WEBVTT

00:00:00.000 --> 00:00:04.600
Hello, today we're chatting with Qiu Yuan, Jin Qiuyuan

00:00:06.200 --> 00:00:07.800
Sandy, you graduated from Yale

00:00:07.900 --> 00:00:09.600
Yes, yes, Yale undergrad

00:00:10.200 --> 00:00:10.900
And then

00:00:12.000 --> 00:00:14.600
Right, so what was your career path afterward

00:00:15.700 --> 00:00:17.000
Yes, I'm a serial entrepreneur

00:00:17.000 --> 00:00:18.700
So after graduating from Yale, what did you do

00:00:18.700 --> 00:00:19.700
I first worked in consulting

00:00:19.700 --> 00:00:21.300
Then later I transitioned into product management

00:00:21.600 --> 00:00:24.900
Before starting my own business, I worked as a product manager at Tantan

00:00:25.000 --> 00:00:26.300
Responsible for growth and monetization

00:00:26.800 --> 00:00:29.300
After Tantan was acquired, I started my own business independently

00:00:29.500 --> 00:00:31.900
I built an online events platform

00:00:31.900 --> 00:00:33.100
Called VLAN

00:00:33.600 --> 00:00:35.700
I previously raised two rounds of funding

00:00:35.900 --> 00:00:36.700
From Shouyuan and Sequoia

00:00:37.300 --> 00:00:38.400
So for this platform of ours

00:00:38.400 --> 00:00:40.500
We mainly serve online enterprises

00:00:40.700 --> 00:00:42.500
Hosting virtual events

00:00:42.900 --> 00:00:46.700
Everything from online recruiting to corporate open days

00:00:47.100 --> 00:00:50.100
Yes, we ran this product for a bit more than four years

00:00:50.500 --> 00:00:51.800
We went through the whole cycle

00:00:52.100 --> 00:00:54.700
During the pandemic we had a lot of

00:00:55.000 --> 00:00:55.800
Momentum

00:00:55.900 --> 00:00:58.200
After the pandemic, many events returned offline

00:00:58.400 --> 00:01:00.400
So last year I paused the company

00:01:00.400 --> 00:01:02.200
And this year I'm exploring new directions

00:01:03.700 --> 00:01:05.400
We've been WeChat friends for many years now

00:01:05.800 --> 00:01:07.700
We were internet friends

00:01:07.700 --> 00:01:09.900
Before we were online friends, now we're friends in real life

00:01:11.200 --> 00:01:12.400
Back then it was at Yale

00:01:12.400 --> 00:01:14.500
What did you study, and how did you end up coming back to start a business

00:01:15.600 --> 00:01:17.800
Right, I studied economics as an undergrad

00:01:17.800 --> 00:01:22.400
At that time, there was a dual entrepreneurship and innovation environment in China

00:01:22.700 --> 00:01:24.100
So back then it was quite emphasized

00:01:24.400 --> 00:01:25.900
Just trying out the path of entrepreneurship

00:01:26.300 --> 00:01:28.600
So after graduation I went back to China to look at various opportunities

00:01:29.600 --> 00:01:31.900
So how long did you work at Tantan

00:01:32.300 --> 00:01:35.000
I was at Tantan around 2018 to 2019

00:01:35.800 --> 00:01:37.500
So it happened right when the pandemic hit

00:01:38.100 --> 00:01:40.600
虚拟社交赛道爆发 -> The virtual social track is booming

00:01:40.700 --> 00:01:41.900
Empecé a trabajar en lo social virtual

00:01:42.600 --> 00:01:46.000
En realidad, siempre he querido dedicarme a conectar a las personas

00:01:46.300 --> 00:01:49.100
Sí, así que en toda mi carrera como product manager

00:01:49.100 --> 00:01:50.000
Básicamente se trataba de lo social

00:01:50.100 --> 00:01:52.700
Luego más adelante trabajé en una plataforma de eventos del metaverso

00:01:52.800 --> 00:01:56.800
También era para, de manera online, dar a los organizadores y participantes de eventos

00:01:57.100 --> 00:01:59.800
la posibilidad de crear una conexión y un entorno social inmersivo

00:02:00.600 --> 00:02:02.500
¿De qué parte te encargabas principalmente en Tantan?

00:02:03.200 --> 00:02:05.500
En Tantan, yo estaba principalmente en el área de crecimiento y monetización

00:02:05.600 --> 00:02:08.400
En monetización básicamente viví la etapa de pasar de cero a uno

00:02:08.700 --> 00:02:12.000
Pasamos de un nivel de ingresos mensuales muy bajo

00:02:12.000 --> 00:02:14.700
a un crecimiento de ingresos de cien millones al mes

00:02:14.900 --> 00:02:17.700
Hicimos muchísimas features y muchísimas

00:02:18.000 --> 00:02:18.700
pruebas A/B

00:02:18.900 --> 00:02:22.300
Así que de ese proceso todavía me siento bastante orgulloso

00:02:22.800 --> 00:02:25.000
¿Cuál crees que fue la feature más importante en ese momento?

00:02:25.500 --> 00:02:26.400
Hubo muchas

00:02:26.700 --> 00:02:28.900
Ahora, dentro de Tantan, entre las funciones de pago, por ejemplo

00:02:29.400 --> 00:02:30.400
ver quién me dio like

00:02:30.700 --> 00:02:31.900
por ejemplo, Super Like

00:02:32.300 --> 00:02:35.400
Cada feature representaba más o menos decenas de millones de RMB en ingresos

00:02:35.900 --> 00:02:38.000
ese punto de pago a nivel de ingresos mensuales

00:02:38.600 --> 00:02:40.900
Entonces, ¿lo de Super Like fue algo que Tantan inventó primero

00:02:40.900 --> 00:02:42.600
o Tinder ya lo tenía originalmente también?

00:02:43.100 --> 00:02:45.900
En realidad también fuimos cruzando el río tanteando las piedras

00:02:46.500 --> 00:02:50.100
Nos basamos en Tinder e hicimos muchas innovaciones y ajustes

00:02:50.100 --> 00:02:52.600
Sí, porque Tinder para entonces ya era una empresa bastante grande

00:02:52.600 --> 00:02:53.900
y también tenía ingresos bastante estables

00:02:54.100 --> 00:02:58.200
así que sus features de monetización y esas funciones

00:02:58.400 --> 00:02:59.300
eran relativamente maduras

00:02:59.500 --> 00:03:02.300
Así que al principio nos basamos sobre todo en sus funciones de pago

00:03:02.500 --> 00:03:04.900
principalmente en la localización para China

00:03:04.900 --> 00:03:08.800
y en cuanto a paquetes y pagos, hicimos algunos diseños nuevos

00:03:10.200 --> 00:03:12.200
Entonces, ¿Tantan salió después de Tinder?

