the agentic applications,

00:07:45.490 --> 00:07:48.589
which are built on top of these foundational

00:07:48.589 --> 00:07:54.449
LLM models. So startups don't actually have hundreds

00:07:54.449 --> 00:07:57.810
of millions of dollars to build these foundational

00:07:57.810 --> 00:08:00.050
models. So they're building kind of the execution

00:08:00.050 --> 00:08:02.790
engine on top of that. And they're trying to

00:08:02.790 --> 00:08:05.740
make it useful. for the rest of us that are hitting

00:08:05.740 --> 00:08:10.100
that block right so these new applications help

00:08:10.100 --> 00:08:13.339
improve that by create are trying to create a

00:08:13.339 --> 00:08:16.519
digital workforce that actually makes the big

00:08:16.519 --> 00:08:19.379
brains useful for the rest of us or as useful

00:08:19.379 --> 00:08:24.040
as it can so um but i want to make this emphasis

00:08:24.040 --> 00:08:26.660
that the foundation models are not meant to be

00:08:26.660 --> 00:08:28.120
replaced. They're kind of used in conjunction

00:08:28.120 --> 00:08:30.699
with each other. You still need that brain to

00:08:30.699 --> 00:08:33.940
perform the work. But agent are the ones that

00:08:33.940 --> 00:08:36.779
are, you know, for example, booking you a flight

00:08:36.779 --> 00:08:41.100
or building you a PowerPoint or launching you

00:08:41.100 --> 00:08:44.240
a web page, right? So they're kind of the application

00:08:44.240 --> 00:08:47.980
layers that are built on top of the LLM agents

00:08:47.980 --> 00:08:52.220
that perform all these things. And so this is

00:08:52.220 --> 00:08:54.259
a backdrop. And this is where I shift to what

00:08:54.259 --> 00:08:57.019
Manus does exactly. So the website is called

00:08:57.019 --> 00:09:00.500
Manus, M -A -N -U -S dot I -M. And this is their

00:09:00.500 --> 00:09:03.100
interface. If you go to that website, you'll

00:09:03.100 --> 00:09:05.500
see a simple interface that looks exactly like

00:09:05.500 --> 00:09:08.620
ChatGPT. It says, what can I do for you? And

00:09:08.620 --> 00:09:11.159
there's a box. And then you enter a prompt just

00:09:11.159 --> 00:09:14.360
like you would with ChatGPT. I mean, you've tried

00:09:14.360 --> 00:09:17.279
it. What's different is it doesn't give you a

00:09:17.279 --> 00:09:20.379
one -off reply. The first thing it'll do for

00:09:20.379 --> 00:09:23.539
most of my queries that actually as a result

00:09:23.539 --> 00:09:29.019
is it will start up something, right? And then

00:09:29.019 --> 00:09:34.379
there will be a box on the bottom that resembles

00:09:34.379 --> 00:09:37.179
a screenshot. And then you click on the box and

00:09:37.179 --> 00:09:40.480
it expands. And when it expands, you'll see something

00:09:40.480 --> 00:09:45.279
called Manus' computer. So what it did in the

00:09:45.279 --> 00:09:48.879
first step is it actually initialized a virtual

00:09:48.879 --> 00:09:51.990
computer in the background. That virtual computer

00:09:51.990 --> 00:09:56.009
then is just dedicated to your task or your prompt

00:09:56.009 --> 00:10:01.090
into solving what you asked it to do. So the

00:10:01.090 --> 00:10:04.269
differentiation is from most apps, it's not just

00:10:04.269 --> 00:10:08.110
a browser window. It's actually a full -on computer.

00:10:08.970 --> 00:10:11.769
It's a Linux workspace installed with its own

00:10:11.769 --> 00:10:14.190
browser, its own text editor, its own terminal,

00:10:14.370 --> 00:10:18.549
its own image editing tools. You will get pre

00:10:18.549 --> 00:10:21.370
-installed Python libraries. Everything that

00:10:21.370 --> 00:10:24.690
you or I would need to operate a computer to

00:10:24.690 --> 00:10:27.870
perform our day -to -day task is there. And you

00:10:27.870 --> 00:10:30.149
get that screenshot, right? It's actually like

00:10:30.149 --> 00:10:34.929
a, it's almost like looking at Manus, the agent,

00:10:35.009 --> 00:10:38.940
work over its shoulder. This is what's called

00:10:38.940 --> 00:10:41.799
the computer -used or the GUI navigation, the

00:10:41.799 --> 00:10:44.379
graphical user interface navigation. And that's

00:10:44.379 --> 00:10:48.059
something that emerged late 2024, early 2025.

00:10:48.960 --> 00:10:52.159
And these models are specifically trained to

00:10:52.159 --> 00:10:55.539
understand how an agent, how a virtual agent

00:10:55.539 --> 00:10:58.879
can use the graphical user interface and to click

00:10:58.879 --> 00:11:01.440
and drag just like we would when we use computers.

00:11:01.799 --> 00:11:04.299
I think what's special about Manos is that when

00:11:04.299 --> 00:11:08.360
you ask it to perform a task, It's very different

00:11:08.360 --> 00:11:13.240
than your chat GPT, right? It's no longer a chatbot.

00:11:13.279 --> 00:11:17.820
It's actually a virtual person almost trying

00:11:17.820 --> 00:11:21.200
to understand your task and try to perform it

00:11:21.200 --> 00:11:22.919
and giving you results and give you a report

00:11:22.919 --> 00:11:25.120
at the same time. And you have the ability to

00:11:25.120 --> 00:11:28.000
overlook it. And I think that's where the biggest

00:11:28.000 --> 00:11:31.340
value add of this group of agents. This is probably

00:11:31.340 --> 00:11:33.840
one of the reasons why it caught the attentions

00:11:33.840 --> 00:11:37.399
of so many people upon release. When we look

00:11:37.399 --> 00:11:39.820
at the story, even though it is a very young

00:11:39.820 --> 00:11:43.399
company, but there were plenty of content to

00:11:43.399 --> 00:11:48.700
be researched upon. Let's talk about those three

00:11:48.700 --> 00:11:54.039
key figures one by one before they started. Maybe

00:11:54.039 --> 00:12:00.700
we can briefly touch upon one critical investment

00:12:00.700 --> 00:12:05.090
fund, an investor of VC. And then when everything

00:12:05.090 --> 00:12:09.090
is glued together, something else happened before

00:12:09.090 --> 00:12:11.330
Manas and then Manas happened, right? Yes, yes.

00:12:11.750 --> 00:12:18.370
Our person number one is Xiao Hong, right? The

00:12:18.370 --> 00:12:23.070
CEO. The CEO. The main guy. Red, right. And how

00:12:23.070 --> 00:12:28.809
would you describe Red? Adjectives. Smart. Knows

00:12:28.809 --> 00:12:32.950
when to quit. This is my, you know, like I think...

00:12:33.179 --> 00:12:37.679
A very... Interesting. Very big pro for founders,

00:12:37.740 --> 00:12:44.860
actually. Yeah. A non -traditional AI agent founder,

00:12:45.059 --> 00:12:48.620
I would say. Because he doesn't have a computer

00:12:48.620 --> 00:12:53.559
science degree from Stanford or... Software engineering.

00:12:53.860 --> 00:12:55.580
He does have a software engineering degree, but

00:12:55.580 --> 00:12:59.299
his school is not exactly Tsinghua, Beidao, or

00:12:59.299 --> 00:13:02.860
Stanford. I would describe him as he's very practical

00:13:02.860 --> 00:13:05.879
because of the product manager work that he does.

00:13:06.480 --> 00:13:10.259
He seems to be a very empathetic person and always

00:13:10.259 --> 00:13:13.740
being rooted for what the customer might want.

00:13:13.940 --> 00:13:17.799
He graduated from Huazhong University of Science

00:13:17.799 --> 00:13:22.620
Technology in 2015, and he was a software engineering

00:13:22.620 --> 00:13:25.649
major. I think very much like JK Liu, he was

00:13:25.649 --> 00:13:27.289
also developing a bunch of different software

00:13:27.289 --> 00:13:30.169
as well. He was in college. He was also on the

00:13:30.169 --> 00:13:32.549
college innovation team. It was called Qiming.

00:13:32.850 --> 00:13:36.370
When did WeChat came out in China? Do you know?

00:13:36.590 --> 00:13:39.690
No idea. 2011, the first year that he went to

00:13:39.690 --> 00:13:44.110
the college. Yeah. He designed a WeChat wall,

00:13:44.509 --> 00:13:47.269
WeChat drift bottle, a lot of small features,

00:13:47.450 --> 00:13:50.470
a lot of tools for WeChat because of WeChat just

00:13:50.470 --> 00:13:55.159
came out. He is also a very solid talent for

00:13:55.159 --> 00:13:57.860
PMF, product market fit. And why do I say that?

00:13:58.440 --> 00:14:03.500
He made a tool within WeChat to benefit the group

00:14:03.500 --> 00:14:06.720
of users as college students, like just his classmates.

00:14:06.899 --> 00:14:10.659
He wanted to create a weather lookup function.

00:14:11.519 --> 00:14:13.679
But obviously, whether looking up wasn't a must

00:14:13.679 --> 00:14:15.860
-have function in the college at all, but looking

00:14:15.860 --> 00:14:19.820
for grades is. So he then tweaked that function

00:14:19.820 --> 00:14:22.279
into a grade -looking function in WeChat, and

00:14:22.279 --> 00:14:25.559
that got viral. Right. Right. And then he became

00:14:25.559 --> 00:14:29.629
the admin for the... the college school account

00:14:29.629 --> 00:14:34.750
later he did a few tech internships during his

00:14:34.750 --> 00:14:36.529
interview i think it was quite funny that he

00:14:36.529 --> 00:14:40.210
mentioned a pro tip for if you want to be starting

00:14:40.210 --> 00:14:43.769
your startup in the college you want to be buying

00:14:43.769 --> 00:14:47.490
your engineering friends dinner or lunch. Because

00:14:47.490 --> 00:14:49.350
in the college, you can just buy them very cheap

00:14:49.350 --> 00:14:51.169
food and then they will just gang up with you

00:14:51.169 --> 00:14:54.350
and then do your features. But if you are graduated,

00:14:54.649 --> 00:14:56.610
then you will be buying them much more expensive

00:14:56.610 --> 00:14:59.210
meals. Like machine learning. That's the pro

00:14:59.210 --> 00:15:02.129
tip. I think it's a pretty good tip. It is acquire

00:15:02.129 --> 00:15:05.110
them while they're young. Pete did the same thing.

00:15:05.289 --> 00:15:07.610
Like when he was not even in college, when he

00:15:07.610 --> 00:15:09.649
was in high school, he was doing that. We'll

00:15:09.649 --> 00:15:12.419
come back, come to that later. And then 2015,

00:15:12.799 --> 00:15:16.700
just a few months before he graduated, he started

00:15:16.700 --> 00:15:20.840
this company called Nightingale Technology. I

00:15:20.840 --> 00:15:24.259
don't know about whether you buy into the story

00:15:24.259 --> 00:15:26.940
of the Nightingale. I don't. You should tell

00:15:26.940 --> 00:15:29.340
the story. Why is name Nightingale? Question.

00:15:29.759 --> 00:15:33.940
So it was claimed. OK, so Nightingale Technology,

00:15:34.399 --> 00:15:39.340
the name Nightingale is from Oscar Wilde's book.

00:15:41.000 --> 00:15:44.679
The Nightingale and the Rose. And if you don't

00:15:44.679 --> 00:15:47.419
know what that is, very quickly, Tale of the

00:15:47.419 --> 00:15:50.620
Arts, a short fairy tale about an anygale who

00:15:50.620 --> 00:15:53.820
sacrifices its life, her life, to create a red

00:15:53.820 --> 00:15:55.840
rose for a young student who left, like just

00:15:55.840 --> 00:15:58.440
a very romantic idea. It's like a sad romance.

00:15:58.679 --> 00:16:00.879
Sad romance. I'm not going to spoil the whole

00:16:00.879 --> 00:16:02.860
story. It's actually not a romance. It's a criticism

00:16:02.860 --> 00:16:08.220
on humans. Right, yeah. So basically, it's a...

00:16:08.429 --> 00:16:13.190
The bird sings all night with her heart pierced

00:16:13.190 --> 00:16:17.389
by a thorn and with the blood that is coming

00:16:17.389 --> 00:16:20.929
out of the heart, staining the rose. And then

00:16:20.929 --> 00:16:23.990
the girl finally gets the red rose, but rejects

00:16:23.990 --> 00:16:26.429
the love, blah, blah. But it's a it's a very

00:16:26.429 --> 00:16:29.809
tragic story. And he interpreted and Red interpreted

00:16:29.809 --> 00:16:34.669
this as. building useful tools yeah it's great

00:16:34.669 --> 00:16:38.330
for humanity like he's he thinks the nightingale

00:16:38.330 --> 00:16:43.549
is a supporter in the story yeah so we're and

00:16:43.549 --> 00:16:46.889
his company is a supporter of wechat ecosystem

00:16:46.889 --> 00:16:49.250
yeah i don't know how i don't know how that ties

00:16:49.250 --> 00:16:52.629
together but that was the reason for naming the

00:16:52.629 --> 00:16:54.889
company nightingale right we're gonna build you

00:16:54.889 --> 00:16:58.090
useful tools And then quickly moving on. So remember,

00:16:58.250 --> 00:17:00.610
he founded the company in January 2015, just

00:17:00.610 --> 00:17:02.409
a few months before he graduated, which is in

00:17:02.409 --> 00:17:07.309
June 2015. He did build a few things when he

00:17:07.309 --> 00:17:10.309
started that company, and mostly WeChat related,

00:17:10.569 --> 00:17:14.509
mostly in the WeChat ecosystem. And those tools

00:17:14.509 --> 00:17:19.170
are still in use today by WeChat. And one is

00:17:19.170 --> 00:17:22.809
called Yiban Zhuzhou. Yiban Assistant, which

00:17:22.809 --> 00:17:27.140
is basically an editing, operating, tool for

00:17:27.140 --> 00:17:29.559
WeChat official accounts. A lot of friends still

00:17:29.559 --> 00:17:32.460
use it today to manage posts. And the other one

00:17:32.460 --> 00:17:36.640
is Weiban Assistant, which is a CRM tool. And

00:17:36.640 --> 00:17:39.880
that business was eventually acquired by another

00:17:39.880 --> 00:17:41.779
company, which is a public company right now,

00:17:41.839 --> 00:17:46.220
Minglu Technology. So these are on the successful

00:17:46.220 --> 00:17:50.380
ends for Xiaohong's projects when he was in school

00:17:50.380 --> 00:17:55.000
and ended up getting millions of users. Did you

00:17:55.000 --> 00:17:58.950
know? Xiao Hong used to have a WeChat blog post

00:17:58.950 --> 00:18:05.430
called Tool Geek. But it's no longer updating.

00:18:05.589 --> 00:18:07.990
But it just shows how much he loves building

00:18:07.990 --> 00:18:12.950
tools. And here is a visual for something that

00:18:12.950 --> 00:18:16.529
he had posted about 10 years ago. Oh, no, actually

00:18:16.529 --> 00:18:17.950
not 10 years ago, just a couple years ago. But

00:18:17.950 --> 00:18:19.690
it was a very old post and it's written by Steve

00:18:19.690 --> 00:18:24.130
Jobs. And here it's Steve Jobs' autobiography.

00:18:24.720 --> 00:18:26.779
But one of the ways that I believe people express

00:18:26.779 --> 00:18:28.640
their appreciation to the rest of the humanity

00:18:28.640 --> 00:18:31.160
is to make something wonderful and put it out

00:18:31.160 --> 00:18:33.400
there. So I think that's what he means. He's

00:18:33.400 --> 00:18:36.160
really into this tool building. He's really into

00:18:36.160 --> 00:18:38.819
tool building. Yeah. Yeah, that I agree. And

00:18:38.819 --> 00:18:41.740
he's very good at it because he started Iban

00:18:41.740 --> 00:18:45.930
understanding that there's a lot of... There's

00:18:45.930 --> 00:18:49.170
a lot of posts that are being posted on official

00:18:49.170 --> 00:18:53.109
WeChat accounts, but then the editors don't have

00:18:53.109 --> 00:18:57.210
a lot of tools. And so he created a tool for

00:18:57.210 --> 00:18:59.930
editors so that it saves their time, like a one

00:18:59.930 --> 00:19:04.329
-click editing for the WeChat interspace. He

00:19:04.329 --> 00:19:07.869
knows what they want. He's born a product manager.

00:19:08.130 --> 00:19:12.259
He is. He knows. where the pain point is at.

00:19:12.440 --> 00:19:16.579
Yeah. You know how he started with Iban for the

00:19:16.579 --> 00:19:19.500
first three years and then only started Weiban

00:19:19.500 --> 00:19:25.599
around 2019. And it wasn't, like when he first

00:19:25.599 --> 00:19:29.079
started Weiban, and Weiban is different from

00:19:29.079 --> 00:19:33.880
you. So Iban is meant for creators to edit their

00:19:33.880 --> 00:19:40.460
posts. Weiban is a... CRM software for businesses,

00:19:40.599 --> 00:19:44.039
small businesses to manage their customers. So

00:19:44.039 --> 00:19:46.160
they're very different. Even though they're both

00:19:46.160 --> 00:19:48.140
in the WeChat ecosystem, they're very different.

00:19:48.720 --> 00:19:53.000
And the first six months of Weiban was not a

00:19:53.000 --> 00:19:55.200
failure, but a waiting period. It wasn't clear.

00:19:55.440 --> 00:19:59.279
It wasn't clear because according to him, in

00:19:59.279 --> 00:20:04.640
2019, he and a bunch of WeChat platform suppliers

00:20:05.869 --> 00:20:09.690
went to a WeChat hosted event where they hinted

00:20:09.690 --> 00:20:13.470
that we're going to figure out how to improve

00:20:13.470 --> 00:20:18.009
our whole WeChat ecosystem with combining the

00:20:18.009 --> 00:20:21.349
business accounts and the personal accounts.

00:20:21.650 --> 00:20:24.589
Because a lot of small businesses before then

00:20:24.589 --> 00:20:27.049
were using personal accounts to manage their

00:20:27.049 --> 00:20:29.049
business accounts because you have a lot of plugin

00:20:29.049 --> 00:20:31.950
that actually made it easier for you to manage

00:20:31.950 --> 00:20:35.390
your customer in the personal accounts. Because

00:20:35.390 --> 00:20:38.470
of the proliferation of this plugin and the fact

00:20:38.470 --> 00:20:41.670
that you can auto message and spam everyone,

00:20:42.369 --> 00:20:47.809
WeChat, he made a bet that WeChat in the future

00:20:47.809 --> 00:20:50.990
would do something about it to make sure that

00:20:50.990 --> 00:20:55.109
their ecosystem is sustainable. But it's like

00:20:55.109 --> 00:20:58.029
the big short. where you make the bet before

00:20:58.029 --> 00:21:01.349
it happens and you wait but the period where

00:21:01.349 --> 00:21:04.930
you wait everyone's telling you that why are

00:21:04.930 --> 00:21:08.990
you doing this what are you doing yeah interesting

00:21:08.990 --> 00:21:12.789
i mean comes with the risk yeah but he was so

00:21:12.789 --> 00:21:16.210
certain of the ecosystem play he just had to

00:21:16.210 --> 00:21:18.680
wait I thought that was pretty interesting story

00:21:18.680 --> 00:21:21.000
because it wasn't a short thing. It wasn't like

00:21:21.000 --> 00:21:23.539
AI, you know, like you just you go in. Right.

