Luyện nói tiếng Anh bằng Shadowing qua video: 24h Inside $5M Silicon Valley AI Startup Building Personal Super Intelligence
Đang tạo bài học...
1
this stack of chicken bake boxes reaches the ceiling.
2
We just gotta go give up and find jobs and get mad because we can't keep doing this.
3
It's doing everything right.
4
Like not just content and not just like the ads and not just like the onboarding, but it has to all work.
5
Most venture capital funded AI startups are building for engineers and enterprises.
6
And almost no one is building for regular people because consumer is one of the most brutal categories you can touch.
7
But on this trip, I met a team taking a more creative approach to consumer AI than most startups have come across so far.
8
That's our definition of personal super intelligence.
9
It is one that understands your goals and is unequivocally for you.
10
What's really special about Tomo is
11
that it's really trying to inspire everyone to pursue their life's work in one aspect or another.
12
This is a huge space.
13
Our team is the entire world.
14
We are excited to build something that will go large and big and go beyond the beta.
15
Today I'm going behind the scenes with Justin and the team from Tomo, a personal AI that lives in your text messages,
16
learns your goals, and is now starting to build custom software around your life.
17
They've just raised a five million dollar seed round led by Bain Capital Ventures
18
and already have over 500,000 people texting Tomo to run their lives.
19
There's like such wide range of interfaces we can create that really motivate people, which is what I'm particularly excited about.
20
A lot of developers and founders try to overcomplicate.
21
But when you actually look at what consumers want and how hard it is to develop just the most simple features, a company like Tomo is building just for the simple stuff,
22
for the everyday people stuff.
23
It's a category so new, it's almost impossible to put a number on.
24
So the real question I want to answer is whether a
25
small team can actually win in the one space most of Silicon Valley is too afraid to build for.
26
Oh my god, walking up a hill.
27
I feel so unfit.
28
I am going behind the scenes with Justin and the team from Tomo.
29
They've raised their seed round from some pretty big names
30
and they are only a month away from sending this live and launching their first proper product to market.
31
Tomo is heavily centered as more of like an AI agent.
32
Infrastructure for the average human being and I've always been skeptical about these kind of plays in AI, but Justin gave me access to the product,
33
full access the other day, and I've actually used it every day since.
34
So if I'm a good indication of using a product like this and something that's actually sticky and defensible,
35
I'd hope that's a good litmus test on whether something's actually pretty cool.
36
So we're going behind the scenes with them.
37
We're gonna go to their house first because I'm pretty sure they just signed the lease for a new office.
38
So we're gonna go check out where it all began, and then we're gonna go to the new place.
39
So, let's go see him.
40
This is where I sat up until like basically yesterday for the last two and a half years.
41
Over there is where Ray has sat for the last year or so.
42
And then Varun about a couple months ago started sitting over there.
43
Yeah, this has been the 100 square feet that we've been working out of.
44
Wild.
45
Yeah, so what's the story of the chicken bag boxes?
46
What is that about?
47
We had worked on a lot of different products for consumer.
48
For the vast majority, almost everything, we didn't charge, but our op-max was quite high.
49
But it got to the point where we just looked at each other and said, OK, if this stack of chicken bag boxes reaches the ceiling, we just got to go give up and find jobs again,
50
man, because we can't keep doing this.
51
And super luckily, in that sort of idea maze, there's like about a box and a half of clearance that you'll see before, yeah, we would've had to go and like, sort of call it quits.
52
Good timing in that there's just enough juice left before things really started working.
53
Literally.
54
Yeah, stuff really was not working for the better half of two years.
55
It's like, maybe a box and a half of clearance, but...
56
That's crazy.
57
You can probably tell by the expiration dates of the boxes when like, things were really rough, and then we were like, okay, we can afford to eat like, a little better from this week.
58
This is insane.
59
Oh my god, that's so close, man.
60
Yeah, but one of the, maybe it's fate, but one of the cool things about the new office is
61
that it's also like a five minute walk from Costco.
62
So we can go straight to the source and we don't have to get the frozen stuff anymore.
63
So dude, real quick, if you were to sum it up in kind of one or two sentences, like really layman's terms, like basic, what is Tomo?
64
Yeah, so Tomo is personal AI for real people.
65
And what that means is that it's not AI necessarily for doing the work in the way that we, in Silicon Valley, perceive doing the work,
66
but it is AI that learns your goals and will do whatever it takes to help you get there.
67
And practice for anyone that oftentimes wants to go and lose weight
68
or go to the gym more often
69
or wants to keep track of their finances or wants to go
70
and do the childhood hobby and maybe they used to go and sing and write together.
71
That's what a lot of people are using Tomo for.
72
Today I'm going through a couple of sort of like meetings that you have to launch product-wise.