00:03:13.600 --> 00:03:14.500
Sí, así es

00:03:17.100 --> 00:03:18.900
Es un producto de tipo copy to China

00:03:19.500 --> 00:03:19.900
Sí

00:03:20.500 --> 00:03:22.900
En ese momento, de hecho, muchas plataformas eran copy to China

00:03:23.000 --> 00:03:28.200
Mi primer trabajo tech, en tecnología, fue en Didi

00:03:28.300 --> 00:03:30.600
当时也是超的Uber的模式 -> At the time it was actually modeled after Uber's approach

00:03:30.800 --> 00:03:35.400
当时其实也是正好把Uber中国并购之后我加入的 -> Actually, I joined right after Uber China was acquired

00:03:35.500 --> 00:03:37.900
所以当时还是很多copy to China的 -> So back then there was still a lot of 'copy to China'

00:03:38.300 --> 00:03:40.400
这样一个移动互联网的创业方式 -> That kind of mobile internet entrepreneurship approach

00:03:40.900 --> 00:03:42.700
探探会培养你一个用户模型 -> Tantan would help you develop a user model

00:03:42.900 --> 00:03:46.500
你会大概脑子里知道用户进入这个产品可能会干什么 -> You'd roughly know in your mind what a user might do once they enter the product

00:03:46.900 --> 00:03:47.600
对 -> Right

00:03:47.600 --> 00:03:49.900
就是相当于是一个验证的路径 -> It's essentially a path of validation

00:03:50.200 --> 00:03:52.200
但是从付费上可能会差一个汇率 -> But in terms of payment, there might be an exchange rate difference

00:03:53.600 --> 00:03:54.700
学会了一些用户模型 -> I learned some user models

00:03:55.200 --> 00:03:59.100
后来做探探这些用户模型 -> Later, working on Tantan with these user models

00:03:59.400 --> 00:04:05.000
你脑子里生成用户模型对你后来做元宇宙社交有帮助吗 -> Did having these user models in your mind help you later when working on metaverse social products?

00:04:05.900 --> 00:04:07.100
这个还是有帮助的 -> It did help

00:04:07.200 --> 00:04:11.600
因为其实很多的付费功能设计还是围绕着人性来展开 -> Because a lot of paid feature design really revolves around human nature

00:04:12.100 --> 00:04:14.300
比如说男女比例很不平衡 -> For example, the male-to-female ratio is very imbalanced

00:04:14.400 --> 00:04:19.900
怎么让男用户有更主动的看女用户的权利 -> How to give male users more proactive rights to view female users

00:04:20.000 --> 00:04:25.300
或者让他可以去显示自己的实力 -> Or let them showcase their strength

00:04:25.700 --> 00:04:29.100
我们其实还有很多的付费功能是围绕这个来设计的 -> We actually designed many paid features around this

00:04:30.200 --> 00:04:34.800
从我们后面创业做线上的使女活动这块角度来说 -> From the perspective of our later venture into online social activities

00:04:35.100 --> 00:04:38.600
其实也是利用了人性中的一个需求 -> It also tapped into a fundamental human need

00:04:38.700 --> 00:04:42.100
就是在疫情条件下 -> That is, under pandemic conditions

00:04:42.300 --> 00:04:44.500
然后很多人开了很多视频会议 -> Many people had a lot of video meetings

00:04:44.500 --> 00:04:46.100
非常的疲惫 -> They were very exhausted

00:04:46.200 --> 00:04:53.300
然后也没有这种沉浸感和互动的方式 -> And there was no sense of immersion or interactive way of engaging

00:04:53.600 --> 00:04:56.300
我们线上其实用游戏化的体验 -> We actually used a gamified experience online

00:04:56.500 --> 00:05:00.000
去还原了线下社交的这样的一个experience -> To recreate that offline social experience

00:05:00.300 --> 00:05:02.300
其实里面每个人有自己的avatar -> Actually, everyone has their own avatar in it

00:05:02.600 --> 00:05:05.400
然后这样个小人走进其他用户之后 -> And when this little character walks up to other users

00:05:05.700 --> 00:05:07.700
会自动产生音视频的一个链接 -> An audio and video connection is automatically established

00:05:08.100 --> 00:05:11.300
从而你会觉得这个人好像和我就在同一个空间 -> So you'd feel as if this person is in the same space as you

00:05:11.700 --> 00:05:13.600
产生着这个社交的体验 -> Creating that social experience

00:05:13.900 --> 00:05:17.400
所以这个其实也是围绕着一个人的需求 -> So this is also built around a fundamental human need

00:05:17.800 --> 00:05:18.800
去展开产品设计 -> To develop the product design

00:05:19.400 --> 00:05:20.500
所以它是线上一对一 -> So is it one-on-one online

00:05:21.200 --> 00:05:22.200
是一对多的 -> It's one-to-many

00:05:22.400 --> 00:05:24.300
它主要就体现了一个空间感 -> It mainly conveys a sense of space

00:05:24.500 --> 00:05:27.600
比如说我们现在在一个大堂的环境 -> For example, we're currently in a lobby environment

00:05:27.700 --> 00:05:29.500
那里面有不同的桌子对吧 -> There are different tables in there, right

00:05:29.600 --> 00:05:31.100
那边在讨论其他的一个话题 -> Over there they're discussing another topic

00:05:31.100 --> 00:05:33.600
然后我们这边在做一个录制的视频 -> And over here we're recording a video

00:05:33.800 --> 00:05:35.800
Luego, por ese lado, puede que también estén organizando un seminario

00:05:36.000 --> 00:05:39.600
En realidad, es el mismo espacio, pero te ofrece una experiencia inmersiva

00:05:40.300 --> 00:05:42.200
Esto en realidad lo logramos en línea a través de

00:05:42.800 --> 00:05:45.100
un espacio así llamado gamificado

00:05:45.700 --> 00:05:48.300
y un entorno inmersivo para recrear y reproducir la experiencia para ti

00:05:48.700 --> 00:05:51.800
De esta manera, tu evento de lanzamiento se siente como si se estuviera realizando en un lugar real

00:05:52.200 --> 00:05:55.300
Tu feria de empleo en línea se siente como poner un puesto en la escuela

00:05:55.500 --> 00:05:59.600
y realmente puedes moverte entre los diferentes booths