00:21:23.720 --> 00:21:27.140
Yeah. I personally been through that period where

00:21:27.140 --> 00:21:29.339
there were a lot of business just sending private

00:21:29.339 --> 00:21:32.420
messages through WeChat and it was very irregulated.

00:21:32.599 --> 00:21:36.619
Right. Yeah. So he made the right bet based on

00:21:36.619 --> 00:21:40.000
and obviously he had a lot of relationship within

00:21:40.000 --> 00:21:44.829
WeChat. the corporate because of his success

00:21:44.829 --> 00:21:48.049
of iban so probably helped with the decision

00:21:48.049 --> 00:21:52.190
but still that's like a lot of enduring that

00:21:52.190 --> 00:21:54.869
he has to he has to tell his employees that don't

00:21:54.869 --> 00:21:58.069
worry i know there's no users right now it's

00:21:58.069 --> 00:21:59.809
gonna happen it's gonna happen and it happened

00:21:59.809 --> 00:22:03.230
yeah and he sold it right after yeah a lot of

00:22:03.230 --> 00:22:08.490
enduring that comes in to red's life that enduring

00:22:08.490 --> 00:22:11.920
moment came One step earlier, actually in 2016,

00:22:12.099 --> 00:22:14.380
when he wasn't even sure whether he wants to

00:22:14.380 --> 00:22:17.460
continue on with that company. Apparently, the

00:22:17.460 --> 00:22:20.380
company wasn't banking on a lot of money and

00:22:20.380 --> 00:22:23.400
also doesn't have much to do at the time besides

00:22:23.400 --> 00:22:27.940
just Yiban. So he felt a little helpless and

00:22:27.940 --> 00:22:31.319
bored. A lot of his classmates went to work for

00:22:31.319 --> 00:22:34.740
the big tech firms in Beijing and anywhere else

00:22:34.740 --> 00:22:38.519
in China. And so he was still staying on in his

00:22:38.519 --> 00:22:41.480
startup. But then one day, this is also one of

00:22:41.480 --> 00:22:44.039
the had he not done this moment. So before he

00:22:44.039 --> 00:22:46.539
officially onboarded with a big tech company

00:22:46.539 --> 00:22:48.960
because he thought that, oh, maybe I should do

00:22:48.960 --> 00:22:52.980
this too. Him and his teammates in 2016 in October

00:22:52.980 --> 00:22:56.700
participated in a hackathon that was hosted by

00:22:56.700 --> 00:23:02.279
Ping West in Beijing. In the hackathon, one of

00:23:02.279 --> 00:23:05.180
the judges was an investor from GenFund and his

00:23:05.180 --> 00:23:08.400
name is Liu Yuan. Red at the time already made

00:23:08.400 --> 00:23:10.319
the decision that, oh, I'm going to, you know,

00:23:10.319 --> 00:23:13.019
go on to take an offer from a big tech company.

00:23:13.119 --> 00:23:15.559
So he wasn't extremely stressed out about the

00:23:15.559 --> 00:23:19.220
result of this hackathon at all. But it was indeed

00:23:19.220 --> 00:23:23.059
very life changing for him. And they were demonstrating

00:23:23.059 --> 00:23:26.640
some of the extensions for Chrome for that hackathon.

00:23:29.630 --> 00:23:31.789
Liu Yuan's attention. And then that was the investor

00:23:31.789 --> 00:23:35.650
from Zhen Fan. So the next day, right after that

00:23:35.650 --> 00:23:38.410
hackathon, Liu Yuan brought him and his team

00:23:38.410 --> 00:23:42.630
to Zhen Fan for an IC to see the investment committee.

00:23:43.670 --> 00:23:46.990
And in that, their name was being added to the

00:23:46.990 --> 00:23:50.609
IC last, handwritten on it. He was not on the

00:23:50.609 --> 00:23:52.910
official list of all the projects that's being

00:23:52.910 --> 00:23:56.430
screened by the IC at the day. Just want to show

00:23:56.430 --> 00:23:59.460
you a bit of a... Red's personality. So by the

00:23:59.460 --> 00:24:01.380
end of the IC, everyone was kind of impressed

00:24:01.380 --> 00:24:03.720
by the work that he's done. And they were saying

00:24:03.720 --> 00:24:05.220
that this is such an interesting project. And

00:24:05.220 --> 00:24:07.839
an investor, unnamed investor from GenFund asked

00:24:07.839 --> 00:24:10.940
him, if I were to give you this much money, how

00:24:10.940 --> 00:24:12.880
are you going to use this money? And Red immediately

00:24:12.880 --> 00:24:15.759
replied, that's not necessary. We can make a

00:24:15.759 --> 00:24:17.740
profit. We don't need that money. So he rejected

00:24:17.740 --> 00:24:21.539
the money in the first place. And on the same

00:24:21.539 --> 00:24:26.430
day, he received a competing offer. That's coming

00:24:26.430 --> 00:24:29.230
from a very big tech company and made an even

00:24:29.230 --> 00:24:32.849
higher offer from the money that Zhen Fan was

00:24:32.849 --> 00:24:37.569
going to offer him. It was allegedly 3 to 4 million

00:24:37.569 --> 00:24:41.130
RMB. Zhen Fan was offering him 1 million RMB.

00:24:41.190 --> 00:24:46.390
Red rejected the 3 to 4 million RMB offer from

00:24:46.390 --> 00:24:49.750
the big tech company of purchasing his whole

00:24:49.750 --> 00:24:53.529
company on Tell You Why. And he eventually decided

00:24:53.529 --> 00:24:56.349
after thorough thinking, he realized that, oh,

00:24:56.349 --> 00:24:59.630
maybe I want to be working on my own projects.

00:24:59.690 --> 00:25:01.950
And if I don't do it now, I don't think he'll

00:25:01.950 --> 00:25:04.869
ever use it. So he ended up taking GenFund's

00:25:04.869 --> 00:25:09.309
1 million RMB. So the tech company investor wanted

00:25:09.309 --> 00:25:11.809
him to do not his own project? There are two

00:25:11.809 --> 00:25:14.410
offers. One is from GenFund and then one is from

00:25:14.410 --> 00:25:16.769
a different tech company. But for the same product?

00:25:18.240 --> 00:25:21.059
For the same product, for the same product, for

00:25:21.059 --> 00:25:22.880
the same chrome extension that he was building

00:25:22.880 --> 00:25:25.160
at the time. And so apparently he's doing a very

00:25:25.160 --> 00:25:27.980
good job with the team. And maybe that tech company

00:25:27.980 --> 00:25:31.160
sees great value of that product thriving at

00:25:31.160 --> 00:25:32.920
the broader ecosystem that they were building.

00:25:33.079 --> 00:25:35.460
But it was not given any name, but it was being

00:25:35.460 --> 00:25:37.759
offered at a higher price. Got it. So he neither

00:25:37.759 --> 00:25:40.980
wants the money and he wants the autonomy of

00:25:40.980 --> 00:25:43.940
development. Clearly, he's prioritizing the autonomy

00:25:43.940 --> 00:25:47.720
and just chasing dreams. So lessons learned from

00:25:47.720 --> 00:25:50.880
this whole period of building the first company

00:25:50.880 --> 00:25:54.869
and Red Set later. If you want to showcase your

00:25:54.869 --> 00:25:56.809
work, you should always find a place to showcase

00:25:56.809 --> 00:25:59.930
it. And if you are more relaxed, you will even

00:25:59.930 --> 00:26:01.950
get better results. That's what happened to him.

00:26:02.009 --> 00:26:04.450
I think it's so true. Yeah, yeah. If you care

00:26:04.450 --> 00:26:06.930
less in general, I think that comes out better,

00:26:07.009 --> 00:26:10.529
right? To be a chill version of yourself. And

00:26:10.529 --> 00:26:13.069
the third one is don't give yourself a fixed

00:26:13.069 --> 00:26:16.599
conclusion too early. Because he... was kind

00:26:16.599 --> 00:26:19.000
of desperate and he wasn't happy with the project.

00:26:19.099 --> 00:26:21.319
And all of a sudden two offers come. I know,

00:26:21.359 --> 00:26:24.000
right? Like things could really turn in a day.

00:26:24.519 --> 00:26:28.599
I think that actually goes into how much we don't

00:26:28.599 --> 00:26:32.380
talk enough about his trials and errors, right?

00:26:32.480 --> 00:26:35.890
His first venture. after graduation, he tried

00:26:35.890 --> 00:26:37.630
a couple of things, failed, and then found this

00:26:37.630 --> 00:26:40.529
one that works. And then with Weiban, he also

00:26:40.529 --> 00:26:42.569
tried a couple of things and found Weiban and

00:26:42.569 --> 00:26:46.410
found that the business side of Weichat worked.

00:26:47.130 --> 00:26:50.849
And then later on, he'll try other things, right?

00:26:50.950 --> 00:26:54.450
And then found that Manus works. So it's like,

00:26:54.509 --> 00:26:57.309
you don't really give up on your first try. And

00:26:57.309 --> 00:27:00.990
then he also got lucky by the time he is at his

00:27:00.990 --> 00:27:05.900
nth try. is about to give up, an investor like

00:27:05.900 --> 00:27:08.960
Jim Fund comes in. Shows up. It's what we call

00:27:08.960 --> 00:27:13.269
fate. Oh, actually, I think he has a third option,

00:27:13.369 --> 00:27:15.390
which is to go work for a tech company. But that

00:27:15.390 --> 00:27:17.349
third option doesn't have any autonomy. So I

00:27:17.349 --> 00:27:20.009
guess he doesn't. Yeah. Or like a million dollar

00:27:20.009 --> 00:27:22.569
funding. For sure. For sure. Million dollar in

00:27:22.569 --> 00:27:25.170
the year 2016. That's not possible for someone

00:27:25.170 --> 00:27:27.390
who is just going to work for a company after

00:27:27.390 --> 00:27:31.190
graduation. Exactly. I would take the investor

00:27:31.190 --> 00:27:34.390
funding. Yeah. For sure. Yeah. I want to save

00:27:34.390 --> 00:27:38.349
the butterfly effect company for later. OK. Can

00:27:38.349 --> 00:27:39.890
we do peak now? Let's do peak. Let's do peak.

00:27:40.279 --> 00:27:43.720
How would you describe Peak? Genius. I put down

00:27:43.720 --> 00:27:50.059
the same word. Anything else? INTJ genius. So

00:27:50.059 --> 00:27:53.339
there's one impression where he's like, I've

00:27:53.339 --> 00:27:56.539
tried to be a CEO. I realize I'm not good at

00:27:56.539 --> 00:27:59.579
it. So therefore, I'd rather be the CEO and I

00:27:59.579 --> 00:28:02.799
just do all the tech stuff. I know. So besides

00:28:02.799 --> 00:28:08.259
genius, I also put down idealistic. I think they're

00:28:08.259 --> 00:28:10.440
all pretty idealistic. That whole team, including

00:28:10.440 --> 00:28:13.460
their investors. Which is so insane. How would

00:28:13.460 --> 00:28:16.160
you, at this day and age, you have a whole team

00:28:16.160 --> 00:28:19.240
of very idealistic, from investor to product

00:28:19.240 --> 00:28:21.140
manager to engineer. Everyone is kind of on the

00:28:21.140 --> 00:28:23.859
same page. Realistic idealistic, though. Not

00:28:23.859 --> 00:28:25.779
just like, oh, we're going to, yeah, practical

00:28:25.779 --> 00:28:28.380
idealistic. I'll start with a how do you not

00:28:28.380 --> 00:28:33.000
moment. So Peak Ji, his Chinese name is Ji Chao,

00:28:33.079 --> 00:28:38.140
right? Goes by Peak. So how do you not? He's

00:28:38.140 --> 00:28:40.839
mom and dad. Oh, this is not very average Chinese

00:28:40.839 --> 00:28:42.839
family. So his dad is a college professor in

00:28:42.839 --> 00:28:46.119
physics. And his mom is a serial entrepreneur

00:28:46.119 --> 00:28:49.460
in Beijing. And before we dive in, I think we

00:28:49.460 --> 00:28:52.079
should just take a moment to acknowledge that

00:28:52.079 --> 00:28:54.359
he is very lucky to be born in a family like

00:28:54.359 --> 00:28:58.720
this. Yeah. I think what resonated with me was

00:28:58.720 --> 00:29:02.079
there is no such term as serial entrepreneur

00:29:02.079 --> 00:29:05.809
in Beijing. At that time. At that time. Either

00:29:05.809 --> 00:29:08.910
it could be like somewhere between late 80s or

00:29:08.910 --> 00:29:13.630
90s. Yeah. There was no venture capitalism in

00:29:13.630 --> 00:29:17.750
China. Right. Peak lived exactly as a combination

00:29:17.750 --> 00:29:21.970
of his parents, I think. Right. His first development

00:29:21.970 --> 00:29:26.490
experience, which is insane, in elementary school.

00:29:27.349 --> 00:29:31.789
So he. He left China at the age of four and went

00:29:31.789 --> 00:29:33.630
to the States with his parents and came back

00:29:33.630 --> 00:29:37.190
in the second grade. And because he has superb

00:29:37.190 --> 00:29:39.369
English, because you're in the States for a few

00:29:39.369 --> 00:29:43.769
years, he got... acquired by the robotics team

00:29:43.769 --> 00:29:47.009
in the elementary school in Peking University

00:29:47.009 --> 00:29:50.990
-affiliated elementary school. And he constantly

00:29:50.990 --> 00:29:55.549
joins competitions on the international championship.

00:29:55.849 --> 00:29:58.269
One of those super smart kids, right? Like those

00:29:58.269 --> 00:30:02.150
kids on the robotics team. At the age of 16,

00:30:02.470 --> 00:30:06.089
okay, another had he not moment. And this gives

00:30:06.089 --> 00:30:08.210
me a little goosebumps because we might share

00:30:08.210 --> 00:30:11.619
the same school principle. based on how Peak

00:30:11.619 --> 00:30:14.799
described his high school. My high school principal,

00:30:15.140 --> 00:30:21.480
headmaster, left my high school two years after

00:30:21.480 --> 00:30:24.579
I graduated and went to Beidafuzhong to be headmaster.

00:30:24.700 --> 00:30:28.619
So there is this possibility. that the age also

00:30:28.619 --> 00:30:30.880
kind of clicked. Look how you guys turning out

00:30:30.880 --> 00:30:33.460
so well. But hold on a second, hold on a second.

00:30:34.160 --> 00:30:37.200
But I think this really shows like the school's

00:30:37.200 --> 00:30:40.240
whole management system feels like, oh, I think

00:30:40.240 --> 00:30:43.920
he was the headmaster of the time. Going back

00:30:43.920 --> 00:30:47.440
to Peak. So in high school, that's when he started

00:30:47.440 --> 00:30:51.539
working as a full stack developer for Mammoth

00:30:51.539 --> 00:30:56.259
browser. Still downloadable today. mammoth browser

00:30:56.259 --> 00:31:01.940
oh really yeah the school is so supportive of

00:31:01.940 --> 00:31:08.299
what he did at the time the vice headmaster verbally

00:31:08.299 --> 00:31:11.259
said to him it doesn't mean that you're an excellent

00:31:11.259 --> 00:31:14.859
student just because you have scored 600 in your

00:31:14.859 --> 00:31:17.940
college entrance examination this is like 99

00:31:17.940 --> 00:31:21.160
percent 99 .99 percentile probably the other

00:31:21.160 --> 00:31:23.259
teachers working at the school will tell them

00:31:23.259 --> 00:31:26.519
that Before you graduate and move on with your

00:31:26.519 --> 00:31:29.579
lives, you need to be an independent person with

00:31:29.579 --> 00:31:32.779
critical thinking before you can do more things

00:31:32.779 --> 00:31:34.759
with your life and do academics or do anything.

00:31:34.960 --> 00:31:37.960
It's so rare. And I want to put this into context

00:31:37.960 --> 00:31:43.859
in like 90s, early 2000s China. This is a very

00:31:43.859 --> 00:31:46.140
different landscape from liberal arts America.

00:31:47.859 --> 00:31:52.380
Kids don't do this. Like Gaokao is... the university

00:31:52.380 --> 00:31:56.180
entrance exam is their life for 18 years you

00:31:56.180 --> 00:31:58.960
don't deviate from that most family will not

00:31:58.960 --> 00:32:01.299
let you most parents will not let you deviate

00:32:01.299 --> 00:32:06.940
from that principals and teachers are half your

00:32:06.940 --> 00:32:09.160
parents and they won't let you they will push

00:32:09.160 --> 00:32:12.640
you from that right yeah so it's a very different

00:32:12.640 --> 00:32:16.700
landscape from the you know your western uh academics

00:32:16.700 --> 00:32:21.599
k -12 is very different so in china this is if

00:32:21.599 --> 00:32:27.079
you have the brains you need to get on that either

00:32:27.079 --> 00:32:29.759
you bypass the exam because you're so smart or

00:32:29.759 --> 00:32:33.960
you get on that exam and you you get you get

00:32:33.960 --> 00:32:37.240
the best score that you can Yeah, absolutely.

00:32:37.500 --> 00:32:39.619
And as Eminem would say, you have one shot in

00:32:39.619 --> 00:32:43.099
your life. That's exactly what he meant. That's

00:32:43.099 --> 00:32:45.880
exactly because, you know, college is an examination.

00:32:46.279 --> 00:32:50.059
You get one shot. You get one shot. So this is

00:32:50.059 --> 00:32:52.240
a very unconventional school that he went to,

00:32:52.279 --> 00:32:54.619
a very unconventional family and system that

00:32:54.619 --> 00:32:59.500
he was in. So by senior year in high school,

00:32:59.839 --> 00:33:04.099
he was being poached by tech companies. And guess

00:33:04.099 --> 00:33:10.009
who was poaching him? xiaomi no yes xiaomi had

00:33:10.009 --> 00:33:14.210
just started at the time okay and there was another

00:33:14.210 --> 00:33:17.710
company that was you know trying to grab the

00:33:17.710 --> 00:33:20.569
talent apple also likes him because of the browser

00:33:20.569 --> 00:33:23.950
business that he started oh right so he got attention

00:33:23.950 --> 00:33:27.670
from apple right and just to to give you a little

00:33:27.670 --> 00:33:30.049
bit of a what size of a business was he starting

00:33:30.049 --> 00:33:34.700
the business model was very simple It was Mammoth's

00:33:34.700 --> 00:33:38.359
browser. You pay by each download. So for every

00:33:38.359 --> 00:33:42.359
copy that's sold on the App Store, you make one

00:33:42.359 --> 00:33:46.220
copy of money. Which is the normal business model

00:33:46.220 --> 00:33:48.240
when Apple App Store first started. That's right.

00:33:48.299 --> 00:33:50.420
They only did freemium like... Per purchase.