73
I made some changes yesterday on Google chats in an effort to make it counter-intuitively less proactive so it feels more assisted.
74
That's the feedback that we've seen.
75
My focus as with the rest of this week has been app building.
76
one that I talked to me about is it's hard to actually iterate on like the
77
or even use the end-to-end system right now of like testing
78
that building without example context of like here's an example chat
79
of like oh this is the goal they care about whether it's like calorie tracking
80
or financial things and it's actually like thus build like a proper app on top of
81
so going to be working on a way to just like implement ten different scenarios
82
that we can we use for the sake of future evals, but primarily now for app building.
83
I've been looking into some of how our ping backoffs happen
84
and some of the messaging around what Tomo is saying when it decides to back off.
85
And instead of merely today backing off and kind of disappearing into the void,
86
reframing that as an opportunity to improve what the user was originally trying to do with Tomo.
87
I took a look and I think I have like a much cleaner version of this
88
which uses a more improved model and also
89
cleaner structure that separates first into the first message being like hey I noticed you did this thing
90
and then the second message is more actionable and offers like three options for you to re-engage.
91
It's kind of related to what we talked about yesterday where
92
smoothness is actually not necessarily conducive to reduction
93
and some of the like the entropy
94
that you might expect within like a real conversation of even
95
some of the sharpness can be helpful obviously you don't know what the backoff is going to be like
96
but there's some level of delight from the kind of like
97
wabi-sabi of like a message that's like hey like what's like why aren't you coming back
98
or like hey what about this thing
99
that you promised me yeah some people like this this contrast in the sharpness
100
and if it came to like a real-life human relationship
101
which there's like push and pull and it's not just like constantly pull like like most other sort of language model products.
102
What are we doing now?
103
We are going to Costco for lunch but then more importantly to make sure
104
that Varun and Jessie get their employee benefit which is a Costco executive membership. And Jackie.
105
Let's go. And Jackie.
106
Very important.
107
I've been waiting for this.
108
I've been telling Tomo this like every day.
109
And getting his badge of adulthood which is a Costco membership card.
110
I'm excited.
111
I appreciate it.
112
He's locked in.
113
That's the one that's the one.
114
Jesse won honestly.
115
Wow.
116
Nice sample.
117
Good job.
118
Oh here it is man.
119
The company would not exist without these guys.
120
Okay, you asked me if there's a Costco in China.
121
Tomo says yes, definitely.
122
They opened the first one in Shanghai.
123
Oh, they have many?
124
Yeah, they have a lot.
125
Oh shit!
126
I'm gonna ask Tomo how many calories I think all of this is.
127
Mmmmm.
128
I don't recall.
129
Holy, you guys really went for it.
130
The spread is actually insane.
131
Just for the whole table, you're looking at at least 6,000 to 7,000 calories easy.
132
Wait, how many thousand?
133
6, 6, 7.
134
One fun thing is like unreleased but
135
Komo is starting to go and generate interfaces for people and
136
so it knows I've been as you know like eating a lot of these chicken bakes
137
and it's like oh it'd be fun to keep track of exactly how many outside of the boxes
138
and so it's developed this chicken bake app for me and so if I go in
139
You can see that it is like 36 this year
140
and I guess the month just started so zero and
141
so I can take a photo and text it like hey
142
and then it'll increment or if it's just easier
143
because you know Sometimes you don't want to go and text with these like AI agents.
144
I can go into this like fun interface
145
and be like oh I just ate one today and then we'll like add on Number of chicken bakes per month.
146
Yeah, but then also Tomo obviously has like comments about About this.
147
Oh, and he's reminding me that like I have to go in fixed or like
148
So it's like actually I need you to do a real thing right now Yeah, I just like okay like after you finish like get back
149
to work like this is what you have to do or what you told me But yeah, that's like the sort of like fun sort of chapter two
150
that we're excited to release pretty soon Where it's based on everything that you're sharing with your Tomo just texting.
151
Yeah What is that personal operating system look like?
152
What are games that we can create based off of this?
153
And you were saying like you've got you've got really like strong API connections with some pretty cool stuff So like yeah,
154
I'm talking about like budget tracking like cuz you got plaid like fitness health like all that stuff Yeah, it's stuff that we've only fitness and health that you'll need after eating this.
155
Yeah.
156
Yeah, especially after this My tone will probably send me a ping about getting back in track.
157
So today we're going to work through everything, like how the communication will go, what is the project like, and also connect a little bit more information from you guys that we needed.
158
So this is gonna be the page that we will share with you.
159
It's kind of our main source of truth.
160
Here you will find all the information about the project.
161
This timeline is in line where the kind of hard launch that we're calling it, that Jackie's helping us with.
162
But we ultimately want to have products that we can get in the hands of friends quite a bit earlier than that.