00:06:00.000 --> 00:06:02.300
Y después de terminar de hablar en un booth

00:06:02.300 --> 00:06:04.100
también puedes cambiar a otro canal

00:06:04.300 --> 00:06:05.100
así no se interrumpen entre sí

00:06:05.100 --> 00:06:08.500
pero al mismo tiempo tienes una experiencia bastante fluida

00:06:09.200 --> 00:06:10.800
Y esto, creo que es bastante divertido

00:06:11.300 --> 00:06:12.400
Es muy divertido

00:06:12.400 --> 00:06:14.300
Para nosotros también es un producto bastante

00:06:15.300 --> 00:06:19.100
bien recibido por estudiantes y usuarios jóvenes

00:06:19.400 --> 00:06:20.800
Porque todos decimos que la Generación Z

00:06:21.100 --> 00:06:23.500
y las generaciones futuras son nativas digitales de los videojuegos

00:06:23.500 --> 00:06:25.500
así que en realidad usamos la forma que a ellos les gusta

00:06:25.500 --> 00:06:29.800
para crear una plataforma con bastante interactividad y valor de entretenimiento

00:06:30.600 --> 00:06:32.800
Lo clave es que el mes pasado hablé con un joven de 19 años

00:06:32.800 --> 00:06:34.900
que vino de Houston

00:06:35.400 --> 00:06:38.500
él almost quería hacer una idea similar

00:06:38.500 --> 00:06:41.900
Su punto de partida fue que antes había participado en una reunión en línea

00:06:41.900 --> 00:06:42.400
Summoner

00:06:42.800 --> 00:06:44.900
y al llegar se encontró con un grupo de trescientas personas

00:06:44.900 --> 00:06:46.800
donde un gran experto le hablaba a un montón de gente

00:06:46.800 --> 00:06:47.800
Él estaba ahí escuchando una lecture

00:06:47.800 --> 00:06:48.800
como los estadounidenses, ya sabes

00:06:48.800 --> 00:06:51.300
incluyendo esto, yo también, como parte de la nueva generación, pienso que es muy molesto

00:06:51.300 --> 00:06:51.700
¿verdad?

00:06:51.700 --> 00:06:53.100
Yo llegué queriendo interactuar con la gente

00:06:53.100 --> 00:06:54.900
y terminó convirtiéndose en un sermón

00:06:54.900 --> 00:06:57.500
Él quería hacer algo en línea similar a esto

00:06:57.500 --> 00:06:58.900
y luego subir y generar un escenario

00:06:59.200 --> 00:07:01.200
para que la gente pudiera acercarse y hablar uno a uno entre sí

00:07:01.200 --> 00:07:04.800
Sí, en esa época fuimos de los primeros, creo

00:07:04.900 --> 00:07:07.700
al menos en la región de Asia fuimos de los primeros en lanzarlo

00:07:07.700 --> 00:07:07.900
Sí

00:07:07.900 --> 00:07:10.300
Así que creo que esta necesidad todavía existe

00:07:10.300 --> 00:07:13.200
Solo que ahora todos pueden llevar mucho tiempo confinados por la pandemia otra vez

00:07:13.200 --> 00:07:14.600
volver a una forma de socializar presencial

00:07:15.700 --> 00:07:18.300
porque vi que ese niño jugaba a ese Roblox

00:07:18.300 --> 00:07:20.600
realmente hay mucho espacio

00:07:20.600 --> 00:07:20.900
sí

00:07:20.900 --> 00:07:22.100
entonces, ¿qué opinas?

00:07:22.100 --> 00:07:24.400
y ese joven de 19 años incluso sentía que

00:07:24.400 --> 00:07:25.900
esta necesidad suya no estaba siendo satisfecha

00:07:25.900 --> 00:07:27.700
él piensa que esto es bastante futurista

00:07:27.700 --> 00:07:28.700
¿qué opinas al respecto?

00:07:28.700 --> 00:07:32.700
este es un problema de separación entre el usuario y quien paga

00:07:32.800 --> 00:07:35.300
es decir, nuestros clientes que pagan

00:07:35.300 --> 00:07:36.300
en realidad son las empresas

00:07:36.300 --> 00:07:38.000
y dentro de la empresa, los que toman las decisiones

00:07:38.000 --> 00:07:39.600
a veces tienen una edad más avanzada

00:07:40.400 --> 00:07:43.400
así que al asignar el presupuesto y pagar por esto

00:07:43.400 --> 00:07:46.000
no necesariamente entienden por completo esta necesidad

00:07:46.000 --> 00:07:48.800
así que nuestros usuarios y participantes

00:07:48.800 --> 00:07:50.500
a todos les encanta desde el primer momento

00:07:50.500 --> 00:07:52.700
el NPS básicamente está por encima del 80%

00:07:52.700 --> 00:07:55.600
así que es una plataforma muy bien recibida por los participantes

00:07:55.600 --> 00:07:59.000
pero quienes pagan son los organizadores de eventos y los clientes corporativos

00:07:59.000 --> 00:08:01.900
así que hay cierto rezago, supongo

00:08:02.000 --> 00:08:03.500
así que esto también es un desafío

00:08:03.500 --> 00:08:05.100
de la monetización de esta plataforma

00:08:06.400 --> 00:08:08.500
pero en ese momento, como las empresas pagaban por esto

00:08:08.500 --> 00:08:10.300
probablemente sí alejó a bastantes empresas

00:08:10.300 --> 00:08:11.900
en realidad, la mayoría de los nuestros son en libras esterlinas

00:08:11.900 --> 00:08:13.900
porque nuestro primer cliente fue Aula

00:08:13.900 --> 00:08:16.600
después muchas empresas nos buscaron por su cuenta

00:08:16.600 --> 00:08:18.200
diciendo que querían organizar un evento similar

00:08:18.200 --> 00:08:19.000
con presencia de marca

00:08:19.000 --> 00:08:22.000
con un espacio de branding propio

00:08:22.000 --> 00:08:22.200
sí

00:08:22.200 --> 00:08:24.900
así que esto también es un punto de venta para las empresas

00:08:24.900 --> 00:08:27.000
básicamente hicimos tanto 2C como 2D

00:08:27.000 --> 00:08:31.500
sí, tenemos participantes del lado C y organizadores del lado B

00:08:31.600 --> 00:08:32.200
un mercado de doble cara

00:08:32.800 --> 00:08:35.600
¿y cuál era más o menos el plan para los usuarios C en ese momento?

00:08:35.600 --> 00:08:38.200
organizamos aproximadamente 3000 eventos al año

00:08:38.200 --> 00:08:43.600
Y luego cada evento tenía decenas, cientos o incluso miles de personas, ¿no?