00:33:50.480 --> 00:33:52.680
Afterwards. Yeah, exactly. And guess how much

00:33:52.680 --> 00:33:56.779
money he made just by Mammoth number one. Not

00:33:56.779 --> 00:34:00.779
sure. He made 300K US dollar. That's not bad.

00:34:01.690 --> 00:34:05.069
two, three million RMB as a high school student.

00:34:05.410 --> 00:34:09.210
Why do you need college interest exams? And how

00:34:09.210 --> 00:34:12.630
did he even spread the news and did the GTM of

00:34:12.630 --> 00:34:15.070
MAMA's browser is by posting on discussion boards.

00:34:15.230 --> 00:34:20.170
So he hoarded a bunch of Western users and he

00:34:20.170 --> 00:34:22.969
knows that these would be the people who will

00:34:22.969 --> 00:34:26.940
be paying for my browser. And he also, like at

00:34:26.940 --> 00:34:31.519
the age of 17 or 18, he is well aware that Chinese

00:34:31.519 --> 00:34:33.960
people don't like paying for a software. So he

00:34:33.960 --> 00:34:36.239
wasn't even trying for the Chinese market. Not

00:34:36.239 --> 00:34:38.460
that he was going global right away, but he was

00:34:38.460 --> 00:34:40.559
just very acutely aware that Chinese people are

00:34:40.559 --> 00:34:43.000
not going to pay for that. Who's the paying users?

00:34:43.579 --> 00:34:45.760
Right. But he posted on the Chinese forum anyways,

00:34:45.840 --> 00:34:49.570
just to get the popularity up. And that's also

00:34:49.570 --> 00:34:52.409
good for the app store at the time. He did the

00:34:52.409 --> 00:34:56.309
UI UX development testing daily operations completely

00:34:56.309 --> 00:34:58.489
on his own. Of course, who else are you going

00:34:58.489 --> 00:35:01.710
to hire as a high school student? Right. So because

00:35:01.710 --> 00:35:05.250
his browser was all over the place and someone

00:35:05.250 --> 00:35:08.010
didn't notice him and it was someone from Sequoia

00:35:08.010 --> 00:35:10.469
Capital. And it was the time when Sequoia Capital,

00:35:10.570 --> 00:35:12.469
Sequoia China was still a part of Sequoia Capital.

00:35:12.590 --> 00:35:14.909
And by way of introduction, he met with a partner

00:35:14.909 --> 00:35:18.539
with Sequoia Capital who later. introduced him

00:35:18.539 --> 00:35:23.880
to the dream investment of his life the fund

00:35:23.880 --> 00:35:26.500
that we'll get on later which is Zhen Fund and

00:35:26.500 --> 00:35:30.039
he get it introduced to Bob Xu also Sequoia didn't

00:35:30.039 --> 00:35:32.800
invest in him but they introduced him to Zhen

00:35:32.800 --> 00:35:35.619
Fund I think they both did okay but it was kind

00:35:35.619 --> 00:35:37.860
of a small circle of people because at the time

00:35:37.860 --> 00:35:42.059
everyone knows each other and this is at the

00:35:42.059 --> 00:35:47.949
time when you know it's the early 2000 and 13,

00:35:47.989 --> 00:35:51.690
15 -ish. Yeah, venture capital just kind of getting

00:35:51.690 --> 00:35:54.409
its way. Right. And also, if you just speak English

00:35:54.409 --> 00:35:56.949
well enough in Beijing, I think that will get

00:35:56.949 --> 00:35:58.510
you a lot of attention. Not to say that you've

00:35:58.510 --> 00:36:02.250
built something really cool, right? And one quick

00:36:02.250 --> 00:36:04.369
note I want to add about Zhen Fan and Xi Xiaoping

00:36:04.369 --> 00:36:10.130
is Bob Xu, before he started Zhen Fan, he was

00:36:10.130 --> 00:36:13.590
also a partner of the largest English tutoring...

00:36:14.590 --> 00:36:18.010
agency literally every single chinese kid would

00:36:18.010 --> 00:36:21.210
know the name new oriental yeah new oriental

00:36:21.210 --> 00:36:26.079
yeah You've never. You didn't need to. I've never.

00:36:26.340 --> 00:36:29.219
But actually, interestingly, a lot of my Canadian

00:36:29.219 --> 00:36:33.460
high school classmates who are Chinese would

00:36:33.460 --> 00:36:37.739
go back for summers. And teach? No, no, no. Students?

00:36:38.900 --> 00:36:41.699
Pre -college. A lot of us wanted to go to the

00:36:41.699 --> 00:36:44.179
U .S. for colleges. The ones that are targeting

00:36:44.179 --> 00:36:46.960
U .S. colleges. They wouldn't go through the

00:36:46.960 --> 00:36:50.280
Canadian tutorial system. They would go back

00:36:50.280 --> 00:36:53.550
to China for this. summer go sign up for new

00:36:53.550 --> 00:36:56.309
oriental and then they come back and their sat

00:36:56.309 --> 00:37:00.329
scores jump and you slay yeah and i didn't get

00:37:00.329 --> 00:37:02.590
on that i didn't get i didn't know this information

00:37:02.590 --> 00:37:06.409
no i didn't know i did i did i just didn't know

00:37:06.409 --> 00:37:10.250
i had information asymmetry back then but anyways

00:37:10.250 --> 00:37:13.789
anyways that's that's my first uh acquaintance

00:37:13.789 --> 00:37:17.570
with new oriental right yeah so so that gives

00:37:17.570 --> 00:37:24.530
bob a very good pool of talent. people in China

00:37:24.530 --> 00:37:28.030
talented kids and like everyone who is going

00:37:28.030 --> 00:37:30.030
abroad and studying right so funny you increase

00:37:30.030 --> 00:37:32.829
their SAT scores they get into better colleges

00:37:32.829 --> 00:37:35.190
and they come back and work exactly that's the

00:37:35.190 --> 00:37:38.050
point that's the whole point yeah I also studied

00:37:38.050 --> 00:37:41.849
very briefly at with New Oriental it was okay

00:37:41.849 --> 00:37:43.889
if you go to some of the smaller classes but

00:37:43.889 --> 00:37:45.809
the one that I've been to I was studying TOEFL

00:37:45.809 --> 00:37:48.349
at the time and I was the only high schooler

00:37:48.349 --> 00:37:51.849
in a in an ocean of college students and anyways

00:37:52.590 --> 00:37:55.170
Very long ago. Back to Peak. Had he not moment,

00:37:55.289 --> 00:37:58.650
he also joined a hackathon event that was hosted

00:37:58.650 --> 00:38:01.670
by GenFund. Not the same one as what? Not the

00:38:01.670 --> 00:38:04.030
same one, but he got asked the same question.

00:38:04.190 --> 00:38:06.010
He got asked, literally asked. Can I give you

00:38:06.010 --> 00:38:09.190
money? Literally the same question. And Bob asked

00:38:09.190 --> 00:38:12.449
him, if I give you money, if you do a startup

00:38:12.449 --> 00:38:16.909
and not go to the college, would you do it? And

00:38:16.909 --> 00:38:22.179
Peak said, do you want to guess? Yes. no let

00:38:22.179 --> 00:38:24.659
me think about it peak said no immediately he

00:38:24.659 --> 00:38:27.860
rejected yeah he went back home and discussed

00:38:27.860 --> 00:38:30.800
his parents his parents said quit school his

00:38:30.800 --> 00:38:32.920
parents said no no his parents didn't say quit

00:38:32.920 --> 00:38:35.260
school but his parents and this is also very

00:38:35.260 --> 00:38:38.840
rare answer guys let's just say kudos to the

00:38:38.840 --> 00:38:41.889
parents the parents said I think you should evaluate

00:38:41.889 --> 00:38:44.070
this opportunity and see what's the opportunity

00:38:44.070 --> 00:38:47.590
cost there. And I'm like, oh, my God. See what's

00:38:47.590 --> 00:38:50.610
the opportunity cost there. I know. Like to a

00:38:50.610 --> 00:38:54.250
17 -year -old who just made $300 ,000 out of

00:38:54.250 --> 00:38:56.769
building a browser. So his parents' suggestion

00:38:56.769 --> 00:38:59.349
is you should reevaluate and see what's the opportunity

00:38:59.349 --> 00:39:01.949
cost. And after very careful thinking, he decided

00:39:01.949 --> 00:39:03.769
to go on with the startup and not go to college

00:39:03.769 --> 00:39:07.610
and start what we later know as Peak Labs. So

00:39:07.610 --> 00:39:09.530
here's a funny thing. He did something that read.

00:39:10.110 --> 00:39:13.190
also did very similar fashion he recruited a

00:39:13.190 --> 00:39:16.130
few students who had already graduated from the

00:39:16.130 --> 00:39:19.969
same high school who got accepted by peak university

00:39:19.969 --> 00:39:23.849
and the three of them worked together as a team

00:39:23.849 --> 00:39:26.130
but did the other two quit the university no

00:39:26.130 --> 00:39:28.190
the other two are in the college already so they

00:39:28.190 --> 00:39:29.909
didn't have to worry about it and there was that

00:39:29.909 --> 00:39:34.150
pku okay and they're older and they all were

00:39:34.150 --> 00:39:38.570
a part of the computer club which is very similar

00:39:38.570 --> 00:39:40.909
to the robotics club idea. So he's always been

00:39:40.909 --> 00:39:43.349
with the whiz kid. And you just grab a couple

00:39:43.349 --> 00:39:46.590
others who don't have to worry about it. The

00:39:46.590 --> 00:39:51.309
600 -plus gal gal group. Yeah, the 600 -grade

00:39:51.309 --> 00:39:53.949
gal gal group. That's right. How did Mammoth

00:39:53.949 --> 00:39:57.949
end? Peak got into natural language processing,

00:39:58.170 --> 00:40:03.789
NLP, by the end of 2011. That's before Jeff Hinton.

00:40:03.929 --> 00:40:06.369
No, it's very early, but he was only reading

00:40:06.369 --> 00:40:09.389
articles about it. He found something interesting

00:40:09.389 --> 00:40:11.469
and then he stopped working on browser. He just

00:40:11.469 --> 00:40:13.670
went to chase the next thing that he got interested

00:40:13.670 --> 00:40:17.150
into. So he was just reading a whole lot about

00:40:17.150 --> 00:40:20.789
NLPU. And this perfectly channels onto the Peak

00:40:20.789 --> 00:40:25.110
Lab development. In Peak Lab, so both Sequoia

00:40:25.110 --> 00:40:29.250
Capital and GenFund together invested about $5

00:40:29.250 --> 00:40:32.429
million in him. That's a lot of money. Yeah.

00:40:33.030 --> 00:40:37.030
And in six months, he developed a couple projects.

00:40:37.690 --> 00:40:41.170
And obviously, like Mammoth still continued to

00:40:41.170 --> 00:40:44.329
evolve. There's Mammoth 1 and all the way it

00:40:44.329 --> 00:40:47.489
went into Mammoth 5. All the talent that he had

00:40:47.489 --> 00:40:51.070
used for Manus, he had already used it for Mammoth

00:40:51.070 --> 00:40:56.179
5. So from design. to the commercial, to the

00:40:56.179 --> 00:40:59.239
voiceover is all done by Peak G. And up until

00:40:59.239 --> 00:41:01.719
Madness, he was also kind of the poster boy for

00:41:01.719 --> 00:41:07.539
Madness. So he did that early on. In 2013, around

00:41:07.539 --> 00:41:12.300
that period was when, okay, this is also one

00:41:12.300 --> 00:41:16.039
of the, had he not or had this not happened,

00:41:16.119 --> 00:41:19.199
maybe he wouldn't ended up in the sector. There

00:41:19.199 --> 00:41:20.920
were a lot of papers being published by NLP.

00:41:21.449 --> 00:41:25.969
right yeah because they yeah there was a after

00:41:25.969 --> 00:41:30.789
the lee feifei alex net alex net that's right

00:41:30.789 --> 00:41:34.610
it blew up everyone's into it right but so he

00:41:34.610 --> 00:41:36.730
was in this whole foundational model thing way

00:41:36.730 --> 00:41:40.860
before Way before. Wow. As a high school slash,

00:41:41.019 --> 00:41:44.679
you know, never went to college kid. And in 2014

00:41:44.679 --> 00:41:47.440
to 15, there was some other event that happened

00:41:47.440 --> 00:41:49.960
that really inspired him into thinking, what

00:41:49.960 --> 00:41:51.900
will the next generation of search be? Do you

00:41:51.900 --> 00:41:54.579
know what happened in 2014 and 15? Hint, hint.

00:41:55.539 --> 00:42:02.440
Hardware. Hardware? 2014? iPhone? iPhone actually.

00:42:03.050 --> 00:42:06.869
No, Apple Watch came out in 2014. But I was quite

00:42:06.869 --> 00:42:09.289
shocked when I realized, oh, Apple Watch came

00:42:09.289 --> 00:42:12.329
out that early ago. It was 2014. 2014 was the

00:42:12.329 --> 00:42:14.489
announcement was made and 2015 was when the prototype

00:42:14.489 --> 00:42:18.309
came out. Got it. All right. So IoT? Wearables?

00:42:18.630 --> 00:42:22.909
Wearables. So the Apple updates and all the NLP

00:42:22.909 --> 00:42:25.949
papers prompted into thinking, will there be

00:42:25.949 --> 00:42:29.050
new ways of like search that will interact with

00:42:29.050 --> 00:42:31.989
people? And he started this way early on, right?

00:42:32.110 --> 00:42:35.389
And then this is still when he was, you know,

00:42:35.389 --> 00:42:38.650
in peak labs. And all of these thinking ended

00:42:38.650 --> 00:42:42.869
up with a second project in peak lab, which is,

00:42:42.929 --> 00:42:45.949
I don't know how to call it. Is it Maggie or

00:42:45.949 --> 00:42:50.250
Maggie? Maggie. Maggie. Okay, cool. It's an AI

00:42:50.250 --> 00:42:54.610
powered search engine. So I remember this one.

00:42:54.849 --> 00:42:57.440
It's not actually. a search engine where you

00:42:57.440 --> 00:42:59.519
could go in and type right it's actually it's

00:42:59.519 --> 00:43:02.139
also in the foundational layer where you need

00:43:02.139 --> 00:43:05.719
to build application on top of it i found the

00:43:05.719 --> 00:43:08.860
demo oh there's actually a demo okay let me show

00:43:08.860 --> 00:43:13.809
you and i will probably cut this bit out Okay,

00:43:13.849 --> 00:43:16.829
end of demo. And I just, I have my laptop here

00:43:16.829 --> 00:43:19.170
just because I want to show you that. I will,

00:43:19.269 --> 00:43:21.510
I don't know what to do with it. I just really

00:43:21.510 --> 00:43:23.409
want to see your reaction to it. Okay, I'll tell

00:43:23.409 --> 00:43:25.849
you exactly why they didn't continue it. Oh man,

00:43:25.849 --> 00:43:31.030
I'm like a fan. Okay, bunch of nerds. Very hardcore

00:43:31.030 --> 00:43:36.730
nerds. They're like, I love, like, not 42. The

00:43:36.730 --> 00:43:40.429
answer to the work should be Magi, not 42. So

00:43:40.429 --> 00:43:42.889
we've both seen it, right? It's an AI -powered

00:43:42.889 --> 00:43:46.610
search. It has knowledge graphs, which has a

00:43:46.610 --> 00:43:48.449
lot of values later on. Actually, a lot of tech

00:43:48.449 --> 00:43:51.969
companies are still using it today. And it has

00:43:51.969 --> 00:43:55.050
structured answers, and you can add your ad queries

00:43:55.050 --> 00:43:56.929
on top of it. I think it's a creation that's

00:43:56.929 --> 00:44:00.750
very much ahead of its time. I wonder if Google

00:44:00.750 --> 00:44:03.590
Knowledge Graph is developing the same thing

00:44:03.590 --> 00:44:06.710
at the same time. A lot of people were developing

00:44:06.710 --> 00:44:08.889
the same thing at the same time. But with like

00:44:08.889 --> 00:44:11.969
a team of a thousand engineers. Right, right.

00:44:12.150 --> 00:44:15.210
You know? Because guess what happened after they

00:44:15.210 --> 00:44:18.809
had this? What happened? ChatGPD 3 for early

00:44:18.809 --> 00:44:21.369
access. Wait, this is so late in the... Yeah,

00:44:21.489 --> 00:44:24.050
this is already pretty late. Okay. Yeah. And

00:44:24.050 --> 00:44:28.309
so ChatGPD came out and Peak felt like, oh, this

00:44:28.309 --> 00:44:30.090
is the end of the world. Well, sorry, this is

00:44:30.090 --> 00:44:31.969
not what he was thinking, but this is just me

00:44:31.969 --> 00:44:34.449
trying to impersonate what he was thinking. But...

00:44:34.650 --> 00:44:37.190
So Peak Lab then took a long time between...

00:44:37.190 --> 00:44:40.309
Lasted a while. Mammoth kind of maintenance mode

00:44:40.309 --> 00:44:44.130
to... I think they were continuously trying to

00:44:44.130 --> 00:44:47.309
try new things, but some of the major events

00:44:47.309 --> 00:44:49.849
that happened in Peak Lab would be first continuing

00:44:49.849 --> 00:44:54.590
mammoth, and then this Magi. Got it. Magi. Magi.

00:44:54.610 --> 00:44:57.610
Right? And Peak actually also mentioned the project

00:44:57.610 --> 00:45:02.030
Magi a couple times in his various interviews

00:45:02.030 --> 00:45:06.460
on... browser and search he had he had a dream

00:45:06.460 --> 00:45:09.260
and then the dream is to build something that

00:45:09.260 --> 00:45:11.579
can take over google that can be the next generation

00:45:11.579 --> 00:45:15.260
of search he has has had that dream before and

00:45:15.260 --> 00:45:18.000
i think magic was trying to yes tackle that and

00:45:18.000 --> 00:45:19.659
he was also into natural language processing

00:45:19.659 --> 00:45:21.400
i think it's just unfortunate that google is

00:45:21.400 --> 00:45:24.559
developing something exactly the same at the

00:45:24.559 --> 00:45:28.079
same time that he's working on it so i mean he

00:45:28.079 --> 00:45:30.840
could have just went to work lead one of the

00:45:30.840 --> 00:45:33.300
teams i know i know transfer the knowledge but

00:45:33.300 --> 00:45:38.079
uh yeah wow so you know this chachi beauty 3

00:45:38.079 --> 00:45:40.559
came out and he felt like that all of his work

00:45:40.559 --> 00:45:43.099
is for nothing it's all moving too fast so his

00:45:43.099 --> 00:45:47.460
first instinct was to sell magi so his reflection

00:45:47.460 --> 00:45:51.159
at the time is um too ahead of his time But he

00:45:51.159 --> 00:45:55.159
tried everything. He coded word by word. There

00:45:55.159 --> 00:45:58.880
was no regrets. After Magic, he went to work

00:45:58.880 --> 00:46:03.420
for a B2B AI tech company. And he was working

00:46:03.420 --> 00:46:05.679
on, well, as a researcher slash engineer, working

00:46:05.679 --> 00:46:07.880
on improving the scores for AI. Like no longer

00:46:07.880 --> 00:46:11.480
founding a company, but employed by a company.