163
You know, even as soon as the next couple weeks, like you've seen even the couple days last week, we're already cooking and thinking through these different experiences.
164
Yeah, yeah, and it keeps us also like early testing, so that's great.
165
So we can also use it on users immediately.
166
So how did that meeting go for you?
167
Like, and actually probably more important question, why are you engaging in an agency from Amsterdam?
168
Yeah, yeah, yeah, really interesting kickoff meeting, like Justin said.
169
I think they're really great cultural people, so I'm excited to get like their perspectives on especially like what are actually meaningful products
170
and apps for again like real people out there not those that are, you know, kind of in sort of the tech bubble that, I mean, obviously we see it live in here.
171
Yeah, I think Kickoff's great.
172
Looking forward to working with them over the next few weeks.
173
You wanna see this?
174
There's like one thing that you get, it's like Jesse is not an engineer.
175
He is our head of growth, but he runs more coding agents than anybody at the company combined.
176
Run me through, what's the setup here?
177
Overall, I am actually really bad at typing, so...
178
The main thing I use a lot, of course, is Whisperflow.
179
Um...
180
Yeah, yeah, you interviewed them recently.
181
Yo, Hector.
182
Yeah.
183
So something funny that I do is that I have my backspace, so that it deletes by words instead of by character.
184
So, it's like...
185
Oh, gangster, that's cool.
186
Yeah, it's a bit weird, but I...
187
But that's because you're speaking into it.
188
So Flo will already correct your words.
189
So you're just faster, right?
190
Yeah, it's just like speed of editing, essentially.
191
Nice.
192
All I do actually all day, I feel like a salesperson, but all I do all day is talk.
193
Tell me about version 2
194
because like we talked a little about version 1 we understand how it works
195
but like personal OS literally custom building software is based on your preferences like tell me about
196
that how does that work? this thought really arrived from all the data
197
that we saw
198
which is like it's crazy how many people were sending things they cared about to Tomo their preferences some context
199
but even more
200
so just like images like things they care about for example like tracking like the food they eat
201
and sort of like calories or like macros associated with it.
202
The fact that they were like going to the gym to work out every time and laying to know.
203
And it's sort of like this realization
204
that like this UX was pretty suboptimal with like iMessage or really just like any raw chat interface.
205
You want to be able to just like talk to this like entity
206
and have it have that context and give you like correct answers about like what to do.
207
But what's even better is like what
208
if you can actually see like visuals of like here's my
209
progress here's how I'm doing another example is like what
210
if I am like trying to save enough money to like
211
buy this car next month reality is like there's actually like
212
a lot of ways you can create like budgeting apps
213
and use like other sort of like finance tracking apps out there
214
but they're not really like optimized for this scenario what would really be cool is like what
215
if it was just like a widget
216
that was like a progress bar to like a car emoji like that's a lot like simpler more motivating
217
and just like clear like here's my goal
218
and here's my progress to reach out rather than like uploaded like budget tracking app
219
so I'm super excited to sort of like bring this like dynamic like interface world to life.
220
One of the things that's really important to understand here is
221
the capability of an app like this actually really simple for the end user I'll show you a simple feature I know.
222
Hey Tomo I'm getting pretty hungry you know how I like healthy foods you've got my location
223
so can you ping me maybe like five of the closest restaurants near here where I kind of like grab
224
and go get something takeaway like I don't want to sit
225
into a fine dining restaurant find something that's like pretty easy
226
also relatively cheap I send it in I'm gonna say on
227
it let me find you some good healthy spots nearby
228
so it's like got a bit of attitude it's got a bit of its own like language
229
and it's just hitting me with locations
230
and the cool part is I can say to it okay cool could you remind me at 5 p.m
231
because I'm a bit busy now
232
but can you remind me to head out of the office
233
and actually go eat something I'll send that to it
234
and then it's going to save that it's going to message me
235
and say yeah no worries but it's a simple thing
236
so it's like help me find recommendations help me go to
237
the gym help me drink water help me do really basic stuff
238
and it's funny when you get into this how hard it is to actually just do basic stuff
239
and to not overcomplicate this especially in consumer world
240
because consumers are very very very difficult to bring on consumer software is one of the most cutthroat categories
241
that you can jump into in consumer okay real quick it is 5.02 pm
242
and right on time tomo just messaged me Wild. So cool.
243
The thing we do on the channel is we ask one piece of advice from founder to founder.
244
So you're in the thick of it right now, right?
245
Like raising a seed here.
246
So is there any bit of advice you'd give for a
247
founder that's maybe just a little bit earlier on in their journey right now?
248
Yeah, I think it's really two things.
249
They say it a lot, but don't die.
250
Like really, really do not die.