00:08:43.600 --> 00:08:43.800
Sí

00:08:43.800 --> 00:08:47.000
Así que es una plataforma bastante grande

00:08:47.000 --> 00:08:47.700
Sí

00:08:47.700 --> 00:08:49.500
Entonces es que

00:08:49.500 --> 00:08:50.900
Pero también tiene un problema

00:08:50.900 --> 00:08:54.400
Y es que no necesariamente es un modelo muy recurrente

00:08:54.400 --> 00:08:55.500
Porque una vez que terminas de organizar un evento

00:08:55.500 --> 00:08:58.000
tal vez el próximo lo organices hasta el próximo trimestre o el próximo semestre

00:08:58.000 --> 00:09:00.600
Así que no tiene un cobro automático tan bueno

00:09:00.600 --> 00:09:00.800
Sí

00:09:00.800 --> 00:09:03.100
Así que ese es también el problema de este tipo de productos

00:09:03.100 --> 00:09:05.000
es decir, las plataformas de eventos

00:09:05.000 --> 00:09:06.500
Tienen su novedad

00:09:06.500 --> 00:09:10.900
pero es difícil que se integren en el flujo de trabajo de una empresa

00:09:10.900 --> 00:09:12.700
para lograr que paguen a largo plazo

00:09:12.700 --> 00:09:16.800
Pero si estos niños de ahora eventualmente desarrollan una necesidad real de socialización virtual

00:09:16.800 --> 00:09:19.700
y empiezan a usarla activamente

00:09:19.700 --> 00:09:20.900
eso sí es una posibilidad

00:09:20.900 --> 00:09:21.300
Sí

00:09:21.300 --> 00:09:24.500
Yo creo que, a largo plazo

00:09:24.500 --> 00:09:29.400
este sigue siendo un producto y una plataforma con bastante valor

00:09:29.400 --> 00:09:32.700
así que más adelante, con los cambios en los dispositivos

00:09:32.700 --> 00:09:34.900
y los cambios en la estructura de los usuarios

00:09:34.900 --> 00:09:37.100
creo que podrían surgir nuevas oportunidades

00:09:37.100 --> 00:09:39.100
Entonces, ¿qué piensas ahora?

00:09:39.100 --> 00:09:42.000
¿Qué te gustaría hacer con la IA?

00:09:42.000 --> 00:09:44.400
Bueno, por mi background e intereses

00:09:44.400 --> 00:09:47.900
siempre he trabajado en el comportamiento humano

00:09:47.900 --> 00:09:50.800
en la intersección entre la conexión humana y la tecnología

00:09:50.800 --> 00:09:51.500
Sí

00:09:51.500 --> 00:09:55.100
Y mi experiencia anterior también se ha desarrollado en torno a esa línea principal

00:09:55.100 --> 00:09:56.500
La dirección que quiero tomar ahora

00:09:56.500 --> 00:10:04.000
es en realidad tratar de que la IA tenga cierta intuición humana y algo de humanidad

00:10:04.000 --> 00:10:05.500
Porque ahora vemos que en muchas ejecuciones de la IA

00:10:05.500 --> 00:10:08.000
sigue un flujo de trabajo

00:10:08.000 --> 00:10:10.500
sigue una regla codificada de forma rígida (hard coded)

00:10:10.500 --> 00:10:11.600
y luego la ejecuta

00:10:11.600 --> 00:10:14.700
Pero si le pides que tome decisiones y juicios por sí misma

00:10:14.700 --> 00:10:17.900
o que actúe según el contexto, ¿verdad?

00:10:17.900 --> 00:10:21.100
Then when making some of these professional choices

00:10:21.100 --> 00:10:23.500
it often produces relatively large deviations

00:10:23.500 --> 00:10:25.200
possibly not even as good as an intern

00:10:25.300 --> 00:10:28.800
So we hope to let AI converge

00:10:28.800 --> 00:10:31.800
so that it can be like a professional with a certain amount of experience

00:10:31.800 --> 00:10:34.400
like a seasoned professional

00:10:34.400 --> 00:10:38.100
and learn to do these professional tasks

00:10:38.100 --> 00:10:39.800
especially long-horizon ones

00:10:39.800 --> 00:10:41.200
and horizon-related ones

00:10:41.200 --> 00:10:44.500
tasks that can be used in actual work and have economic value

00:10:44.500 --> 00:10:45.400
Right

00:10:45.400 --> 00:10:50.800
So basically we're going to build a product platform

00:10:50.800 --> 00:10:54.500
to help these professionals

00:10:54.500 --> 00:10:55.400
share their knowledge

00:10:55.400 --> 00:10:58.100
share their workflow and some of their professional

00:10:58.100 --> 00:11:00.400
computer use operations on the computer

00:11:00.400 --> 00:11:05.000
Then from these I extract some key, effective behavioral information

00:11:05.000 --> 00:11:09.200
and teach AI how to do all kinds of work

00:11:09.200 --> 00:11:12.300
So that's our product positioning

00:11:12.300 --> 00:11:14.800
So do you think it's not a technical problem then

00:11:14.800 --> 00:11:16.000
It's a data problem

00:11:16.000 --> 00:11:17.300
It's actually a data problem

00:11:17.300 --> 00:11:20.500
Because we've found that all the large model companies

00:11:20.500 --> 00:11:24.400
allocate two-thirds of their manpower and resources

00:11:24.400 --> 00:11:28.000
to data collection, cleaning, labeling, and pre-training work

00:11:28.000 --> 00:11:31.400
So actually the quality of the data determines the quality of the training

00:11:31.400 --> 00:11:33.600
So as the saying goes, garbage in, garbage out

00:11:33.600 --> 00:11:35.900
I think data is extremely important

00:11:35.900 --> 00:11:40.200
Now AI training has just run into this kind of data bottleneck

00:11:40.200 --> 00:11:42.800
because data available online

00:11:42.800 --> 00:11:46.100
as well as data that can be manually labeled

00:11:46.100 --> 00:11:47.500
has basically been exhausted

00:11:47.500 --> 00:11:48.900
So there's a huge amount of real-world data

00:11:48.900 --> 00:11:52.800
that hasn't been effectively filtered and collected by anyone

00:11:52.800 --> 00:11:54.800
So this is a relatively big problem we want to solve

00:11:54.800 --> 00:11:58.500
That is, once there's scaled human demonstration

00:11:58.500 --> 00:12:04.200
that can teach AI how to operate all kinds of hands-on trades and actual work processes