00:46:11.780 --> 00:46:14.139
Probably did that for fun. And he, in fact, did

00:46:14.139 --> 00:46:16.340
that for fun. What was he doing at the time?

00:46:16.860 --> 00:46:21.159
So he's... Primary goal is to make sure the AI

00:46:21.159 --> 00:46:25.300
is getting as high score as possible on all the

00:46:25.300 --> 00:46:29.400
international scoring systems. And those who

00:46:29.400 --> 00:46:32.579
made the AI to the leaderboard will get more

00:46:32.579 --> 00:46:34.840
graphics card. And guess who ranked number one

00:46:34.840 --> 00:46:37.199
on that leaderboard? Peak. Peak was on number

00:46:37.199 --> 00:46:40.900
one at all times. And this is a positively reinforcing

00:46:40.900 --> 00:46:45.099
mechanism. Like the better scoring you get, the

00:46:45.099 --> 00:46:47.619
more graphics card you get. And then the better

00:46:47.619 --> 00:46:49.519
score. The better results you get, the more graphics

00:46:49.519 --> 00:46:51.760
cards you get. So at the end of the day, he ended

00:46:51.760 --> 00:46:53.960
up hoarding a bunch of graphics cards he wasn't

00:46:53.960 --> 00:46:59.119
even using. That's not very nice. But very capable,

00:46:59.360 --> 00:47:02.340
you know, doing it for fun. And if you're interested

00:47:02.340 --> 00:47:05.380
in more of Peak's work. First of all, he was

00:47:05.380 --> 00:47:08.519
listed on MIT Technology Review in the Innovator

00:47:08.519 --> 00:47:13.119
Recognition in 2025. Really, Innovators 135 did

00:47:13.119 --> 00:47:16.420
a bunch of cool things. And he is also on GitHub

00:47:16.420 --> 00:47:19.519
under the handle PeakG. So that's where you can

00:47:19.519 --> 00:47:24.679
get his most updated work. So this is him before

00:47:24.679 --> 00:47:29.440
he met Red. Or let's say Butterfly Effect and

00:47:29.440 --> 00:47:33.010
how that materialized. Young genius. Yeah. Very

00:47:33.010 --> 00:47:35.630
cool. Yeah. All over the place with a lot of

00:47:35.630 --> 00:47:39.230
ideals. But a constant supply of people who are

00:47:39.230 --> 00:47:42.449
willing to invest in him and his hobby projects.

00:47:42.670 --> 00:47:45.809
I know. Yeah. It's great. If you're that talented

00:47:45.809 --> 00:47:48.190
and put yourself out there, people will notice

00:47:48.190 --> 00:47:51.550
you and they will want to lift you up. Yeah.

00:47:51.550 --> 00:47:55.710
Yeah. He was very lucky. Okay. On to number three.

00:47:56.449 --> 00:48:01.190
Zhang Tao. Tao. Hide Cloud. His name is Hyde

00:48:01.190 --> 00:48:06.489
Cloud. I love that name. How would you describe

00:48:06.489 --> 00:48:10.269
him? So he's the one that I read up the least

00:48:10.269 --> 00:48:13.989
about, but I think he's like a product manager,

00:48:14.190 --> 00:48:18.329
kind of your more typical path, right? By going

00:48:18.329 --> 00:48:22.389
to the tech firms in China and paying his dues

00:48:22.389 --> 00:48:24.750
and understanding how products are developed.

00:48:25.269 --> 00:48:28.989
I listed almost the exact same set of words.

00:48:29.449 --> 00:48:33.030
I said solid PM. Yes. With a lot of practical

00:48:33.030 --> 00:48:36.010
experience on the ground. Yes. Super grounded

00:48:36.010 --> 00:48:40.070
person. He's a bit older though. Oh, okay. He

00:48:40.070 --> 00:48:44.349
was born in 1986. And he's from Chongqing. It's

00:48:44.349 --> 00:48:48.050
a pretty thick, cute Chongqing accent. Homie's

00:48:48.050 --> 00:48:51.070
accent. Okay. So he's kind of the older brother

00:48:51.070 --> 00:48:55.469
for the team later on in Manus. But he is actually

00:48:55.469 --> 00:48:59.739
the one with the most PM experience. And how

00:48:59.739 --> 00:49:03.159
experienced are we talking about? Did you know,

00:49:03.300 --> 00:49:06.480
fun fact, he started working in tech companies

00:49:06.480 --> 00:49:09.599
in 2014. To give you a little more reference,

00:49:09.860 --> 00:49:13.219
have you ever heard of a system that's called

00:49:13.219 --> 00:49:19.820
the Symbian OS? Saiban? Vaguely. Vaguely? Do

00:49:19.820 --> 00:49:22.340
you know which cell phone company? Wait, is it?

00:49:22.670 --> 00:49:26.710
That's not Huawei? So SymbianOS was a pioneering

00:49:26.710 --> 00:49:29.869
discontinued mobile operating system for smartphones,

00:49:30.070 --> 00:49:34.909
dominating in the 2000s, especially on Nokia

00:49:34.909 --> 00:49:39.090
devices. Oh, Nokia! Known for its early multitasking,

00:49:39.170 --> 00:49:42.130
customization, and robust features like messaging,

00:49:42.210 --> 00:49:44.869
web browsing, blah, blah, blah. So Nokia. So

00:49:44.869 --> 00:49:49.110
he goes way back. Wow. Yeah. Goes way back. And

00:49:49.110 --> 00:49:52.650
he also had the experience of working in a series

00:49:52.650 --> 00:49:54.969
of tech companies and a lot of relations to Google.

00:49:55.269 --> 00:49:57.789
First, Tencent, obviously, because that's the

00:49:57.789 --> 00:50:02.909
biggest slash the first one. And SnapP, Wandojia,

00:50:03.050 --> 00:50:07.829
does that ring a bell? So Dr. Li, Dr. Kaifu Li,

00:50:08.030 --> 00:50:11.670
one of his first ventures with Sinovision was

00:50:11.670 --> 00:50:16.030
SnapP and it was an app store. A lot of Googlers

00:50:16.030 --> 00:50:18.090
actually ended up working for Snapchat. For Android.

00:50:18.489 --> 00:50:22.670
Yeah. In China. Global. But it was physically

00:50:22.670 --> 00:50:26.610
based in China, based in Beijing. So Tao was

00:50:26.610 --> 00:50:29.809
also a product manager for ByteDance, naturally.

00:50:30.429 --> 00:50:33.590
And for the first couple of years, he was working

00:50:33.590 --> 00:50:36.110
towards more on the data end of things. There

00:50:36.110 --> 00:50:40.440
was a website called sensorsdata .cn. so that's

00:50:40.440 --> 00:50:43.820
where he started and later he went to work on

00:50:43.820 --> 00:50:46.880
lark which is the international version of the

00:50:46.880 --> 00:50:52.980
collaborating software facial yes he also worked

00:50:52.980 --> 00:50:56.559
in a company which also discontinued but was

00:50:56.559 --> 00:50:58.900
very interesting which is also related to one

00:50:58.900 --> 00:51:02.119
of the later investors to manus So Wang Huiwen,

00:51:02.199 --> 00:51:04.719
this name, you're familiar with, right? So his

00:51:04.719 --> 00:51:07.960
name showed up in the Manus list of investors.

00:51:08.300 --> 00:51:11.579
Yes. And they used to work with each other. So

00:51:11.579 --> 00:51:14.980
Tao and Huiwen used to work with each other in

00:51:14.980 --> 00:51:17.380
a company called Guanglian Zhiyuan. which is

00:51:17.380 --> 00:51:20.199
also an AGI company. And I couldn't find the

00:51:20.199 --> 00:51:23.099
proper English name for it because it has unregistered.

00:51:23.099 --> 00:51:25.619
That company no longer exists. But that company,

00:51:25.639 --> 00:51:28.159
I guess, can roughly be translated into Light

00:51:28.159 --> 00:51:30.579
Years Away. It's an AGI -focused company, I guess,

00:51:30.599 --> 00:51:33.219
developing a bunch of products that's related

00:51:33.219 --> 00:51:37.840
to AI. And Wang Huiwen is a co -founder of Meituan.

00:51:39.699 --> 00:51:43.619
So the company was eventually taken over by Meituan.

00:51:43.820 --> 00:51:46.750
That's correct. Because Wang Huiwen and Wang

00:51:46.750 --> 00:51:49.289
Xing, the Meituan CEO Wang Xing, they used to

00:51:49.289 --> 00:51:51.610
be college dorm buddies. They were in the same

00:51:51.610 --> 00:51:53.969
dorm. Okay. So I guess if you're in the dorm

00:51:53.969 --> 00:51:55.829
in Tsinghua and starting a business together,

00:51:55.949 --> 00:51:58.429
you're kind of brothers for life, right? Yeah.

00:51:58.489 --> 00:52:00.829
So that's why that company got bought out by

00:52:00.829 --> 00:52:08.920
Meituan. So HiCloud. is a serial PM. Serial and

00:52:08.920 --> 00:52:12.119
very serious PM. And a few other things that

00:52:12.119 --> 00:52:14.079
he did, I have to point out. Had he not done

00:52:14.079 --> 00:52:16.280
this, he wouldn't have been picked up by the

00:52:16.280 --> 00:52:20.880
group. In year 2024, he was still working for

00:52:20.880 --> 00:52:26.139
the AGI company light years away. In his sharing

00:52:26.139 --> 00:52:29.579
as the PM in that specific company, he expressed

00:52:29.579 --> 00:52:32.820
a couple times. which is also kind of his ethos

00:52:32.820 --> 00:52:36.260
for working as a PM, which is as a PM in the

00:52:36.260 --> 00:52:40.340
AI era, it is simply not enough to find the PMF.

00:52:40.840 --> 00:52:43.619
You need to find the best model product fit,

00:52:43.760 --> 00:52:46.840
which is exactly, I guess, what you described

00:52:46.840 --> 00:52:50.019
earlier, that man is championed, right? They're

00:52:50.019 --> 00:52:53.280
so good at it. Use the right model to create

00:52:53.280 --> 00:52:55.400
the right product, and then you find the product

00:52:55.400 --> 00:52:58.360
market fit. So that's what he said. In February

00:52:58.360 --> 00:53:02.340
2025, and specifically February 10th, 2025, remember

00:53:02.340 --> 00:53:04.739
what happened in the end of January last year?

00:53:05.739 --> 00:53:08.639
End of January last year? Around Chinese New

00:53:08.639 --> 00:53:11.559
Year time. Deep Seek. Yes, exactly. Deep Seek

00:53:11.559 --> 00:53:16.400
came out, right? And Zhang Tao did a very famous

00:53:16.400 --> 00:53:23.480
two -hour video on Deep Seek's R1 model. He did

00:53:23.480 --> 00:53:27.119
what? he did that video to educate the non -technical

00:53:27.119 --> 00:53:31.900
everyone that why is it good? How did I not come

00:53:31.900 --> 00:53:34.059
across this in the deep -seek research? It's

00:53:34.059 --> 00:53:37.179
just so well hidden. But when I click on that

00:53:37.179 --> 00:53:39.860
video, I have so many friends who like that video.

00:53:39.940 --> 00:53:42.099
And every single person who clicked liked on

00:53:42.099 --> 00:53:44.260
that video. Where is this published? I will send

00:53:44.260 --> 00:53:46.059
it to you later. It's published on his WeChat

00:53:46.059 --> 00:53:49.780
blog post. And those people who like that video

00:53:49.780 --> 00:53:52.400
are either working for an active AI research

00:53:52.400 --> 00:53:55.179
institution or a very big tech firm. So it shows

00:53:55.179 --> 00:53:58.260
the credits of how good that content is. And

00:53:58.260 --> 00:54:01.739
the name of that video is called The best way

00:54:01.739 --> 00:54:04.699
to pay tribute is through learning. I don't know

00:54:04.699 --> 00:54:06.639
whether learning here means whatever learning,

00:54:06.739 --> 00:54:11.800
but it's giving the background of what is R1.

00:54:11.980 --> 00:54:16.800
And he was explaining in very layman's term why

00:54:16.800 --> 00:54:22.820
DeepSeek was good technically and compared a

00:54:22.820 --> 00:54:25.099
bunch of models. And I think it was probably

00:54:25.099 --> 00:54:28.099
ranked by a lot of people in the tech world,

00:54:28.199 --> 00:54:31.980
tech video of the year. Wow. Yeah. Actually,

00:54:32.079 --> 00:54:35.019
this comes to all three co -founders that we

00:54:35.019 --> 00:54:38.039
covered so far. They do have this commonality,

00:54:38.039 --> 00:54:43.079
which is they're all very active, not only with

00:54:43.079 --> 00:54:45.159
the investors, but within the tech community.

00:54:45.599 --> 00:54:47.719
The influencers kind of. Yeah, they're like your

00:54:47.719 --> 00:54:50.159
influencers within the tech community. And they

00:54:50.159 --> 00:54:53.969
like posting. Their ideas and their thoughts.

00:54:54.110 --> 00:54:57.110
And I think that actually helped with popularity.

00:54:57.110 --> 00:54:59.969
Do you want to guess? So for the two -hour video

00:54:59.969 --> 00:55:04.909
that Tao created, who was the initiator of that

00:55:04.909 --> 00:55:07.949
video? Whose platform was hosting that video?

00:55:08.690 --> 00:55:13.510
Platform? Or whose channel? Who initiated that

00:55:13.510 --> 00:55:19.010
idea? Pink or red? GenFund. GenFund. Oh, GenFund

00:55:19.010 --> 00:55:23.429
has a... Oh, podcast? It was on their channel.

00:55:23.630 --> 00:55:25.769
Wow. Yeah, it was kind of on the public education

00:55:25.769 --> 00:55:28.869
side of things. Okay. Yeah, so everything all

00:55:28.869 --> 00:55:30.409
comes together, right? Everything all comes together.

00:55:31.210 --> 00:55:34.210
Doofan also likes posting stuff. Oh, very active,

00:55:34.369 --> 00:55:38.590
very active. So one last thing about Tao is he

00:55:38.590 --> 00:55:42.750
said that he reads paper every day and he would

00:55:42.750 --> 00:55:45.050
very jokingly mention that it's not that I have

00:55:45.050 --> 00:55:46.929
such a good habit of reading paper every day,

00:55:46.989 --> 00:55:50.269
but factually in this industry, One new paper

00:55:50.269 --> 00:55:53.730
is published every day. At least. At least. Yeah.

00:55:53.789 --> 00:55:55.929
So he's still reading paper every single day.

00:55:56.110 --> 00:56:00.829
Oh, I know. Like all three of them are, I'd say,

00:56:00.889 --> 00:56:06.619
rightfully on top. on the respective things that

00:56:06.619 --> 00:56:10.519
they're working on. And they're still learning

00:56:10.519 --> 00:56:12.980
and studying and staying active in the communities.

00:56:13.219 --> 00:56:16.420
How do you bring these three together? How do

00:56:16.420 --> 00:56:18.019
you find them? How do you bring them together?

00:56:18.340 --> 00:56:23.039
What an excellent question. Let's get to that

00:56:23.039 --> 00:56:26.360
bit. But before we can really connect the three

00:56:26.360 --> 00:56:30.820
of them, we should definitely just quickly talk

00:56:30.820 --> 00:56:32.820
about Zhen Fan before we go into Liu Yuan and

00:56:32.820 --> 00:56:36.199
how they all meet each other. So GenFund, we

00:56:36.199 --> 00:56:40.559
all heard of this company before. GenFund has

00:56:40.559 --> 00:56:45.579
a sort of an investment manifesto. And it goes,

00:56:45.920 --> 00:56:50.139
GenFund believes in backing visionary founders,

00:56:50.320 --> 00:56:53.619
fostering technological innovation, empowering

00:56:53.619 --> 00:56:57.539
bold entrepreneurs with... Disruptive ideas,

00:56:57.820 --> 00:57:00.239
early stage tech angel investing, emphasizing

00:57:00.239 --> 00:57:02.559
speed, resilience and supportive partnerships

00:57:02.559 --> 00:57:06.159
with over strict control, focusing on areas like

00:57:06.159 --> 00:57:10.119
AI, biotech, new consumption and high end manufacturing.

00:57:10.320 --> 00:57:12.619
Yeah. And if you haven't heard of them, they're

00:57:12.619 --> 00:57:18.800
one of the largest, well -known early stage venture

00:57:18.800 --> 00:57:21.900
capital. Early stage. Early stage. And they're

00:57:21.900 --> 00:57:25.050
very known for. not just investing on the project,

00:57:25.110 --> 00:57:27.809
but investing in people. So like founder first

00:57:27.809 --> 00:57:31.070
kind of theory. They're all very liked by the

00:57:31.070 --> 00:57:34.909
investor community. Right. Yes. So Liu Yuan was

00:57:34.909 --> 00:57:39.650
the key investor in Red's former company. Yes.

00:57:39.750 --> 00:57:42.829
19U, right? Yeah. Liu Yuan actually invested

00:57:42.829 --> 00:57:46.289
in Red five times. Yeah. So Liu Yuan was obviously

00:57:46.289 --> 00:57:49.550
very... Like four, five rounds. So Liu Yuan was

00:57:49.550 --> 00:57:52.030
very much linked with Red, and they're actually

00:57:52.030 --> 00:57:55.369
both from the same place, Hubei, Wuhan. And Pete

00:57:55.369 --> 00:57:59.449
got connected with them because of Anna. I thought

00:57:59.449 --> 00:58:02.090
it was with Bob. Bob was the one that tried to

00:58:02.090 --> 00:58:04.630
convince him, the founding partner. But linking

00:58:04.630 --> 00:58:07.409
onto a specific project with Red, that was for

00:58:07.409 --> 00:58:11.730
Anna. And Zhang Tao was brought into the team

00:58:11.730 --> 00:58:15.610
later because of Yu Sen, Dai Yu Sen. Okay, so

00:58:15.610 --> 00:58:19.719
basically... Was a lot of matchmaking here. Yeah,

00:58:19.780 --> 00:58:22.280
and they're all the names. They're just directors

00:58:22.280 --> 00:58:24.599
or founding partner, like partners or founding

00:58:24.599 --> 00:58:28.260
partners of GenFound. Yeah, right. Apparently,

00:58:28.260 --> 00:58:30.559
they collaborate a whole lot. And when they're

00:58:30.559 --> 00:58:34.760
working on a new project, exchange information

00:58:34.760 --> 00:58:37.400
and bring the relevant folks together. So that's

00:58:37.400 --> 00:58:40.320
the way of working. So a bit more about Liu Yuan.

00:58:40.860 --> 00:58:44.900
Obviously, he is the initiator. And he's the

00:58:44.900 --> 00:58:47.860
sort of the core investment partner in this project.

00:58:48.039 --> 00:58:51.840
Yeah. The person who gave Red the first pot of

00:58:51.840 --> 00:58:53.599
gold, basically, after the 26th. And continued

00:58:53.599 --> 00:58:57.940
to give him four more pots of gold. Yeah. So

00:58:57.940 --> 00:59:02.260
Yuan graduated from Washington Lee in the US,

00:59:02.500 --> 00:59:05.079
which is a liberal arts college. And his major

00:59:05.079 --> 00:59:08.159
was accounting. And as he described, it was not

00:59:08.159 --> 00:59:10.820
the sexiest. major at the time. I think it's

00:59:10.820 --> 00:59:13.699
generally known. No, but at the time, it was

00:59:13.699 --> 00:59:15.579
useful. It was practical, right? At the time,

00:59:15.599 --> 00:59:18.940
it was never sexy. Before he joined GenFund,

00:59:19.019 --> 00:59:22.260
he was working for another investment firm called

00:59:22.260 --> 00:59:27.480
Green Spring Associates in Maryland. And at the

00:59:27.480 --> 00:59:30.880
time, he was, you know, doing a lot of research

00:59:30.880 --> 00:59:35.489
about all the funds in the world. So he is well

00:59:35.489 --> 00:59:37.550
aware of all the backgrounds and he did a lot

00:59:37.550 --> 00:59:41.369
of deals. And then he came back to China and

00:59:41.369 --> 00:59:44.710
then joined GenFund. And as he described on his

00:59:44.710 --> 00:59:47.869
official page, he said that because I only studied,

00:59:48.070 --> 00:59:52.690
and sort of sarcastically to himself, because

00:59:52.690 --> 00:59:54.849
I only studied accounting, so the investment

00:59:54.849 --> 00:59:57.730
portfolio that I cover are quite broad, ranging

00:59:57.730 --> 01:00:02.030
from... uh hard tech like companies like momenta

01:00:02.030 --> 01:00:05.389
and credit ai to like consumption everyday use

01:00:05.389 --> 01:00:10.369
like amira and ramen talk to uh content and even

01:00:10.369 --> 01:00:13.809
a company that we both know casbox which is a

01:00:13.809 --> 01:00:18.590
ex -googler started company And out of all the

01:00:18.590 --> 01:00:20.809
companies, I found other companies that I actually

01:00:20.809 --> 01:00:24.969
know personally. Plantag was one of my classmates'

01:00:25.269 --> 01:00:27.190
company and that he also invested in. Apparently,

01:00:27.710 --> 01:00:31.309
it's a very small world. So he seems to really

01:00:31.309 --> 01:00:34.809
invest it in a broad range of things. But nothing

01:00:34.809 --> 01:00:40.170
super big up until Manus. Exactly. Very good

01:00:40.170 --> 01:00:46.940
eye on that. And he had a moment. on that he

01:00:46.940 --> 01:00:49.440
had a moment on that yeah i'll tell you that

01:00:49.440 --> 01:00:53.860
just in a minute but he invested in red all together

01:00:53.860 --> 01:00:56.579
five times and i'll break it down for you the

01:00:56.579 --> 01:00:59.320
first one is the 1 million rmb at the hackathon

01:00:59.320 --> 01:01:05.800
the second time is with 19 go technology um and

01:01:05.800 --> 01:01:12.179
the third time which I can only find in one of

01:01:12.179 --> 01:01:16.280
the videos that he had done a lot early on, which

01:01:16.280 --> 01:01:20.940
is with a company called Benchling. It's not

01:01:20.940 --> 01:01:24.380
Benchling, the US Benchling, but it's Red's idea

01:01:24.380 --> 01:01:27.139
of getting Benchling to China. Right, right.

01:01:27.179 --> 01:01:30.000
Getting Benchling to China. And Benchling is,

01:01:30.099 --> 01:01:36.610
for context, it's a Slack for... biotech or science

01:01:36.610 --> 01:01:39.510
scientific research and development that's right

01:01:39.510 --> 01:01:43.510
that's right so getting that to china and it

01:01:43.510 --> 01:01:49.050
is i think when you say red has a taste or knows

01:01:49.050 --> 01:01:51.789
when to stop did you mean that because everything

01:01:51.789 --> 01:01:55.329
i mean when he sold uh the wechat assistance

01:01:55.329 --> 01:01:58.989
when he stopped benchling and later on when he

01:01:58.989 --> 01:02:03.690
stopped the ai browser yeah yeah yeah so So Benchling

01:02:03.690 --> 01:02:08.070
or slash Jianji, the Benchling China version,

01:02:08.210 --> 01:02:12.329
was the third time that Yuan has invested in

01:02:12.329 --> 01:02:18.429
him. And then the fourth time is Monica. Monica

01:02:18.429 --> 01:02:21.010
.im, yeah, the butterfly effect, which we'll

01:02:21.010 --> 01:02:25.710
get to later. To your point on why Yuan also

01:02:25.710 --> 01:02:29.929
needed this victory, he apparently is a very

01:02:29.929 --> 01:02:35.280
sentimental person. Right? Yes. Right? How would

01:02:35.280 --> 01:02:39.219
you describe him? I heard one podcast where he

01:02:39.219 --> 01:02:42.699
self -proclaims that he cries at every moment

01:02:42.699 --> 01:02:46.599
that he can. Healthy. That's really healthy.

01:02:46.840 --> 01:02:49.260
It's very healthy. It's very healthy. Yeah. So

01:02:49.260 --> 01:02:54.980
he seems to be a very loving person. They're

01:02:54.980 --> 01:02:57.829
usually not this type. i know right yeah so i

01:02:57.829 --> 01:03:00.710
was just checking sense check are they all like

01:03:00.710 --> 01:03:05.510
that uh is this average would someone who work

01:03:05.510 --> 01:03:08.929
in the vc or p sector be as sentimental or as

01:03:08.929 --> 01:03:12.829
caring as as you know i think it actually prevents

01:03:12.829 --> 01:03:16.440
you from I don't want to say that, but I agree.

01:03:16.579 --> 01:03:18.760
I know what you mean. You're supposed to be very

01:03:18.760 --> 01:03:21.559
number oriented. You calculate your, you know,

01:03:21.559 --> 01:03:24.420
like how much would this company worth and, you

01:03:24.420 --> 01:03:27.579
know, but so adding that extra emotional touch

01:03:27.579 --> 01:03:31.619
is probably detrimental as an investor. I agree.

01:03:31.760 --> 01:03:35.699
But at the same time, I think it, it is what

01:03:35.699 --> 01:03:40.559
made him stick with red for so long. And so you

01:03:40.559 --> 01:03:45.110
can argue that this personality is, what created

01:03:45.110 --> 01:03:48.969
his ultimate success. Yeah, I agree. Yeah, that's

01:03:48.969 --> 01:03:52.269
a very good way to put it. So as a very compassionate

01:03:52.269 --> 01:03:58.650
and gentle investor, in the year 2022, Yuan wrote

01:03:58.650 --> 01:04:04.739
Bob a very short note. And he said, And because

01:04:04.739 --> 01:04:07.440
he started working in 2014, so by 2022, it was

01:04:07.440 --> 01:04:10.519
eight years. For eight years, I failed to produce

01:04:10.519 --> 01:04:13.400
any work I could be proud of, perhaps the peak

01:04:13.400 --> 01:04:15.739
of my career when I was promoted as a partner.

01:04:16.320 --> 01:04:19.340
Since joining GenFan, I've grown under the immense

01:04:19.340 --> 01:04:21.920
favor and encouraging words of you and Anna.

01:04:22.199 --> 01:04:24.980
If I have achieved success, it will be a perfect

01:04:24.980 --> 01:04:28.480
story of repaying your kindness. And if I don't,

01:04:28.480 --> 01:04:31.179
my robot Zhen Fan will be like a favored eunuch

01:04:31.179 --> 01:04:35.170
in the imperial court. That's what he said. you

01:04:35.170 --> 01:04:39.909
see yeah yeah and i think he actually meant what

01:04:39.909 --> 01:04:42.550
he said about how do you respond to a message

01:04:42.550 --> 01:04:47.829
like this he didn't respond um yeah but i think

01:04:47.829 --> 01:04:52.050
um and you know like how how much uh so obviously

01:04:52.050 --> 01:04:53.949
i think everyone at jinfang was helping each

01:04:53.949 --> 01:04:58.579
other out and yuan helped xiao hung so much in

01:04:58.579 --> 01:05:01.360
the past 10 years, besides investing in him like

01:05:01.360 --> 01:05:05.019
five different times, he also pulled, do you

01:05:05.019 --> 01:05:06.940
know how many WeChat groups he pulled for him?

01:05:07.199 --> 01:05:13.139
Hundreds, probably. 130. But I, so, okay. I think

01:05:13.139 --> 01:05:16.699
he stuck, he sticked with Red, continued to invest

01:05:16.699 --> 01:05:21.139
in his ventures because the first exit with the

01:05:21.139 --> 01:05:24.940
WeChat assistants, It was positive. It wasn't

01:05:24.940 --> 01:05:27.119
like huge. It wasn't $2 billion exit, but it

01:05:27.119 --> 01:05:32.159
was a good, decent exit. I think the BenchLink

01:05:32.159 --> 01:05:35.019
equivalent didn't end up making money or not

01:05:35.019 --> 01:05:38.460
that they talk about it. Monica was cash positive.

01:05:38.739 --> 01:05:42.460
It was great. Right? Yeah. And so there is no

01:05:42.460 --> 01:05:45.920
reason to not invest in this guy. It was on the

01:05:45.920 --> 01:05:48.880
positive trajectory. Yeah. And for the only other

01:05:48.880 --> 01:05:51.340
portfolio company that I know, which is CastBox.

01:05:52.199 --> 01:05:54.659
they're doing pretty well. It's just, you're

01:05:54.659 --> 01:05:57.099
not, it's not madness, but it's, you know, it's

01:05:57.099 --> 01:06:00.820
doing well, but maybe he wants to be the person.

01:06:01.079 --> 01:06:03.239
Explosive. He wants to be the person that's known

01:06:03.239 --> 01:06:06.300
for investing in the Red Book, which is what

01:06:06.300 --> 01:06:11.980
Jim found is, you know, known for. And being

01:06:11.980 --> 01:06:16.099
that and being promoted, you know, into his position,

01:06:16.239 --> 01:06:18.699
he thought that there should be at least one

01:06:18.699 --> 01:06:21.320
of these companies to show for. Makes sense.

01:06:21.400 --> 01:06:23.659
So he was aiming for something really high. Right.

01:06:23.719 --> 01:06:26.820
And it doesn't mean that's. by any means a failure

01:06:26.820 --> 01:06:32.599
because it must be net positive. It must be 1000x.

01:06:33.340 --> 01:06:37.260
But Xiao Hong did get rejected by a lot of other

01:06:37.260 --> 01:06:42.219
VCs in China. That's a fact. And he got invested

01:06:42.219 --> 01:06:44.559
by Hongshan and Sequoia and Tencent because that's

01:06:44.559 --> 01:06:47.239
probably the only ones that he's got. And all

01:06:47.239 --> 01:06:50.239
the rest of the objections are very, very similar.

01:06:50.440 --> 01:06:53.579
And they always ask, come back to Xiao Hong with,

01:06:53.679 --> 01:06:56.260
oh, what is remote? doesn't seem to have a lot

01:06:56.260 --> 01:06:58.539
of moot is the market big enough for the things

01:06:58.539 --> 01:07:00.659
that you develop it's so interesting because

01:07:00.659 --> 01:07:07.380
it reminds me of pop mart because it was so hard

01:07:07.380 --> 01:07:10.619
for wanting to get investment as well and and

01:07:10.619 --> 01:07:14.719
both of them came from you know not so i wonder

01:07:14.719 --> 01:07:19.320
if there's this bias towards certain type of

01:07:19.320 --> 01:07:22.219
founders from the vc community at that stage

01:07:22.219 --> 01:07:26.000
and so it's extra hard for them to raise money

01:07:26.000 --> 01:07:31.139
and as a result a lot of people miss out in like

01:07:31.139 --> 01:07:33.599
we realize a lot of people miss out recently

01:07:33.599 --> 01:07:37.960
because of all the uh growth trajectories and

01:07:37.960 --> 01:07:41.000
all the ipos and whatnot but the investment happened

01:07:41.000 --> 01:07:44.219
like six seven eight years ago yeah so maybe

01:07:44.219 --> 01:07:48.460
there's a fundamental mindset back then yeah

01:07:48.460 --> 01:07:51.380
that doesn't favor any of these guys i agree

01:07:51.380 --> 01:07:53.860
yeah yeah it wasn't the right time or people

01:07:53.860 --> 01:07:56.360
didn't believe yeah and and now i don't know

01:07:56.360 --> 01:07:59.260
if like i don't know if they shifted this mindset

01:07:59.260 --> 01:08:02.400
or if it's still like oh we're gonna maybe it's

01:08:02.400 --> 01:08:05.300
starting to change yeah yeah but we'll go on

01:08:05.300 --> 01:08:08.500
to the the sass bit because sass is man is is

01:08:08.500 --> 01:08:11.400
we're we're almost there we're almost there so

01:08:11.400 --> 01:08:15.119
almost there So the role of Zhen Fan in MANAS

01:08:15.119 --> 01:08:20.199
was very critical, right? Day one investor, right?

01:08:20.279 --> 01:08:24.520
Also first institutional backer. And Zhen Fan

01:08:24.520 --> 01:08:27.460
also attracted some other people who came on

01:08:27.460 --> 01:08:36.079
to invest for Monica later on. And while Liu

01:08:36.079 --> 01:08:40.199
Yuan was leading the investment, literally everyone

01:08:40.199 --> 01:08:42.319
at Zhen Fan, I think, was participating in helping

01:08:42.319 --> 01:08:46.380
out. uh the crew yeah it's a small company it's

01:08:46.380 --> 01:08:48.300
a it's a pretty small company that's right that's

01:08:48.300 --> 01:08:51.979
right butterfly effect was officially registered

01:08:51.979 --> 01:08:56.720
in june 2022 okay um but it was around that time

01:08:56.720 --> 01:09:01.220
it was very blurry right red was the person who

01:09:01.220 --> 01:09:08.579
started uh monica .im uh so he acquired oh right

01:09:08.579 --> 01:09:14.890
okay he bought the chrome uh extension which

01:09:14.890 --> 01:09:20.649
is called chat gpt for google that was such a

01:09:20.649 --> 01:09:24.869
such a funny story okay so what i got here which

01:09:24.869 --> 01:09:27.210
which i think will perfectly lead on to the chat

01:09:27.210 --> 01:09:30.449
gpt for google story step number this was before

01:09:30.449 --> 01:09:33.369
the two other founder even exactly this is before

01:09:33.369 --> 01:09:38.600
this is just red And his old co -founders. Exactly.

01:09:38.640 --> 01:09:40.460
Who are still in the team. Who are still in the

01:09:40.460 --> 01:09:43.100
team. But this is when it's starting to take

01:09:43.100 --> 01:09:47.939
shape, like monica .im. And at the time, all

01:09:47.939 --> 01:09:52.159
the AI tools are starting to pop out. Red thinks

01:09:52.159 --> 01:09:55.579
that, okay, I can't be depressed. Gotta do something.

01:09:55.760 --> 01:10:01.720
Gotta do something. And so in 2022, he realized

01:10:01.720 --> 01:10:05.170
that, oh, maybe it could be a good idea. To build

01:10:05.170 --> 01:10:08.409
something that I'm familiar with. An extension

01:10:08.409 --> 01:10:11.510
on a browser. Yeah. Maybe I can start with that.

01:10:11.670 --> 01:10:14.829
Yeah. Productivity tools. I'm a tools guy. I

01:10:14.829 --> 01:10:17.050
think because he benefited from the WeChat platform,

01:10:17.310 --> 01:10:19.229
he thought that, you know, my next startup will

01:10:19.229 --> 01:10:22.170
always be platform based. And what's another,

01:10:22.350 --> 01:10:24.550
what's a better platform than Google Chrome?

01:10:25.399 --> 01:10:28.279
I think they started doing all right with the

01:10:28.279 --> 01:10:30.220
extension because there's some usage. And then

01:10:30.220 --> 01:10:32.439
they quickly realized there's a cap to extension

01:10:32.439 --> 01:10:35.520
usage because most people don't use Chrome with

01:10:35.520 --> 01:10:38.539
extensions. Right. And once again, there will

01:10:38.539 --> 01:10:41.340
be a lot of people criticizing that. Even if

01:10:41.340 --> 01:10:45.319
monica .im, what is monica basically? It's kind

01:10:45.319 --> 01:10:49.039
of an aggregator of all the existing LLMs. You

01:10:49.039 --> 01:10:53.239
package it, right? And so that when you open

01:10:53.239 --> 01:10:56.220
that extension, you can freely switch around

01:10:56.220 --> 01:10:59.539
different LLMs to do your work. So a lot of people

01:10:59.539 --> 01:11:04.300
are criticizing. On Red's post, he posted that.

01:11:04.880 --> 01:11:07.300
Also, I think this is very much from a product

01:11:07.300 --> 01:11:09.880
manager point of view. And he's very less so

01:11:09.880 --> 01:11:13.560
about, oh, going like deep, deep tech, having

01:11:13.560 --> 01:11:16.239
my own model. He's more so about how do I make

01:11:16.239 --> 01:11:20.739
people use? What value do I bring to the users?

01:11:21.350 --> 01:11:25.510
even if it's just an aggregator, he's like really

01:11:25.510 --> 01:11:27.649
going at it. And then they wanted to benchmark

01:11:27.649 --> 01:11:33.090
to Jasper. Oh. Yeah, Jasper AI, which was a writing

01:11:33.090 --> 01:11:35.869
aid. But the problem is now I have the product.

01:11:35.909 --> 01:11:38.989
I don't have enough traffic. I need to get on

01:11:38.989 --> 01:11:42.329
or at least learn from the best and how to drive

01:11:42.329 --> 01:11:45.250
that traffic to the extension that I have. And

01:11:45.250 --> 01:11:48.649
Red was very much inspired by... And I'm not

01:11:48.649 --> 01:11:52.909
kidding. He was reading Anta's CEO talking about

01:11:52.909 --> 01:11:56.170
him acquiring Amersports. So if you're interested,

01:11:56.289 --> 01:11:59.369
you can also have a listen to the Anta episode.

01:11:59.510 --> 01:12:02.229
Besides Anta's story, and he was also looking

01:12:02.229 --> 01:12:06.590
at Constellation Software. He was wanting to

01:12:06.590 --> 01:12:10.829
just acquire the traffic and not building it

01:12:10.829 --> 01:12:14.949
on his own. And he did one thing, which is to...

01:12:15.229 --> 01:12:18.289
get the traffic from ChatGPT for Google and that

01:12:18.289 --> 01:12:24.199
product is made by Red's alum. a guy who was

01:12:24.199 --> 01:12:27.619
older, graduated a little earlier, and ChatGPT

01:12:27.619 --> 01:12:30.659
for Google got its traffic from 100K to 500K

01:12:30.659 --> 01:12:33.720
user in just a month. It is a fantastic extension,

01:12:33.899 --> 01:12:39.600
about 200 lines of code and took about a weekend

01:12:39.600 --> 01:12:43.340
to accomplish. So Rhett was not the only person

01:12:43.340 --> 01:12:47.039
who wants to get that extension. There were...