00:12:04.200 --> 00:12:08.500
it can actually greatly improve AI's applicability in real-world scenarios

00:12:08.500 --> 00:12:09.900
and its usability

00:12:09.900 --> 00:12:13.700
So this is one of our product goals

00:12:13.700 --> 00:12:17.100
Puede clasificarse dentro de la industria del etiquetado de datos

00:12:17.100 --> 00:12:18.800
No queremos llamarlo la industria del etiquetado de datos

00:12:18.800 --> 00:12:22.900
Somos más bien una behavioral intelligence layer

00:12:22.900 --> 00:12:26.300
Es decir, basándonos en nuestros canales de datos y datos reales

00:12:26.300 --> 00:12:29.000
podemos construir algo similar a un modelo de comportamiento

00:12:29.000 --> 00:12:32.800
Este modelo puede hacer que los agentes de IA sean más eficientes

00:12:32.800 --> 00:12:36.900
y ejecuten tareas de tipo 'hammer' a menor costo

00:12:36.900 --> 00:12:38.400
En comparación con un modelo grande tradicional

00:12:38.400 --> 00:12:40.000
puede requerir menos pasos

00:12:40.000 --> 00:12:41.400
y menos tokens

00:12:41.400 --> 00:12:45.600
y en cambio aportar conocimientos y una intuición especializados en el ámbito 'hammer'

00:12:45.600 --> 00:12:46.700
que permiten hacerlo mejor

00:12:46.700 --> 00:12:49.100
Así que esto de hecho ya ha sido validado en el ámbito de la investigación

00:12:49.100 --> 00:12:52.400
porque ahora, por ejemplo, Amazon

00:12:52.400 --> 00:12:55.600
su equipo de research desarrolló un shopping agent

00:12:55.600 --> 00:12:58.400
utilizando el trace de los compradores

00:12:58.400 --> 00:13:02.000
es decir, las trayectorias de comportamiento reales de compradores reales

00:13:02.000 --> 00:13:06.600
para entrenar el shopping agent de IA

00:13:06.600 --> 00:13:08.600
el agente de compras

00:13:08.600 --> 00:13:11.100
el resultado real, comparado con el modelo grande base

00:13:11.100 --> 00:13:12.100
al hacer que elija productos

00:13:12.100 --> 00:13:14.100
y complete todo el proceso de compra

00:13:14.100 --> 00:13:15.500
muestra una mejora de entre un veinte y un treinta por ciento

00:13:15.700 --> 00:13:19.100
así que esto ya ha sido validado en un ámbito 'hammer'

00:13:19.100 --> 00:13:21.500
Por eso queremos, desde el lado de la obtención de datos

00:13:21.500 --> 00:13:24.700
y desde la construcción a nivel de producto de la infraestructura,

00:13:24.700 --> 00:13:29.100
poder generar un resultado diferente

00:13:29.100 --> 00:13:32.400
y ayudar a que la IA se despliegue y ejecute de manera realmente efectiva

00:13:33.700 --> 00:13:36.400
¿Entonces esperas poder vendérselo a las empresas de modelos grandes

00:13:36.400 --> 00:13:38.200
para que mejoren su producto de propósito general

00:13:38.200 --> 00:13:40.500
o venderlo a clientes específicos

00:13:40.500 --> 00:13:43.900
para que mejoren su flujo de trabajo?

00:13:43.900 --> 00:13:45.500
Esto tiene un orden de prioridad

00:13:45.500 --> 00:13:47.300
Es decir, actualmente las empresas de modelos grandes

00:13:47.300 --> 00:13:52.200
tienen mucha demanda por comprar este tipo de datos

00:13:52.200 --> 00:13:54.400
y eso también nos puede generar ingresos tempranos

00:13:54.400 --> 00:13:56.000
En cuanto al futuro

00:13:56.000 --> 00:13:59.300
en realidad nuestra visión sigue siendo convertirnos en una behavioral layer

00:13:59.300 --> 00:14:03.200
es decir, poder tener, por encima de estos modelos grandes base,

00:14:03.200 --> 00:14:08.100
una capa similar a esta de comportamiento

00:14:08.100 --> 00:14:10.600
可以让agent可以自己去教学 -> To allow the agent to teach itself

00:14:10.600 --> 00:14:12.000
自己去学习 -> Learn on its own

00:14:12.000 --> 00:14:12.900
从人的示范 -> From human demonstrations

00:14:12.900 --> 00:14:14.900
从专家的这个经验来说 -> Speaking from the experience of experts

00:14:14.900 --> 00:14:18.000
可以可以就是服务agent -> Yes, it's about serving the agent

00:14:18.000 --> 00:14:19.400
这样一个expert marketplace -> An expert marketplace like this

00:14:20.700 --> 00:14:23.300
我也没想服务agent -> I hadn't thought about serving the agent either

00:14:23.300 --> 00:14:24.700
对因为未来的话 -> Right, because in the future

00:14:24.700 --> 00:14:30.700
其实agent他会自己去找很多的学习内容 -> The agent will actually go find a lot of learning content on its own

00:14:30.700 --> 00:14:31.600
对因为现在agent -> Right, because right now the agent

00:14:31.600 --> 00:14:35.500
其实你需要喂他很多上下文 -> You actually need to feed it a lot of context

00:14:35.500 --> 00:14:38.400
告诉他很多的documentation以及skill -> Tell it a lot of documentation and skills

00:14:38.400 --> 00:14:40.600
但是其实后面agent会自己去找 -> But actually later the agent will go find it on its own

00:14:40.800 --> 00:14:42.700
我这类不懂或是没做过 -> Things like this that I don't understand or haven't done

00:14:42.700 --> 00:14:44.800
我应该怎么去学习 -> How should I go about learning this

00:14:44.800 --> 00:14:48.100
那这时候就需要一个有大量人类的 -> At that point, you need a large repository of human

00:14:48.100 --> 00:14:51.300
这个示范和经验的一个库 -> Demonstrations and experience

00:14:51.300 --> 00:14:55.100
这个库是可以让AI来去学习示范的 -> This repository is meant to let AI learn from demonstrations

00:14:55.100 --> 00:14:56.800
然后他自己可以教自己 -> And then it can teach itself

00:14:56.800 --> 00:14:59.400
去执行这些具体的工作 -> To carry out these specific tasks

00:14:59.400 --> 00:15:01.000
因为其实每个工作人都做过 -> Because actually every job has been done by someone

00:15:01.000 --> 00:15:02.000
只不过这些数据 -> It's just that this data

00:15:02.000 --> 00:15:03.600
现在没有被很好的去留存 -> Hasn't been well preserved so far

00:15:06.100 --> 00:15:09.000
我想一想就是就是你觉得 -> Let me think, so you think that

00:15:09.100 --> 00:15:12.400
虽然模型比方说模型现在回答问题 -> Although models, for example, when answering questions now

00:15:12.400 --> 00:15:15.600
他有很多数据是他训练金里面的 -> A lot of the data comes from within its training set

00:15:15.600 --> 00:15:18.900
但他有一些东西是现场去网上搜的 -> But some of it is searched online in real time

00:15:18.900 --> 00:15:22.000
然后你觉得他之后就agent -> And you think that afterward, for the agent

00:15:22.000 --> 00:15:25.000
做事情的那部分也会是这样 -> The part where it does things will also be like this

00:15:25.000 --> 00:15:25.800
对对对 -> Right, right, right

00:15:25.800 --> 00:15:27.600
因为其实就像你现在 -> Because actually, just like now

00:15:27.600 --> 00:15:29.300
如果作为开发者来说 -> If you're a developer

00:15:29.300 --> 00:15:32.400
你不知道一个就是solution -> If you don't know whether a solution

00:15:32.400 --> 00:15:34.000
有没有现成的API和SDK的话 -> Has a ready-made API and SDK

00:15:34.000 --> 00:15:35.600
你自己会上Github搜 -> You'll go search on Github yourself

00:15:35.600 --> 00:15:36.800
其实这是类似一个逻辑 -> This is actually a similar logic

00:15:36.800 --> 00:15:40.300
就是最后agent要学一个锤类的知识 -> In the end, the agent needs to learn a certain type of knowledge

00:15:40.300 --> 00:15:42.000
一个具体任务人是怎么做的 -> How a person does a specific task

00:15:42.000 --> 00:15:44.000
他也会去找对应的reference -> It will also go find the corresponding reference

00:15:44.000 --> 00:15:45.300
所以我们希望提供这样一个 -> So we hope to provide something like this

00:15:45.300 --> 00:15:46.800
Equivale a decir que es un sistema de referencia

00:15:46.800 --> 00:15:48.900
que le permite aprender rápidamente

00:15:50.800 --> 00:15:51.600
Voy a repetir un punto

00:15:51.600 --> 00:15:53.700
Es decir, tú crees que no

00:15:53.700 --> 00:15:57.100
Las empresas de grandes modelos no van a entrenar todo dentro de su modelo

00:15:57.100 --> 00:16:02.100
sino que necesitan una referencia externa para poder hacerlo

00:16:02.100 --> 00:16:02.700
Sí

00:16:02.700 --> 00:16:06.200
Creo que la razón es que hacer eso con grandes modelos no es económico ni eficiente

00:16:06.200 --> 00:16:08.700
Es decir, ahora mismo usando un modelo general

00:16:08.700 --> 00:16:12.000
es imposible resolver todas las tareas comunes

00:16:12.000 --> 00:16:13.200
Este es un gran juicio mío

00:16:13.200 --> 00:16:14.700
Porque una investigación reciente se llama

00:16:14.700 --> 00:16:16.400
es la agent last exam

00:16:16.400 --> 00:16:17.700
de hecho nosotros también tenemos investigación al respecto

00:16:17.700 --> 00:16:19.700
básicamente compara

00:16:19.700 --> 00:16:24.000
tal vez entre 150 y 200 tipos de trabajo reales concretos

00:16:24.000 --> 00:16:26.500
que son todos bastante complejos

00:16:26.500 --> 00:16:27.800
con múltiples operaciones de software

00:16:27.800 --> 00:16:29.200
y que requieren múltiples pasos

00:16:29.200 --> 00:16:32.900
y resulta que incluso en los modelos SOTA de frontera

00:16:32.900 --> 00:16:34.600
en modo one-shot

00:16:34.600 --> 00:16:38.000
la proporción que logra completarse end-to-end es solo del 2%

00:16:38.000 --> 00:16:40.400
así que en este aspecto creo

00:16:40.400 --> 00:16:44.300
que en realidad será muy difícil lograr una solución única para todo en tres a cinco años

00:16:44.300 --> 00:16:47.400
Creemos que en realidad cada campo debería tener su propia acumulación

00:16:50.900 --> 00:16:51.600
Voy a añadir un poco más

00:16:51.600 --> 00:16:53.400
porque este tema realmente me interesa mucho

00:16:53.400 --> 00:16:55.800
porque ahora se puede ver que incluso él

00:16:55.800 --> 00:16:56.900
como acabo de decir, ¿verdad?

00:16:56.900 --> 00:16:58.800
incluso cuando un gran modelo responde preguntas

00:16:58.800 --> 00:17:01.500
desde cierto año ya no se limita solamente

00:17:01.500 --> 00:17:03.600
a mirar dentro de la sala de entrenamiento de su propio modelo

00:17:03.600 --> 00:17:07.200
sino que también recurre a búsquedas externas en línea

00:17:07.200 --> 00:17:09.800
esto en realidad va a continuar de manera constante

00:17:09.800 --> 00:17:11.700
por eso el nuevo aprendizaje externo siempre es útil

00:17:11.700 --> 00:17:16.200
así que ahora esto se extiende a la capa de comportamiento del agente

00:17:16.200 --> 00:17:19.600
desde la perspectiva del agente

00:17:19.600 --> 00:17:21.400
sí, porque esto también tiene cierto eco

00:17:21.400 --> 00:17:22.900
con la dirección en la que tú trabajas

00:17:22.900 --> 00:17:25.400
es decir, tú en realidad proporcionas las relaciones sociales de una persona