01:12:47.739 --> 01:12:51.279
a couple other contenders who want to get the

01:12:51.279 --> 01:12:53.159
Chateaubriand for gold because everyone was rushing

01:12:53.159 --> 01:12:56.199
at the same time. And Rhett wasn't also necessarily

01:12:56.199 --> 01:13:01.180
pitching the alum at the highest price. His pitch

01:13:01.180 --> 01:13:05.060
was actually on the lower end. And the alum went

01:13:05.060 --> 01:13:08.500
back home and told Rhett that I need to negotiate

01:13:08.500 --> 01:13:11.979
with my wife whether that price was okay. So

01:13:11.979 --> 01:13:15.220
Rhett waited until the alum came back to him

01:13:15.220 --> 01:13:18.289
and said, oh. i can sell it to you now you can

01:13:18.289 --> 01:13:21.369
clear it and red did not wait for the investors

01:13:21.369 --> 01:13:25.250
fund he made the transfer to the alum immediately

01:13:25.250 --> 01:13:28.609
on his phone and rejected the higher uh the the

01:13:28.609 --> 01:13:31.130
alum rejected the higher offer and then this

01:13:31.130 --> 01:13:34.489
tool later became very critical in monica's traffic

01:13:34.489 --> 01:13:37.689
the acquisition of amor sport by anta prompted

01:13:37.689 --> 01:13:42.039
his decision to acquire rather than build from

01:13:42.039 --> 01:13:45.159
scratch even though he builds monica from scratch

01:13:45.159 --> 01:13:48.260
but chat gpt for google helped with the traffic

01:13:48.260 --> 01:13:50.560
don't start from scratch when you don't have

01:13:50.560 --> 01:13:55.340
to onto the right so browser steady cash cow

01:13:55.340 --> 01:14:00.000
12 million dollar arr and they realize that it's

01:14:00.000 --> 01:14:02.279
not going to continue to grow and it's not going

01:14:02.279 --> 01:14:05.399
to be there forever so let's find something else

01:14:06.220 --> 01:14:11.600
And Red has an idea for an AI browser. So he

01:14:11.600 --> 01:14:15.720
has the AI extension. He thought AI browser would

01:14:15.720 --> 01:14:18.760
be the natural next step. I don't know why somehow

01:14:18.760 --> 01:14:23.659
taking over Chrome is on everyone's mind. So

01:14:23.659 --> 01:14:27.300
that's his first idea. And he went to different

01:14:27.300 --> 01:14:30.880
people with the idea, hoping that they would

01:14:30.880 --> 01:14:35.880
join his team. And that's where, I guess, GenFund

01:14:35.880 --> 01:14:40.960
and Peak and Heidkau comes in. How did Red know

01:14:40.960 --> 01:14:43.659
Peak? So he was not really working for anything

01:14:43.659 --> 01:14:46.220
at the time. He was working as an entrepreneur

01:14:46.220 --> 01:14:49.960
in residence at GenFund at the time. So he was

01:14:49.960 --> 01:14:52.340
just observing and sitting in all of the meetings.

01:14:53.020 --> 01:14:55.720
And he had browser development experience with

01:14:55.720 --> 01:15:00.520
Mammoth. And Red had a proposal to Peak. And

01:15:00.520 --> 01:15:03.140
his proposal goes... I know you have built a

01:15:03.140 --> 01:15:06.619
web browser before and you've also built an LLM.

01:15:07.380 --> 01:15:10.479
You've built a search engine. Do you want to

01:15:10.479 --> 01:15:13.699
build a three -in -one with me? And Peak said

01:15:13.699 --> 01:15:20.739
yes. So the end of 2023, that's when Peak and

01:15:20.739 --> 01:15:23.319
Red started engaging on more conversations about

01:15:23.319 --> 01:15:26.659
building. You know, something like Manus, but

01:15:26.659 --> 01:15:28.979
it wasn't called Manus at the time. It's something

01:15:28.979 --> 01:15:33.800
completely different now. The AI Chrome. Yeah,

01:15:33.880 --> 01:15:38.800
the browser thing. March 2024, that's when Peak

01:15:38.800 --> 01:15:43.539
officially joined the team. And later in May

01:15:43.539 --> 01:15:48.529
2024, that's when Tao joined the team. a similar

01:15:48.529 --> 01:15:53.310
time period early 2024 right and i i think the

01:15:53.310 --> 01:15:56.569
setup was that the original monica im team is

01:15:56.569 --> 01:15:59.689
still in wuhan at that time right and then both

01:15:59.689 --> 01:16:03.010
peak and tower in beijing and so this new project

01:16:03.010 --> 01:16:06.050
it's kind of like game science they have like

01:16:06.050 --> 01:16:11.039
a A small new project team. And then you kind

01:16:11.039 --> 01:16:13.979
of gradually shift people over as the new project

01:16:13.979 --> 01:16:16.539
gets bigger. But at first, it's just a very small

01:16:16.539 --> 01:16:20.279
team trying to develop this new product. And

01:16:20.279 --> 01:16:24.199
the new product, I think, started in March, April,

01:16:24.340 --> 01:16:27.180
like right after they joined. And then lasted

01:16:27.180 --> 01:16:32.539
all the way until second half. I want to say

01:16:32.539 --> 01:16:38.199
August or so of 2024. uh so maybe a few five

01:16:38.199 --> 01:16:41.279
months of development time yeah for the ai browser

01:16:41.279 --> 01:16:45.819
and somehow during those five months all of them

01:16:45.819 --> 01:16:50.979
eventually realized that it might not not be

01:16:50.979 --> 01:16:55.439
the product yeah again going back to you know

01:16:55.439 --> 01:16:58.560
they already developed the product by august

01:16:58.560 --> 01:17:02.090
so they could either release it or scratch it

01:17:02.090 --> 01:17:05.149
and i would imagine a lot of people would release

01:17:05.149 --> 01:17:07.909
it just to see where it goes so for the three

01:17:07.909 --> 01:17:11.090
of them to actually say let's scratch this and

01:17:11.090 --> 01:17:14.510
start from scratch it's really bad it's well

01:17:15.319 --> 01:17:17.500
I think it's courage, right? Because you could

01:17:17.500 --> 01:17:19.560
just say release and then maintenance, but they

01:17:19.560 --> 01:17:22.340
don't want the opportunity costs because for

01:17:22.340 --> 01:17:24.760
every hour spent on maintenance and trying to

01:17:24.760 --> 01:17:27.659
figure it out and still... Cost. Exactly. So

01:17:27.659 --> 01:17:30.479
you either shut it down completely and pivot

01:17:30.479 --> 01:17:34.539
or you don't let go of your babies and then you...

01:17:34.890 --> 01:17:37.109
It turns into something else that you don't want

01:17:37.109 --> 01:17:38.989
it to turn into and it eats up your time. So

01:17:38.989 --> 01:17:41.550
I think it's actually a good decision. The other

01:17:41.550 --> 01:17:44.350
thing that's related to kind of the AI period

01:17:44.350 --> 01:17:48.069
of software development, right? They develop

01:17:48.069 --> 01:17:50.369
something very quickly and it's all application

01:17:50.369 --> 01:17:53.770
layer. The opportunity cost of pivoting, the

01:17:53.770 --> 01:17:58.859
sunk cost is not as big as a few years ago. So

01:17:58.859 --> 01:18:02.039
pivoting to Manus and developing Manus. I actually

01:18:02.039 --> 01:18:04.239
looked at the Manus development period that they

01:18:04.239 --> 01:18:08.239
quoted. It started from September, probably mid

01:18:08.239 --> 01:18:11.359
to late September, like a month after they scratched

01:18:11.359 --> 01:18:14.979
the AI browser project. And the first prototype

01:18:14.979 --> 01:18:17.500
was actually done in mid -January. That's three

01:18:17.500 --> 01:18:19.560
and a half months. That's like not even four

01:18:19.560 --> 01:18:23.819
months. Manus was created. And they waited until

01:18:23.819 --> 01:18:27.359
March for the release only because Cloud Sonnet

01:18:28.010 --> 01:18:32.109
release timeline was 3 .7 was in late february

01:18:32.109 --> 01:18:34.890
and they wanted to wait for that for both a better

01:18:34.890 --> 01:18:39.369
model base model and for publicity like you can

01:18:39.369 --> 01:18:41.890
kind of use the leverage the extra publicity

01:18:41.890 --> 01:18:46.899
carry over from from cloud i guess relatively

01:18:46.899 --> 01:18:49.579
it was a hard decision to just say forget about

01:18:49.579 --> 01:18:51.899
this thing that i originally promised you guys

01:18:51.899 --> 01:18:55.680
that we'll do um but it's also like very very

01:18:55.680 --> 01:18:58.100
low opportunity cost to say oh let's develop

01:18:58.100 --> 01:19:02.920
something else yeah and this something else is

01:19:02.920 --> 01:19:05.890
just Turned out to be the right decision because

01:19:05.890 --> 01:19:09.569
there are other AI browsers that are still in

01:19:09.569 --> 01:19:12.729
development right now. And I think the amount

01:19:12.729 --> 01:19:17.430
of splash in the market compared to Manas to

01:19:17.430 --> 01:19:20.130
another AI browser. They're doing decently well,

01:19:20.250 --> 01:19:23.869
but just not to the degrees that Manas has been

01:19:23.869 --> 01:19:26.489
gaining traction. When they decided to drop the

01:19:26.489 --> 01:19:29.970
browser business, they realized that all three

01:19:29.970 --> 01:19:32.689
are very, you know. They all have a lot of product

01:19:32.689 --> 01:19:35.449
experience. So for three, I would say two and

01:19:35.449 --> 01:19:38.449
a half season product managers to see this and

01:19:38.449 --> 01:19:42.289
to say, maybe let's not ship this. They must

01:19:42.289 --> 01:19:47.649
have like an instinctual feeling that this is

01:19:47.649 --> 01:19:49.369
not going to work. It's not going to work, right?

01:19:49.630 --> 01:19:52.710
The three of them, the whole team had about 2

01:19:52.710 --> 01:19:57.189
.5 weeks just wandering around and not knowing

01:19:57.189 --> 01:19:59.909
how is this going to turn into something else.

01:20:00.880 --> 01:20:05.699
And in their office, they've observed a lot of

01:20:05.699 --> 01:20:09.800
the non -technical folks, like the non -engineers,

01:20:09.819 --> 01:20:15.539
were using cursor. I think for most non -technical

01:20:15.539 --> 01:20:21.000
folks, aka like me, to see a lot of codes running

01:20:21.000 --> 01:20:24.479
on your laptop and commanding to those codes

01:20:24.479 --> 01:20:28.840
can be very intimidating. And that was the issue.

01:20:29.609 --> 01:20:34.109
for the non -technical folks in that office they

01:20:34.109 --> 01:20:39.250
didn't know what's the best way to use it but

01:20:39.250 --> 01:20:41.850
them coming from a technical background and observing

01:20:41.850 --> 01:20:45.649
this they started with a few conclusions they

01:20:45.649 --> 01:20:50.569
realized that oh so instead of us giving better

01:20:50.569 --> 01:20:53.569
prompts is there a way that we can make better

01:20:53.569 --> 01:20:57.600
use of the like we can interpret better to the

01:20:57.600 --> 01:21:00.140
machines that what human want to do. So can we

01:21:00.140 --> 01:21:04.420
possibly do like a mid layer in between that,

01:21:04.539 --> 01:21:09.159
like to do a better UI and UX and slash make

01:21:09.159 --> 01:21:12.300
agents run on clouds and make this interface

01:21:12.300 --> 01:21:16.600
much better to create a tool, not just for engineers,

01:21:16.720 --> 01:21:21.220
but for professional users. So this got them

01:21:21.220 --> 01:21:24.020
thinking that can we implement these? So they

01:21:24.020 --> 01:21:29.270
started trying. So a new project was born shortly

01:21:29.270 --> 01:21:33.489
two and a half weeks after they had the failed

01:21:33.489 --> 01:21:36.829
browser launch. The browser that they were launched.

01:21:37.090 --> 01:21:39.850
Yeah, it decided not to launch. The new project

01:21:39.850 --> 01:21:43.550
isn't called Manus yet. And it's called Browser

01:21:43.550 --> 01:21:50.810
in Browser in the Air. Airbnb. Browser in Browser

01:21:50.810 --> 01:21:55.079
in the Air. Okay. And this project started, like

01:21:55.079 --> 01:21:58.340
you said, September 2024. Oh, that's interesting.

01:21:58.659 --> 01:22:01.140
Yeah. So the original idea is actually just a

01:22:01.140 --> 01:22:03.659
browser, not an OS in the air. That's right.

01:22:03.779 --> 01:22:07.079
Because they all have very rich experience in

01:22:07.079 --> 01:22:11.180
designing browsers. Right. I'm glad they didn't

01:22:11.180 --> 01:22:14.979
keep this Airbnb concept. Right. So obviously

01:22:14.979 --> 01:22:18.760
you can't really call yourself Airbnb. Once that

01:22:18.760 --> 01:22:22.500
product consolidates and everyone was, you know,

01:22:22.500 --> 01:22:25.920
feeling that, oh, this is a ready to ship product.

01:22:26.489 --> 01:22:29.810
And we all feel that the quality of the product

01:22:29.810 --> 01:22:32.529
that it's paying, it's really going somewhere,

01:22:32.670 --> 01:22:35.149
right? So they decided that, oh, maybe we should

01:22:35.149 --> 01:22:38.430
name this project. They had a few options. Option

01:22:38.430 --> 01:22:41.069
number one, which I don't think you'll even like

01:22:41.069 --> 01:22:42.930
it. I didn't know they had options. Option number

01:22:42.930 --> 01:22:44.489
one is called Dodo Bird or something like that.

01:22:44.489 --> 01:22:46.590
Oh, I do know they have options. Okay, yeah.

01:22:46.630 --> 01:22:48.829
And Anna didn't like it. And then option number

01:22:48.829 --> 01:22:51.130
two was called Vida, which is life and energy

01:22:51.130 --> 01:22:52.770
because they want something that was energetic,

01:22:53.069 --> 01:22:56.640
vetoed. And option number three. was manners

01:22:56.640 --> 01:23:02.739
which was the mit slogan hand yes and peak picked

01:23:02.739 --> 01:23:06.560
it right i mean dodo burst never gonna work because

01:23:06.560 --> 01:23:10.920
it's extinct and it doesn't fly so right then

01:23:10.920 --> 01:23:14.079
a good call yeah yeah yeah um but yeah the opposite

01:23:14.079 --> 01:23:18.220
of that manners is a is a very good way of describing

01:23:18.960 --> 01:23:21.960
This project, because it is the hand to the mind

01:23:21.960 --> 01:23:24.380
that is already been created with the likes of

01:23:24.380 --> 01:23:27.020
ChatGPT and Gemini. It's perfect. Yeah, it's

01:23:27.020 --> 01:23:30.979
perfect. We're about to get to the March 6th

01:23:30.979 --> 01:23:34.159
and March 7th product release, right? Yeah. And

01:23:34.159 --> 01:23:37.819
so just, I guess, as context for anyone who doesn't

01:23:37.819 --> 01:23:40.600
know what happened last year, not before this,

01:23:40.699 --> 01:23:45.159
before March 7th, in January last year, DeepSeek

01:23:45.159 --> 01:23:49.539
launched its first... I guess, interface, right?

01:23:49.680 --> 01:23:53.380
Yes. DeepSeek R1. Blew up the internet. Blew

01:23:53.380 --> 01:23:56.119
up the internet. And if you want to learn more

01:23:56.119 --> 01:23:58.359
about how it blew the internet, you can listen

01:23:58.359 --> 01:24:00.420
to the DeepSeek episode that we launched last

01:24:00.420 --> 01:24:05.220
year. Season one, episode two. Yeah. And then

01:24:05.220 --> 01:24:09.960
in March, they launched. Yeah. So it's been called

01:24:09.960 --> 01:24:13.000
another... It's one of those other China DeepSeek

01:24:13.000 --> 01:24:17.699
moment. OpenAI doing some... similar stuff and

01:24:17.699 --> 01:24:20.140
they said we might come up with a agent that

01:24:20.140 --> 01:24:21.840
will do your work for you and we might charge

01:24:21.840 --> 01:24:24.600
you that ten thousand dollars already yeah the

01:24:24.600 --> 01:24:27.319
rumors like reddit has it you know like ten thousand

01:24:27.319 --> 01:24:28.920
dollars we're gonna charge you for this agent

01:24:28.920 --> 01:24:33.079
so before they're like If you do listen to our

01:24:33.079 --> 01:24:35.020
Deep Seek episode, they said, we're going to

01:24:35.020 --> 01:24:39.220
charge you maybe $1 ,000 for a reasoning model.

01:24:39.399 --> 01:24:42.960
Deep Seek comes out, reasoning model free. And

01:24:42.960 --> 01:24:47.439
a reasonable price for a reasoning model. Sorry.

01:24:48.979 --> 01:24:53.600
No puns intended. And then a year later, another

01:24:53.600 --> 01:24:58.840
Chinese comes out. See your pattern. Yeah. We're

01:24:58.840 --> 01:25:01.020
going to charge you $40 for what you wanted to

01:25:01.020 --> 01:25:06.859
charge $10 ,000 for. And we just talked about

01:25:06.859 --> 01:25:11.600
the scrapped AI browser project, right? The feeling

01:25:11.600 --> 01:25:16.520
in the team pre -release comparing Manos to the

01:25:16.520 --> 01:25:19.939
AI browser project was night and day. With the

01:25:19.939 --> 01:25:22.500
AR Bros project, everyone's like, do we really?

01:25:22.560 --> 01:25:24.840
No, I mean, I know we'd spend effort, but let's

01:25:24.840 --> 01:25:27.840
not release it. We're not sure, right? With Manos,

01:25:27.939 --> 01:25:31.619
everyone's like, this is gonna be the bomb. So

01:25:31.619 --> 01:25:36.119
dope. We're just gonna drop this on March 6th

01:25:36.119 --> 01:25:39.439
and wait for publicity. Everyone had this confidence.

01:25:39.739 --> 01:25:42.729
Oh my God. Or at least... peak set that yeah

01:25:42.729 --> 01:25:47.210
right it's like we know this is gonna blow up

01:25:47.210 --> 01:25:50.550
the internet and so they filmed this very simple

01:25:50.550 --> 01:25:53.710
video yeah you might have more stuff but it's

01:25:53.710 --> 01:25:56.930
a very simple video put it up on youtube did

01:25:56.930 --> 01:25:59.390
blow up the internet do you know what was my

01:25:59.390 --> 01:26:01.229
first reaction to that video when i watched it

01:26:01.229 --> 01:26:04.800
last year Was that an AI talking? Oh, really?