00:17:25.400 --> 00:17:29.900
和这个社交这个资产

00:17:29.900 --> 00:17:31.400
作为一个API

00:17:31.400 --> 00:17:34.800
或作为一个agent可以去铺的一个network

00:17:34.800 --> 00:17:38.000
对然后我其实把大家脑海里的知识和经验

00:17:38.000 --> 00:17:39.200
也做成这样一个network

00:17:39.200 --> 00:17:40.400
让大家可以去铺

00:17:40.400 --> 00:17:42.100
对所以这个逻辑

00:17:42.100 --> 00:17:43.500
我觉得是有相通之处的

00:17:43.500 --> 00:17:45.400
因为只要人存在吧

00:17:45.400 --> 00:17:46.200
他的关系

00:17:46.200 --> 00:17:50.200
他的这个经验和这种专业判断和直觉

00:17:50.200 --> 00:17:54.400
其实是很难被AI就是穷尽的去学完的

00:17:54.400 --> 00:17:55.900
那我这时候我就冒出来一个想法

00:17:55.900 --> 00:18:02.500
就是这个专业的数据库

00:18:02.500 --> 00:18:04.400
专业的行为层数据库

00:18:04.400 --> 00:18:05.500
它已经不是知识层了

00:18:05.500 --> 00:18:06.900
知识层是别人解决的

00:18:06.900 --> 00:18:09.400
它是一个怎么操作电脑啊

00:18:09.400 --> 00:18:11.800
这么一个层次

00:18:11.800 --> 00:18:14.000
那是因为这个层次里面的数据

00:18:14.000 --> 00:18:15.600
它会产生很多动态的变化吗

00:18:15.600 --> 00:18:17.900
所以你要它是一个实时的

00:18:17.900 --> 00:18:20.300
他说他只是因为常美

00:18:21.700 --> 00:18:24.400
这个东西不值得大魔星自己去内化的train

00:18:26.100 --> 00:18:27.200
两方面都有吧

00:18:27.200 --> 00:18:29.700
确实他太常为太分散

00:18:29.700 --> 00:18:31.800
所以你没法学所有东西

00:18:31.800 --> 00:18:34.100
然后第二个就是说

00:18:34.100 --> 00:18:39.300
他其实也是一个需要很多reasoning的一个行为

00:18:39.300 --> 00:18:41.400
因为人在操作不同的软件

00:18:41.400 --> 00:18:43.500
不同的这个键盘鼠标操作的时候

00:18:43.500 --> 00:18:44.400
他其实会想吧

00:18:44.400 --> 00:18:46.900
每一步他其实经过了一个思考

00:18:46.900 --> 00:18:51.400
然后说我考就是考察哪些信息

00:18:51.400 --> 00:18:53.300
然后就是take into account对吧

00:18:53.300 --> 00:18:54.500
考虑了哪些因素

00:18:54.500 --> 00:18:56.600
然后我和谁可能还沟通什么什么内容

00:18:56.600 --> 00:18:58.800
然后最后才做了这个案件

00:18:58.800 --> 00:19:00.800
去提交或者说去分享对吧

00:19:00.800 --> 00:19:04.200
然后这个过程中其实有很多很多的复杂的

00:19:04.200 --> 00:19:07.500
This is a thinking chain of multi stamp

00:19:07.500 --> 00:19:08.800
A chain of thought inside it

00:19:08.800 --> 00:19:13.900
This thing actually isn't being captured and recorded very well right now

00:19:13.900 --> 00:19:17.100
So what we hope is that, beyond the log of everyone's operations

00:19:17.100 --> 00:19:21.800
there's also this layer that can attach itself on top of the raw data

00:19:21.800 --> 00:19:23.800
so that in the end it can be handed over to AI

00:19:23.800 --> 00:19:25.300
I encountered a similar task

00:19:25.300 --> 00:19:27.900
can we learn to think about it the way a human would

00:19:27.900 --> 00:19:29.300
OK

00:19:29.300 --> 00:19:31.100
And since you graduated from Yale

00:19:31.100 --> 00:19:33.700
you came to the US as an undergraduate

00:19:33.700 --> 00:19:34.100
Right

00:19:34.100 --> 00:19:35.600
I actually came to the US in high school

00:19:35.600 --> 00:19:36.900
Came to the US in high school

00:19:36.900 --> 00:19:39.900
So then, given our age group

00:19:39.900 --> 00:19:43.700
you probably have a lot of classmates who are already at large model companies

00:19:43.700 --> 00:19:44.100
Yes

00:19:44.100 --> 00:19:48.400
In terms of the network we have as founders

00:19:48.400 --> 00:19:51.900
because we get in touch with a lot of researchers

00:19:51.900 --> 00:19:55.600
who move between different frontier labs

00:19:55.600 --> 00:19:55.700
Right

00:19:55.700 --> 00:19:59.200
And we have an advisor who works on external machine learning at scale AI

00:19:59.200 --> 00:20:02.700
so his network and his students are also largely at frontier labs

00:20:02.700 --> 00:20:02.900
Right

00:20:04.300 --> 00:20:07.600
So based on your past entrepreneurial experience

00:20:07.600 --> 00:20:09.900
you have the ability to handle enterprise clients

00:20:09.900 --> 00:20:12.500
as well as personal relationships

00:20:12.500 --> 00:20:17.200
so you're able to get in touch with large model companies

00:20:17.200 --> 00:20:19.300
going back to that initial point about the money-making need

00:20:19.300 --> 00:20:20.100
Right

00:20:20.600 --> 00:20:22.500
And you've also communicated with them directly

00:20:22.500 --> 00:20:24.300
so this need is indeed real

00:20:24.300 --> 00:20:24.500
Right

00:20:24.500 --> 00:20:25.700
This is a very real need

00:20:25.700 --> 00:20:29.700
Because recently I heard that companies like Salesforce

00:20:29.700 --> 00:20:31.500
spent two hundred thousand dollars

00:20:31.500 --> 00:20:35.400
to buy twenty long-thread operation trajectory datasets

00:20:35.400 --> 00:20:38.500
meaning using their own software to complete a task

00:20:38.500 --> 00:20:43.100
and a single high-quality trajectory like that is worth about ten thousand dollars

00:20:43.100 --> 00:20:43.600
Right

00:20:43.600 --> 00:20:48.800
然后其他所有的很多公司都需要这样的操作轨迹来训练他们自己的AI translates to: Then all the other many companies need this kind of operational trajectory data to train their own AI

00:20:48.800 --> 00:20:54.400
So this is what we think is the next relatively big opportunity and bottleneck

00:20:54.400 --> 00:20:59.000
That is to say, for AI to truly achieve the so-called Long Horizon AGI goal

00:20:59.000 --> 00:21:01.800
It still requires a large amount of human experience

00:21:01.800 --> 00:21:05.300
And then use that to demonstrate and teach AI

00:21:05.300 --> 00:21:07.000
So this demand is very real

00:21:07.000 --> 00:21:07.500
And then

00:21:07.500 --> 00:21:10.800
You can also reach the next step, which is finding people to do it

00:21:10.800 --> 00:21:11.300
Yes, yes, yes

00:21:11.300 --> 00:21:11.600
Yes

00:21:11.600 --> 00:21:11.900
Yes

00:21:11.900 --> 00:21:12.100
Right

00:21:12.100 --> 00:21:17.300
Right now we're focused on building an early prototype of this kind of product

00:21:17.300 --> 00:21:19.400
And then getting it launched as quickly as possible

00:21:19.400 --> 00:21:22.300
And then selling the first wave of contracts

00:21:23.700 --> 00:21:25.100
I feel like it's actually a pretty good fit

00:21:25.100 --> 00:21:28.500
Because before I also did online recruiting in the virtual world

00:21:28.500 --> 00:21:32.500
This sounds quite similar

00:21:32.500 --> 00:21:32.800
Right

00:21:32.800 --> 00:21:33.100
Yes

00:21:33.100 --> 00:21:33.400
Yes

00:21:33.400 --> 00:21:33.600
Right

00:21:33.600 --> 00:21:37.400
This is really in line with my previous interests and background

00:21:37.400 --> 00:21:41.400
And then we also hope to have a more productized, standardized solution

00:21:41.400 --> 00:21:44.300
That can source and export at scale

00:21:44.300 --> 00:21:46.000
And then give them the corresponding incentives and rewards

00:21:46.000 --> 00:21:47.900
And then let them get onboarded

00:21:47.900 --> 00:21:50.300
And then we can be relatively dynamic

00:21:50.300 --> 00:21:52.400
Depending on different needs

00:21:52.400 --> 00:21:54.100
Source different kinds of expertise

00:21:55.300 --> 00:21:55.600
Right

00:21:55.600 --> 00:21:56.800
This is very similar

00:21:56.800 --> 00:21:59.600
Because before, you were basically dealing with top-tier major clients

00:21:59.600 --> 00:22:01.700
Top-tier major clients have recruiting budgets

00:22:01.700 --> 00:22:02.300
So what about now

00:22:02.300 --> 00:22:04.000
These model companies also have recruiting budgets

00:22:04.000 --> 00:22:05.200
So they're not short on money

00:22:05.200 --> 00:22:05.800
Yes, yes, yes

00:22:05.800 --> 00:22:09.900
Because each model company basically has a one billion budget

00:22:09.900 --> 00:22:11.700
For data procurement

00:22:11.700 --> 00:22:11.900
Right

00:22:11.900 --> 00:22:13.700
And this part is actually very certain

00:22:13.700 --> 00:22:15.000
It's also a very large market

00:22:15.000 --> 00:22:15.500
They lack people

00:22:15.500 --> 00:22:18.000
And then there was also the other side of that recruiting event

00:22:18.000 --> 00:22:19.100
Which is people looking for jobs

00:22:19.100 --> 00:22:20.000
Right, right, right, right, right

00:22:20.000 --> 00:22:22.600
So actually dating is similar too

00:22:22.600 --> 00:22:24.000
It's also a two-sided marketplace

00:22:24.000 --> 00:22:24.200
Right

00:22:24.200 --> 00:22:25.600
So I think a marketplace

00:22:25.600 --> 00:22:29.900
is still part of the entrepreneurial architecture and thinking I have as a product manager

00:22:29.900 --> 00:22:30.300
It's like

00:22:30.300 --> 00:22:30.900
Oh no, oh no

00:22:30.900 --> 00:22:36.300
Is this the common dream of product managers with a background in China, in the industry in China?

00:22:36.300 --> 00:22:38.900
Will people think we're outdated?

00:22:38.900 --> 00:22:39.600
I don't think so

00:22:39.600 --> 00:22:42.700
In the AI era, there's still wisdom in this kind of marketplace

00:22:42.700 --> 00:22:44.100
And once a marketplace gets started

00:22:44.100 --> 00:22:45.600
it still has a flywheel effect

00:22:45.600 --> 00:22:48.000
So it will have relatively fast growth

00:22:49.000 --> 00:22:53.100
My feeling is that the people around us all quite like this network

00:22:53.100 --> 00:22:53.500
and things like that

00:22:53.500 --> 00:22:56.300
But some people think this is something left over from the mobile era

00:22:57.700 --> 00:22:59.100
Right?

00:22:59.100 --> 00:23:00.800
How do you see this, are we old-fashioned

00:23:00.800 --> 00:23:04.600
or is this the next generation of product-type opportunities that should naturally emerge

00:23:06.100 --> 00:23:07.200
I think there is an opportunity

00:23:07.200 --> 00:23:09.900
otherwise you and I wouldn't be sitting here discussing a topic like this

00:23:09.900 --> 00:23:12.000
Doing similar things all have networks, right?

00:23:12.000 --> 00:23:13.400
directions with potential

00:23:13.400 --> 00:23:20.600
But I don't think every field or direction is suitable for solving with a network

00:23:20.600 --> 00:23:20.800
Right

00:23:20.800 --> 00:23:23.100
So for example, some vertical AI in niche areas

00:23:23.100 --> 00:23:25.100
it might not inherently have a network effect

00:23:25.100 --> 00:23:28.000
it might just need to be forward deployed and implemented one by one

00:23:29.000 --> 00:23:30.800
There's mainly another point of view

00:23:30.800 --> 00:23:33.100
As soon as you hear this, it sounds very mobile-internet-era, very outdated

00:23:33.100 --> 00:23:33.400
Right?

00:23:33.400 --> 00:23:35.600
Your mind has been fixed in the mold of mobile internet

00:23:35.600 --> 00:23:37.400
Esta también es una opinión bastante común

00:23:37.400 --> 00:23:38.100
¿Cómo la refutas?

00:23:39.900 --> 00:23:42.600
Creo que hay que partir de los principios fundamentales

00:23:42.700 --> 00:23:44.600
Es decir, para resolver este problema en el fondo

00:23:44.600 --> 00:23:47.300
Qué tipo de arquitectura y método es mejor para resolverlo

00:23:47.300 --> 00:23:47.400
Sí

00:23:47.400 --> 00:23:50.600
Luego vemos este tipo de cuellos de botella en los datos

00:23:50.600 --> 00:23:55.400
En realidad es un suministro de long tail bastante disperso

00:23:55.400 --> 00:23:56.900
En realidad es bastante adecuado resolverlo con esta arquitectura

00:24:01.700 --> 00:24:03.400
Entonces, ¿te vas a quedar en Estados Unidos de ahora en adelante?

00:24:03.400 --> 00:24:03.600
Sí

00:24:03.600 --> 00:24:05.900
Ahora me estoy reubicando, con base en San Francisco

00:24:07.800 --> 00:24:08.300
Bien

00:24:08.300 --> 00:24:09.300
¿Hay algo más de lo que quieras hablar?

00:24:09.300 --> 00:24:09.500
Bien

00:24:09.500 --> 00:24:11.100
Creo que con esto es suficiente por hoy

00:24:11.100 --> 00:24:11.400
Gracias