01:26:05.100 --> 01:26:06.880
Yeah, because there were so many AI -generated

01:26:06.880 --> 01:26:09.520
videos. There were so many backdrops. They said

01:26:09.520 --> 01:26:11.399
they filmed it in the... Because they didn't

01:26:11.399 --> 01:26:13.420
have a proper office, they filmed it on the first

01:26:13.420 --> 01:26:16.579
floor of wherever they were working, their co

01:26:16.579 --> 01:26:18.619
-working space, and there was just people walking

01:26:18.619 --> 01:26:21.159
around. And then they had to black out a section

01:26:21.159 --> 01:26:23.500
because there was a janitor walking around or

01:26:23.500 --> 01:26:26.220
something. So anyways, it was pretty scrappy,

01:26:26.340 --> 01:26:29.479
but it was, you know, like... goes back to the

01:26:29.479 --> 01:26:32.199
product mindset if you believe very much in your

01:26:32.199 --> 01:26:34.680
product who cares about the marketing who cares

01:26:34.680 --> 01:26:38.420
about how you reveal it right and the video itself

01:26:38.420 --> 01:26:43.039
was cut really nicely yeah um when it was published

01:26:43.039 --> 01:26:45.939
it took a few days for silicon valley to catch

01:26:45.939 --> 01:26:49.680
up but when it did it just because of all the

01:26:49.680 --> 01:26:55.020
uh endorsement from investors uh from the tech

01:26:55.020 --> 01:27:01.020
community it did take it so it blew up in the

01:27:01.020 --> 01:27:03.619
term that a lot of influencers a lot of youtubers

01:27:03.619 --> 01:27:09.739
in and out of china are are uh how do you say

01:27:09.739 --> 01:27:13.060
like created a secondary market for the code

01:27:13.060 --> 01:27:21.060
right and so that moment although the madness

01:27:21.060 --> 01:27:24.819
team expected the explosion they weren't expecting

01:27:24.819 --> 01:27:28.399
that level of explosion and so when you get too

01:27:28.399 --> 01:27:31.239
much attention for them at least they actually

01:27:31.239 --> 01:27:35.239
shut themselves down rather than going back out

01:27:35.239 --> 01:27:38.619
into the market as they would typically to tell

01:27:38.619 --> 01:27:42.020
people you know we we didn't do any marketing

01:27:42.020 --> 01:27:44.859
it's not us uh we're gonna do this and this and

01:27:44.859 --> 01:27:47.600
that expect like 1 .5 to launch or whatnot they

01:27:47.600 --> 01:27:49.600
actually kept really quiet during that whole

01:27:49.600 --> 01:27:53.810
phase afterwards I think one is to focus on the

01:27:53.810 --> 01:27:57.229
product development and to ensure that they have

01:27:57.229 --> 01:28:01.689
enough bandwidth for their growth potential and

01:28:01.689 --> 01:28:04.489
to work with their cloud providers, their vendors,

01:28:04.689 --> 01:28:07.329
et cetera, to really focus on the product and

01:28:07.329 --> 01:28:11.050
let the product speak for itself. But also, I'm

01:28:11.050 --> 01:28:14.029
guessing they just don't know how to react because

01:28:14.029 --> 01:28:17.369
nobody had... They were confident, but not that

01:28:17.369 --> 01:28:20.930
confident. I found that the total expenditure

01:28:20.930 --> 01:28:27.390
for the March 6th or 7th product released that

01:28:27.390 --> 01:28:31.789
video, in total, if you don't count Peak's price,

01:28:32.090 --> 01:28:37.369
should account to 39 RMB. For what? That was

01:28:37.369 --> 01:28:40.289
the price to ship the camera from Yusen to Manus'

01:28:40.430 --> 01:28:44.430
hide -in office. That's how much it takes. Wow.

01:28:44.970 --> 01:28:49.439
Peak's appearance. Yeah, Pete's appearance. It's

01:28:49.439 --> 01:28:51.420
like a million dollars by itself. And then the

01:28:51.420 --> 01:28:54.819
reason that he was the person who ended up doing

01:28:54.819 --> 01:28:56.500
the product launch because he's the only one

01:28:56.500 --> 01:28:59.260
who speaks better English on the team. Oh. Right?

01:28:59.359 --> 01:29:01.560
That's actually a potential. We might talk about

01:29:01.560 --> 01:29:03.600
it later on that now they're acquired by Meta.

01:29:04.000 --> 01:29:07.039
How are they supposed to work with Meta? I'm

01:29:07.039 --> 01:29:10.060
sure it can all work out. But it's just when

01:29:10.060 --> 01:29:13.520
they pick the representation of Manus for that

01:29:13.520 --> 01:29:16.340
particular video, Pete became the base. It's

01:29:16.340 --> 01:29:20.300
a good call. It is a good call. Zhen Fan was

01:29:20.300 --> 01:29:24.619
helping so much on reaching the international

01:29:24.619 --> 01:29:29.680
crowds. Anna Fang, the GP, was helping getting

01:29:29.680 --> 01:29:32.560
people's attention on LinkedIn. And Yuan himself

01:29:32.560 --> 01:29:35.479
also personally reached out to a lot of KOLs

01:29:35.479 --> 01:29:38.699
overseas. And they were all really excited for

01:29:38.699 --> 01:29:40.920
her. Because it blew up so much that Junfeng

01:29:40.920 --> 01:29:44.220
had to work for Manos for a period of time. Yeah,

01:29:44.220 --> 01:29:47.899
yeah, yeah. That's right. Which is how they ended

01:29:47.899 --> 01:29:53.439
up getting from March to December 125 million.

01:29:54.199 --> 01:29:57.680
Yeah. Yeah. And they didn't disclose the exact

01:29:57.680 --> 01:30:01.699
number of paid users. Um, but I did a rough calculation

01:30:01.699 --> 01:30:05.220
based on their per month price that they charge.

01:30:05.300 --> 01:30:09.539
It's probably somewhere between 200 ,000 users

01:30:09.539 --> 01:30:15.079
to three, three 30, because, um, there is the

01:30:15.079 --> 01:30:18.140
$20 package, but a lot of their users do pay

01:30:18.140 --> 01:30:21.500
for the $40 package. And some of the power users

01:30:21.500 --> 01:30:27.710
end up paying. not only the one is a $200 package,

01:30:27.930 --> 01:30:31.310
but also they top up on tokens because it only

01:30:31.310 --> 01:30:33.609
gives you a fixed amount and then you would need

01:30:33.609 --> 01:30:39.270
to top up in the end. Sidebar, I created, and

01:30:39.270 --> 01:30:41.810
I just showed you this, I created our Ascend

01:30:41.810 --> 01:30:45.710
podcast website, or I was trying it out if it

01:30:45.710 --> 01:30:48.630
would create the website. It created a great

01:30:48.630 --> 01:30:53.560
website. In the matter of 15 minutes after three

01:30:53.560 --> 01:30:57.220
back and forth prompts. The amount of effort

01:30:57.220 --> 01:31:01.680
that I used to create our old website. Was. A

01:31:01.680 --> 01:31:07.319
lot. Definitely more than 15 minutes. And almost

01:31:07.319 --> 01:31:11.460
zero brain cells. I think it took me like a week

01:31:11.460 --> 01:31:16.460
or so. Right. And it wasn't as nearly as good

01:31:16.460 --> 01:31:21.130
as this one. In terms of. the appearance right

01:31:21.130 --> 01:31:23.810
because i had to i had to generate my own images

01:31:23.810 --> 01:31:28.930
and whatnot so uh so the amount of time if you

01:31:28.930 --> 01:31:32.270
okay i don't know i'm like my hourly wage right

01:31:32.270 --> 01:31:36.350
now so the amount of hourly if i were to pay

01:31:36.350 --> 01:31:38.670
myself let's say the minimum hourly wage right

01:31:38.670 --> 01:31:44.729
now i would pay at least 40 50 but i would pay

01:31:44.729 --> 01:31:46.869
up to like you know a thousand dollars for this

01:31:46.869 --> 01:31:53.569
product alone uh just a year ago so i think there

01:31:53.569 --> 01:31:57.930
is so much potential in charging your consumers

01:31:57.930 --> 01:32:02.710
and most of them are prosumers who uh who have

01:32:02.710 --> 01:32:06.409
this high willingness to pay so um i think i'm

01:32:06.409 --> 01:32:09.369
jumping ahead a little bit in that i think 125

01:32:09.369 --> 01:32:13.130
million arr is nothing compared to manis's potential

01:32:13.130 --> 01:32:20.039
which directs me toward this Probably inaccurate

01:32:20.039 --> 01:32:23.020
hypothesis that they might have sold themselves

01:32:23.020 --> 01:32:28.579
short with $2 billion. But there's other things

01:32:28.579 --> 01:32:32.000
at play in terms of the grand landscape of AI

01:32:32.000 --> 01:32:37.579
agents because it's so proliferated that maybe

01:32:37.579 --> 01:32:40.600
they see that next year the valuation won't be

01:32:40.600 --> 01:32:43.880
as high for your generic AI agents. Maybe there's

01:32:43.880 --> 01:32:47.250
other entrants. Totally possible. Because building

01:32:47.250 --> 01:32:51.109
an OS, right, and creating an environment for

01:32:51.109 --> 01:32:55.329
your agent to operate, it's not as hard as building,

01:32:55.390 --> 01:32:59.890
let's say, an LLM model. So the barrier of entry

01:32:59.890 --> 01:33:03.409
is lower. And as much as I think that this group

01:33:03.409 --> 01:33:06.789
has some technical geniuses and product geniuses,

01:33:06.789 --> 01:33:12.170
the amount of effort that's needed for copy and

01:33:12.170 --> 01:33:14.569
pasting manners is actually much lower than...

01:33:15.260 --> 01:33:18.819
what you expect a similar product to come out

01:33:18.819 --> 01:33:21.819
a few years ago without the proliferation of

01:33:21.819 --> 01:33:25.699
AI. So maybe they see that there is a roadblock

01:33:25.699 --> 01:33:32.000
to grow from 125 million ARR to 500. But going

01:33:32.000 --> 01:33:36.500
from zero to 125 took them nine months. And they've

01:33:36.500 --> 01:33:38.279
been consistently, and they've been quoting a

01:33:38.279 --> 01:33:43.539
12, 20. 20 % month over month increase in ARR.

01:33:43.819 --> 01:33:47.319
So hitting, I'm just imagining if they really,

01:33:47.399 --> 01:33:49.579
you know, do the promotions well and everything,

01:33:49.739 --> 01:33:55.399
do the marketing plus improve your product, hitting

01:33:55.399 --> 01:34:00.180
a 500 million ARR in maybe one or two years isn't

01:34:00.180 --> 01:34:03.800
unfathomable, right? Given that they managed

01:34:03.800 --> 01:34:06.720
to produce something that would cost me. maybe

01:34:06.720 --> 01:34:09.939
$4 ,000, $5 ,000 to hire a full stack developer

01:34:09.939 --> 01:34:14.199
to do. And when you have a $500 ARR product,

01:34:14.520 --> 01:34:18.699
even with a much lower revenue to, like a revenue

01:34:18.699 --> 01:34:23.560
multiplier of, let's say, 10 rather than 16 to

01:34:23.560 --> 01:34:28.539
20, it's still a $5 billion company. Absolutely.

01:34:28.720 --> 01:34:33.039
Right? The other side of me, because we did this

01:34:33.039 --> 01:34:37.039
background research on the team's entire idealism

01:34:37.039 --> 01:34:41.199
and pragmatism, a part of me wants to say that

01:34:41.199 --> 01:34:44.579
maybe they see the future. So they know when

01:34:44.579 --> 01:34:46.680
to quit. We know that they know when to quit.

01:34:46.779 --> 01:34:50.380
Maybe they see that 2025 is the year and maybe

01:34:50.380 --> 01:34:55.439
the only year for such an agent to create such

01:34:55.439 --> 01:34:59.880
high multiples and a good exit. Or maybe they

01:34:59.880 --> 01:35:02.859
just are putting money on the table. They don't

01:35:02.859 --> 01:35:06.720
care if they can grow manas to a $5 billion valuation.

01:35:07.260 --> 01:35:10.020
They just want their foot in the door to meta

01:35:10.020 --> 01:35:15.159
so that they can get the resources of meta ASAP

01:35:15.159 --> 01:35:19.060
so they can do more things with the amount of

01:35:19.060 --> 01:35:24.340
resources. I think we both have referenced one.

01:35:25.050 --> 01:35:28.090
interview of Peak. And in that interview, I think

01:35:28.090 --> 01:35:31.850
Peak said it very clearly. It was just throughout

01:35:31.850 --> 01:35:37.409
the message that to be able to compete at today's

01:35:37.409 --> 01:35:41.789
AI landscape, we're just a bigger size developer,

01:35:42.189 --> 01:35:45.470
independent developer. And to be able to compete

01:35:45.470 --> 01:35:47.989
better in the future, you must become a big tech

01:35:47.989 --> 01:35:53.810
company. Because it just requires so much, whether

01:35:53.810 --> 01:35:58.210
it's computing like the model the money the capex

01:35:58.210 --> 01:36:03.930
yes i think that interview was done before meta's

01:36:03.930 --> 01:36:07.090
acquisition so i think maybe in their mind uh

01:36:07.090 --> 01:36:10.649
there was that direction of we're either going

01:36:10.649 --> 01:36:14.250
to be working very closely with a big tech company

01:36:14.250 --> 01:36:17.949
or we're going to become a big tech company yeah

01:36:17.949 --> 01:36:21.340
but still i mean It's a lot of money on the table.

01:36:21.539 --> 01:36:24.680
If you have a firm belief that you can grow your

01:36:24.680 --> 01:36:28.239
ARR to 500 million, and if you only take 2 million

01:36:28.239 --> 01:36:33.000
for the deal, there's something else at play,

01:36:33.100 --> 01:36:37.500
right? That we don't know about. We can only

01:36:37.500 --> 01:36:42.020
make our hypothesis. But for the users, for us,

01:36:42.159 --> 01:36:46.319
I think it's great that Meta acquired Manos because,

01:36:46.539 --> 01:36:51.619
and I'm, On Meta's perspective, right? Why would

01:36:51.619 --> 01:36:53.880
Meta acquire Manos? I think there's a few things

01:36:53.880 --> 01:36:58.420
that folks quoted. One being that Lama 4 was

01:36:58.420 --> 01:37:02.399
a flop. And they do need to use acquisition or

01:37:02.399 --> 01:37:04.880
other means to quickly get into this game because

01:37:04.880 --> 01:37:08.659
this game is a war right now. And obviously there

01:37:08.659 --> 01:37:10.920
are other companies within the Mach 7 that are

01:37:10.920 --> 01:37:13.899
winning. And so if Meta still wants to be in

01:37:13.899 --> 01:37:16.319
the game, they need to do this through acquisition

01:37:16.319 --> 01:37:19.979
because that's the fastest way of entering the

01:37:19.979 --> 01:37:23.039
arena. And so not only did they acquire Manos,

01:37:23.060 --> 01:37:28.180
they acquired a 49 % stake of scale AI. So that

01:37:28.180 --> 01:37:30.779
all speaks to their determination. The other

01:37:30.779 --> 01:37:32.760
thing that was quoted was, what are they going

01:37:32.760 --> 01:37:35.880
to do to generate that $2 million or more back?

01:37:36.279 --> 01:37:39.220
And I did a little bit of number crunching. So

01:37:39.220 --> 01:37:42.539
one of the things they quoted was to apply AI

01:37:42.539 --> 01:37:49.279
agents in WhatsApp. So WhatsApp, did you know

01:37:49.279 --> 01:37:51.739
WhatsApp is one of the most under -monetized

01:37:51.739 --> 01:37:56.210
apps out there? Really? So let me give you some

01:37:56.210 --> 01:38:02.489
figures. Go. So we all know Meta has three main

01:38:02.489 --> 01:38:05.170
products. You have Facebook and Instagram and

01:38:05.170 --> 01:38:09.069
WhatsApp. Instagram, they have a $70 billion

01:38:09.069 --> 01:38:12.850
revenue. Roughly, let's give or take $70 billion

01:38:12.850 --> 01:38:18.729
revenue with about 2 to 3 billion users. Primarily

01:38:18.729 --> 01:38:23.699
advertising? Advertising, yes. More than 95 %

01:38:23.699 --> 01:38:27.560
advertising. For the same amount of users that's

01:38:27.560 --> 01:38:34.180
on WhatsApp, $1 .8 billion versus $70 billion.

01:38:35.340 --> 01:38:38.039
Because it's hard to, with Instagram, you can

01:38:38.039 --> 01:38:40.039
have reels, you can have stories, and you can

01:38:40.039 --> 01:38:43.279
just put your ads everywhere. With a chat messaging

01:38:43.279 --> 01:38:48.300
app, people are averse to ads. And so they'd

01:38:48.300 --> 01:38:51.340
never, or the product never kind of proliferated

01:38:51.340 --> 01:38:54.680
as they only started ads in 2018. And that's

01:38:54.680 --> 01:38:58.260
how they got to the 1 .8. But what they've been

01:38:58.260 --> 01:39:01.539
building in WhatsApp is what WeChat tried to

01:39:01.539 --> 01:39:04.840
do in 2020, which is the business side of things.

01:39:05.420 --> 01:39:07.880
Charging your business customers. So WeChat,

01:39:08.079 --> 01:39:12.439
sorry, WhatsApp now has WhatsApp Pay. It has

01:39:12.439 --> 01:39:15.789
WhatsApp for Business. which gives you like the

01:39:15.789 --> 01:39:19.770
group messaging functionalities. And so like

01:39:19.770 --> 01:39:23.369
a company like Delta Airlines can use WhatsApp

01:39:23.369 --> 01:39:25.550
for business and they will be charged for conversation.

01:39:25.949 --> 01:39:30.130
Oh, that makes so much sense. That's so agentic

01:39:30.130 --> 01:39:34.470
integrated. Exactly. Because it's customer service

01:39:34.470 --> 01:39:37.350
in the US. You don't know how like in China,

01:39:37.390 --> 01:39:40.909
right? How often do you receive AI generated

01:39:40.909 --> 01:39:44.390
message from Taobao right now? Like. like like

01:39:44.390 --> 01:39:47.630
a million a day i don't know like hey your promotion's

01:39:47.630 --> 01:39:50.229
about to start hi can you give me like a five

01:39:50.229 --> 01:39:55.409
-star rating hi do you like me so so it's just

01:39:55.409 --> 01:39:58.909
so integrated and merchants do pay even in china

01:39:58.909 --> 01:40:01.550
merchants pay a few thousand yuan for these for

01:40:01.550 --> 01:40:04.350
a month per month like the the bigger merchants

01:40:04.350 --> 01:40:10.109
um so there's huge leeway a huge potential in

01:40:10.109 --> 01:40:15.060
this area that whatsapp has the $2 billion, sorry,

01:40:15.199 --> 01:40:19.960
the 2 billion or so, 3 billion users, and almost

01:40:19.960 --> 01:40:24.500
little to no monetization. Jeez, I cannot imagine

01:40:24.500 --> 01:40:26.859
a day where my WhatsApp's going to be proliferated

01:40:26.859 --> 01:40:30.319
with SMBs asking me, oh, do you want to buy this?

01:40:30.340 --> 01:40:33.180
Do you want to buy that? But isn't it a natural...

01:40:33.180 --> 01:40:36.819
It's reasonable. Right? And I think it's actually,

01:40:36.920 --> 01:40:40.909
maybe it's not the large... but the small mom

01:40:40.909 --> 01:40:43.750
and pop shops actually relies on this Facebook

01:40:43.750 --> 01:40:47.149
and WhatsApp ecosystem to survive. And if they

01:40:47.149 --> 01:40:51.010
manage to do what WeChat did for China or what

01:40:51.010 --> 01:40:54.710
Line did for Southeast Asia and Japan, imagine

01:40:54.710 --> 01:41:00.189
the profitability potentials. And so Mr. Zuckerberg,

01:41:00.229 --> 01:41:05.939
he famously purchased Instagram, right? for $1

01:41:05.939 --> 01:41:09.279
billion, 13 employees, and zero revenue. And

01:41:09.279 --> 01:41:11.640
look at where it is now. I know. So with the

01:41:11.640 --> 01:41:17.180
right resources and trajectory, the $2 billion

01:41:17.180 --> 01:41:21.680
that WhatsApp spent on Manos is nothing. So every

01:41:21.680 --> 01:41:25.119
two weeks, they have a net profit of this much.

01:41:25.300 --> 01:41:27.399
So every two weeks, they can purchase a Manos.

01:41:28.270 --> 01:41:32.149
So for them, it's peanuts. Just tossing... For

01:41:32.149 --> 01:41:34.949
us, it's like, whoa, $2 billion. You're throwing

01:41:34.949 --> 01:41:37.750
away money if you grow your ARR by $500 million.

01:41:37.869 --> 01:41:40.449
But for them, it's like... I can pay this. I

01:41:40.449 --> 01:41:42.909
can afford this. Yeah, like even if it flops,

01:41:42.949 --> 01:41:45.989
right? But the potential, the upside potential

01:41:45.989 --> 01:41:49.300
of getting WhatsApp for business right. A lot

01:41:49.300 --> 01:41:54.739
of imagination. Wow. Goodbye. Goodbye. But the

01:41:54.739 --> 01:41:56.899
downside is, how are you going to integrate this

01:41:56.899 --> 01:41:59.000
Chinese team who's now based in Singapore and

01:41:59.000 --> 01:42:02.460
with the two countries at geopolitical tension

01:42:02.460 --> 01:42:04.579
with each other? We will continue to monitor

01:42:04.579 --> 01:42:07.859
the two teams. Well, actually, this episode is

01:42:07.859 --> 01:42:09.960
recorded the end of January. It will likely come

01:42:09.960 --> 01:42:11.800
out at the end of this month. So we actually

01:42:11.800 --> 01:42:13.880
don't know what's going to happen with the purchase.

01:42:13.880 --> 01:42:16.439
Is it actually going to go through? Exactly.

01:42:17.140 --> 01:42:21.699
So a lot of... room for um imagination should

01:42:21.699 --> 01:42:23.899
we give some context around this so there was

01:42:23.899 --> 01:42:30.239
a first uh in uh may of last year when the usvc

01:42:30.239 --> 01:42:33.180
and we i think we skipped this where a usvc called

01:42:33.180 --> 01:42:36.300
benchmark put in 500 million dollars for manos

01:42:36.300 --> 01:42:41.829
um this triggered a U .S. investigation, U .S.

01:42:41.829 --> 01:42:44.850
side investigation, because there was an executive

01:42:44.850 --> 01:42:49.189
order in, I think, 2019, 20 period that actually

01:42:49.189 --> 01:42:54.010
would not allow any U .S. company to invest in

01:42:54.010 --> 01:42:58.270
Chinese startups that are in the AI ecosystem

01:42:58.270 --> 01:43:00.529
or a lot of ecosystem, including AI, because

01:43:00.529 --> 01:43:04.350
you don't want U .S. money to fund AI develop

01:43:04.350 --> 01:43:08.770
in China for obvious reasons. And so this benchmark

01:43:08.770 --> 01:43:11.310
investment of 500 million in a Beijing -based

01:43:11.310 --> 01:43:15.270
company triggered something of an investigation,

01:43:15.649 --> 01:43:20.420
right? What they've done is... Move this whole

01:43:20.420 --> 01:43:22.399
team. They actually fired a lot of people in

01:43:22.399 --> 01:43:25.220
Beijing. Move this whole team to Singapore to

01:43:25.220 --> 01:43:28.699
Singapore wash the team. I don't know. There's

01:43:28.699 --> 01:43:30.720
other terms out there, but like. I like your

01:43:30.720 --> 01:43:35.140
term. It is Singapore wash. Right. So now the

01:43:35.140 --> 01:43:39.180
team is established in Singapore. Everyone moved

01:43:39.180 --> 01:43:42.739
to Singapore. You have zero data retained in

01:43:42.739 --> 01:43:47.359
China. And so U .S. government, as a result,

01:43:47.399 --> 01:43:51.680
stopped investigations. right until december

01:43:51.680 --> 01:43:55.000
when this is announced china comes back and like

01:43:55.000 --> 01:43:59.279
hey wait a second okay us is okay but we're not

01:43:59.279 --> 01:44:02.460
okay we're not okay with it you developed the

01:44:02.460 --> 01:44:06.659
technology in beijing and now you're benefiting

01:44:06.659 --> 01:44:12.279
non -chinese users Like, what is this? You're

01:44:12.279 --> 01:44:14.960
to claim it, Reds. Right. And so the Chinese

01:44:14.960 --> 01:44:17.840
Department of Commerce then sets up another investigation,

01:44:18.119 --> 01:44:21.159
which is happening right now. And there is a

01:44:21.159 --> 01:44:24.619
potential that they'll block the sale. But, and

01:44:24.619 --> 01:44:26.699
this is, again, my personal opinion, because

01:44:26.699 --> 01:44:29.359
a lot of the recent acquisitions in Silicon Valley

01:44:29.359 --> 01:44:32.340
are acquihires. So it's not really, they built

01:44:32.340 --> 01:44:34.359
the product in three and a half months, right?

01:44:34.579 --> 01:44:38.229
Worst case scenario, everything. Goes back to

01:44:38.229 --> 01:44:40.310
scratch. I don't think that's going to happen,

01:44:40.369 --> 01:44:42.310
but it's just... Worst case scenario. If you

01:44:42.310 --> 01:44:44.609
moved everything already, your databases and

01:44:44.609 --> 01:44:48.989
everything already outside of China, your personnels

01:44:48.989 --> 01:44:53.630
are outside of China, there's a lot of legal

01:44:53.630 --> 01:44:57.909
ways to work around this to enable Meta to still

01:44:57.909 --> 01:45:02.489
acquire what they want, even if it's not Manus

01:45:02.489 --> 01:45:05.880
as Manus today. They can acquire monosystem technology.

01:45:06.000 --> 01:45:08.619
They can license whatever, right? So there are

01:45:08.619 --> 01:45:12.600
ways around it that will ultimately enable this

01:45:12.600 --> 01:45:16.220
whole technology to be under meta. It's a one

01:45:16.220 --> 01:45:19.020
plus one equals a hundred story. And all the

01:45:19.020 --> 01:45:21.899
users out there in the world will want this to,

01:45:22.000 --> 01:45:24.220
well, I mean, at least I would want this to happen

01:45:24.220 --> 01:45:27.810
because I've directly benefited. From using manners.

01:45:28.229 --> 01:45:32.409
Yeah. Yeah. And I think Liu Yuan also shared

01:45:32.409 --> 01:45:37.250
his fair comments to the future of manners. He

01:45:37.250 --> 01:45:40.210
said, we don't worry about things that will take

01:45:40.210 --> 01:45:42.569
place tomorrow. If it's geopolitics, because

01:45:42.569 --> 01:45:44.949
anything can happen tomorrow. Like AI development

01:45:44.949 --> 01:45:47.649
will happen very fast. But the one thing that

01:45:47.649 --> 01:45:51.779
we're for sure. will know is like all of those

01:45:51.779 --> 01:45:54.560
people who are working on the product will just

01:45:54.560 --> 01:45:56.859
continue to work on these things they will never

01:45:56.859 --> 01:45:59.340
stop working on this kind of thing why do we

01:45:59.340 --> 01:46:02.699
think it has ended why do we think it's on our

01:46:02.699 --> 01:46:05.439
list i feel like with the combination of these

01:46:05.439 --> 01:46:08.819
three people and the amount of project they worked

01:46:08.819 --> 01:46:13.460
on and how much they are willing to throw things

01:46:13.460 --> 01:46:16.899
in the trash bin even if they didn't develop

01:46:16.899 --> 01:46:20.000
manners they probably have will end up with something

01:46:20.000 --> 01:46:25.659
else yeah just my take um also i think there's

01:46:25.659 --> 01:46:29.340
a there's a big learning with all of red's ventures

01:46:29.340 --> 01:46:32.800
rely on the platform like don't go against it

01:46:32.800 --> 01:46:36.279
right with wechat they created iban and weban

01:46:36.279 --> 01:46:40.000
and they kind of caught the the flow of wechat

01:46:40.000 --> 01:46:45.380
going from personal to business uh assistance

01:46:45.380 --> 01:46:49.880
uh with monica it was kind of they were leveraging

01:46:49.880 --> 01:46:54.100
on Chrome as the platform. With Manos, they're

01:46:54.100 --> 01:46:58.399
leveraging on all the combination of LLMs out

01:46:58.399 --> 01:47:01.279
there rather than competing against them. So

01:47:01.279 --> 01:47:04.739
having your product set on an existing platform

01:47:04.739 --> 01:47:09.279
and not, it's less risky. And the other thing

01:47:09.279 --> 01:47:12.079
I think really helped them is that the whole

01:47:12.079 --> 01:47:16.380
scrum process. Again, in one of the interviews,

01:47:16.520 --> 01:47:20.180
they mentioned that with three co -founders,

01:47:20.199 --> 01:47:23.300
they actually have like six people with the two

01:47:23.300 --> 01:47:27.619
other previous co -founders that worked on the

01:47:27.619 --> 01:47:31.380
Monaco project, right? The decision is, you would

01:47:31.380 --> 01:47:34.640
imagine when there's a clash in terms of decisions,

01:47:34.920 --> 01:47:38.319
it would be difficult because it's not a one

01:47:38.319 --> 01:47:41.859
-person decision. But in the end, for them, actually,

01:47:41.880 --> 01:47:43.979
they make all the decisions very fast because

01:47:43.979 --> 01:47:46.920
they don't care if it's the wrong decision because

01:47:46.920 --> 01:47:51.880
the amount of time that's needed to correct a

01:47:51.880 --> 01:47:55.640
decision nowadays is so low. So why don't we

01:47:55.640 --> 01:47:58.460
just, if we disagree, fine, let's go with one,

01:47:58.479 --> 01:48:01.020
like make the decision, we'll go with one direction.

01:48:02.140 --> 01:48:04.640
Very quickly, in two weeks, three weeks, our

01:48:04.640 --> 01:48:06.939
customers, our users will tell us if it's the

01:48:06.939 --> 01:48:10.039
right decision. So let your users, because the

01:48:10.039 --> 01:48:12.760
development process is very condensed nowadays,

01:48:13.119 --> 01:48:16.539
you can actually let your users tell you that,

01:48:16.579 --> 01:48:22.060
Manos, I like using you for PPT. I like using

01:48:22.060 --> 01:48:24.939
you for website creation. I'm okay with using

01:48:24.939 --> 01:48:28.020
you for other things, right? So I think that's

01:48:28.020 --> 01:48:31.760
how they managed to really... get their product

01:48:31.760 --> 01:48:35.560
out so soon and make pivots very quickly. It's

01:48:35.560 --> 01:48:38.500
a very democratic way of product management.

01:48:38.760 --> 01:48:41.520
You just let the data tell you. Right, and you

01:48:41.520 --> 01:48:44.659
have one decision maker, but you don't fix that.

01:48:45.079 --> 01:48:47.800
You let the product then tell you what happens

01:48:47.800 --> 01:48:50.699
with the decision. For sure. They made it to

01:48:50.699 --> 01:48:52.800
our list. Well, first of all, they are requested

01:48:52.800 --> 01:48:55.920
by a lot of our friends. There are plenty of

01:48:55.920 --> 01:48:57.680
people who drop you a message and say, you must

01:48:57.680 --> 01:49:00.300
get on Manos. I'm like... OK, let's do it. Yeah.

01:49:00.460 --> 01:49:03.640
And in fact, Lena and I, we had a moment where

01:49:03.640 --> 01:49:08.380
we were about to record and I last minute pinged

01:49:08.380 --> 01:49:10.859
her and said, hey, I can't do this this afternoon.

01:49:10.979 --> 01:49:13.930
Can you give me two more days? Because. I completely

01:49:13.930 --> 01:49:16.470
underestimated how much information it was out

01:49:16.470 --> 01:49:19.010
there, even though it's a very young company.

01:49:19.409 --> 01:49:22.010
But I don't know why. They're all so talkative.

01:49:22.050 --> 01:49:24.970
That's why. They're everywhere. There's a lot

01:49:24.970 --> 01:49:26.649
of information. There's a lot of information.

01:49:27.050 --> 01:49:29.390
Compared to... What was the one that we had a

01:49:29.390 --> 01:49:32.609
lot of trouble on? Deep Seek. Oh, deep -seek.

01:49:32.609 --> 01:49:34.689
Yeah. Because no one ever makes an appearance

01:49:34.689 --> 01:49:36.890
anywhere. Yeah, exactly. He doesn't do any talks.

01:49:37.149 --> 01:49:39.310
Yeah, whereas this one is just like overwhelming

01:49:39.310 --> 01:49:42.710
amount of information. Which is a great thing

01:49:42.710 --> 01:49:46.350
for doing research. But I also grant you that

01:49:46.350 --> 01:49:49.890
the research done for this episode is completely

01:49:49.890 --> 01:49:52.489
human -driven. There's nothing AI about it because

01:49:52.489 --> 01:49:55.869
I have to go through everything. I like them.

01:49:56.900 --> 01:50:00.000
Because I think there was some sort of a consensus

01:50:00.000 --> 01:50:03.380
on day one when they started the company, Butterfly

01:50:03.380 --> 01:50:06.260
Effect, that we're not going to create another

01:50:06.260 --> 01:50:09.619
LLM. We're just going to go for products. So

01:50:09.619 --> 01:50:13.600
I think that was quite clear. Nobody was going

01:50:13.600 --> 01:50:15.880
for reinventing the wheels. They just wanted

01:50:15.880 --> 01:50:17.520
to build a really great product that people will

01:50:17.520 --> 01:50:20.319
use. Going back to the tool geek kind of mindset.

01:50:21.130 --> 01:50:24.470
The other thing is, it's like Zhen Fan up until

01:50:24.470 --> 01:50:26.689
today, neither confirms or denies that Manus

01:50:26.689 --> 01:50:29.529
has already made it because they think that it's

01:50:29.529 --> 01:50:33.189
nice to be the incumbent for a bit, but the hype

01:50:33.189 --> 01:50:35.750
changes very fast. So they don't know, which

01:50:35.750 --> 01:50:37.829
I think is a very modest way of looking at that.

01:50:37.930 --> 01:50:39.869
And I think it's fair. And I think they deserve

01:50:39.869 --> 01:50:42.590
some spotlight. Let's just be honest. Of course.

01:50:42.750 --> 01:50:47.170
I think like, I mean, just to put it plainly,

01:50:47.189 --> 01:50:49.430
the monetary reward itself has been great, right,

01:50:49.470 --> 01:50:52.579
for them. I think in that perspective, they've

01:50:52.579 --> 01:50:55.520
already made it. I think what we want or what

01:50:55.520 --> 01:50:58.979
everyone wants to see is what would happen post,

01:50:59.039 --> 01:51:02.739
once you have the meta resources, can you make

01:51:02.739 --> 01:51:08.619
this into an epic? Exactly. Is it, prove the

01:51:08.619 --> 01:51:11.220
dissenters wrong that this is just a wrapper

01:51:11.220 --> 01:51:13.699
without a future, right? Prove to the dissenters

01:51:13.699 --> 01:51:18.199
that with the support of meta this is going to

01:51:18.199 --> 01:51:21.439
be something that can generate 70 million dollars

01:51:21.439 --> 01:51:25.819
for whatsapp make whatsapp instagram um yeah

01:51:25.819 --> 01:51:29.699
scary one thing i do want to last last thought

01:51:29.699 --> 01:51:34.680
i promise where is amazon in all of this why

01:51:34.680 --> 01:51:39.199
why didn't amazon purchase because hear me out

01:51:39.199 --> 01:51:43.319
the virtual machines they're amazon The virtual

01:51:43.319 --> 01:51:45.579
machines, the 80 million virtual machines that

01:51:45.579 --> 01:51:50.220
Manos created, the vendor is Amazon. Oh my goodness.

01:51:50.420 --> 01:51:54.340
The company that is most like Taobao that needs

01:51:54.340 --> 01:52:01.800
agents is Amazon. Logically, yes. Right? So naturally...

01:52:01.800 --> 01:52:04.460
Customer service. I would think Amazon would

01:52:04.460 --> 01:52:07.840
be competing. And maybe that's why Zuckerberg...

01:52:07.840 --> 01:52:13.740
Have to buy it fast. Got it in 10 days. Wow.

01:52:13.740 --> 01:52:15.899
Fair guess. We'll see what it is. Yeah. Yeah.

01:52:16.020 --> 01:52:20.460
Guess of the year. Very cool. Okay. Yeah. When

01:52:20.460 --> 01:52:24.140
we shut off the mic, you can tell me more. This

01:52:24.140 --> 01:52:27.699
whole story at the end of the whole research

01:52:27.699 --> 01:52:32.079
gives me a feeling that it takes a village to

01:52:32.079 --> 01:52:35.939
raise a kid. Don't you think? And they can't

01:52:35.939 --> 01:52:39.939
be modest. Yeah. It's such a, like, with the

01:52:39.939 --> 01:52:44.539
founders, with the funds. And these group of

01:52:44.539 --> 01:52:46.500
people have been working on the project like

01:52:46.500 --> 01:52:50.260
that for about a decade. And so I'm very glad

01:52:50.260 --> 01:52:53.170
that the product. was able to graduate and now

01:52:53.170 --> 01:52:56.489
going out to the world. And obviously, I will

01:52:56.489 --> 01:52:59.270
not finish this episode without Manus writing

01:52:59.270 --> 01:53:02.670
something for the Manus episode. So I did give

01:53:02.670 --> 01:53:04.590
a little prompt. What did it say? Hear me out.

01:53:05.189 --> 01:53:09.810
And so the story closes where it began. Mammoth

01:53:09.810 --> 01:53:14.090
to Maggie, Monica to Manus, Manus to Meta, different

01:53:14.090 --> 01:53:17.789
names, same instinct. The quiet belief that tools

01:53:17.789 --> 01:53:22.470
can think with us, not just for us. Call it evolution,

01:53:22.829 --> 01:53:26.989
coincidence or fate. The spirit of M kept shipping.

01:53:29.010 --> 01:53:32.750
That's it. It is a coincidence that they all

01:53:32.750 --> 01:53:36.970
started with M. Why it all started with M, I

01:53:36.970 --> 01:53:39.949
don't know. But best of luck to Manus. Best of

01:53:39.949 --> 01:53:43.250
luck to Manus. I'm looking forward to more developments

01:53:43.250 --> 01:53:46.850
in that area. All right. Thank you for listening.

01:53:47.189 --> 01:53:50.479
And if you like this episode, please... Comment

01:53:50.479 --> 01:53:54.100
or rate us on Apple and Spotify. We'll see you

01:53:54.100 --> 01:53:55.279
in the next episode. Bye.