251
There are a lot of moments where it's just like we were pretty close to death, but it's that that scar tissue becomes burned insights that someone starting out doesn't have.
252
And the other thing is as part of that, like you can be the smartest person in the world, but I really do subscribe to the belief that you just gotta get stuff out there.
253
And if you aren't, you are doing a disservice to your team and your company and your investors.
254
And so if you aren't dying, and you're getting in front of real people and building real product for real people, your success is guaranteed.
255
Like there's actually no way, given a long time horizon that you fail.
256
And so we've seen the good and the bad and the ugly of that and I'm excited for other people, especially building consumer while outside of WFOS to build some cool stuff for real people.
257
Nice, man.
258
Appreciate you.
259
All right.
📺 Cùng kênh
✨ Video gợi ý
B22
108 câu
We made YOUR Dream Phone
Business EnglishUniversity StudentsWorking Adults
B21
220 câu
I Copied A Huge App. Now I Make $80K/Month
Business EnglishUniversity StudentsWorking Adults
B24
166 câu
Can This Robot Be My Personal Assistant
Business EnglishDaily ConversationHigh School Students
B21
259 câu
Silicon Valley | Season 1-5 | The Best of Gilfoyle
Business EnglishUniversity StudentsWorking Adults
B21
232 câu
Life Lessons From The World’s Youngest Self-Made Woman Billionaire
Business EnglishUniversity StudentsWorking Adults
B22
43 câu
AI is changing the world, but is it also trying to kill us? @RonnyChieng investigates #DailyShow
Business EnglishDaily ConversationUniversity Students
B21
142 câu
How Americans see EU Tech
Business EnglishUniversity StudentsWorking Adults
B21
380 câu
You Know What You Should Do… So Why Aren't You Doing It
Business EnglishUniversity StudentsWorking Adults
B22
477 câu
Tom Holland on Going Sober & Building a Brand That Matters
Business EnglishDaily ConversationUniversity Students
B22
278 câu
AI Food Slop is Here
Business EnglishUniversity StudentsWorking Adults
B21
201 câu
Full Video: Anna Chapman on her startup Property Finder
Business EnglishInterview EnglishUniversity Students
Từ vựng và ghi chú luyện nói cho bài học này
Bài luyện nói trình độ B2 này dựa trên video “24h Inside $5M Silicon Valley AI Startup Building Personal Super Intelligence”. Video này có 258 câu và 3220 từ để luyện shadowing. Phần lời nói dài 17:05. Người nói nói nhanh, khoảng 189 từ mỗi phút, nên sẽ có nhiều chỗ nối âm và nuốt âm. 87% số từ nằm trong 3.000 từ tiếng Anh thông dụng nhất, nên từ vựng dễ theo.
Từ vựng quan trọng trong video
15 từ khó nhất trong video, kèm phiên âm và nghĩa:
| Từ | Phiên âm | Nghĩa |
|---|---|---|
| calorie danh từ | /ˈkæləɹi/ | ca-lo, ca-lo-ri |
| bake động từ | /beɪk/ | nướng |
| sharpness danh từ | độ nét | |
| motivate động từ | /ˈmoʊ.tɪ.veɪt/ | khích lệ, động viên |
| stack danh từ | /stæk/ | chồng |
| ceiling danh từ | /ˈsilɪŋ/ | trần, trần nhà |
| scenario danh từ | /sɪˈnɛəɹioʊ/ | trường hợp, kịch bản |
| backspace danh từ | /ˈbækˌspeɪs/ | phím lùi |
| litmus danh từ | /ˈlɪtməs/ | quỳ |
| salesperson danh từ | người bán hàng | |
| maze danh từ | /ˈmeɪz/ | mê cung |
| scar danh từ | /skɑɹ/ | sẹo, thẹo |
| sticky tính từ | /ˈstɪki/ | dính |
| luckily trạng từ | /ˈlʌk.ɪ.li/ | may, may mắn |
| hobby danh từ | /ˈhɑ.bi/ | sở thích riêng, sở thích |
Cụm động từ bạn sẽ nghe
| Từ | Nghĩa |
|---|---|
| give up động từ | đầu hàng |
| check out động từ | tính tiền, trả phòng |
| come back động từ | trở lại, trở về |
| sit up động từ | ngồi dậy |
Những câu đáng luyện lại
Các câu ngắn, trọn ý trong video mà bạn có thể dùng lại khi giao tiếp hằng ngày:
- That's our definition of personal super intelligence.
- That's the feedback that we've seen.
- What are we doing now?
- I've been waiting for this.
- Run me through, what's the setup here?
Ngữ pháp trong video
Những cấu trúc người nói dùng nhiều nhất, kèm đúng cụm từ trong video:
| Cấu trúc | Trong video |
|---|---|
| Thì hiện tại hoàn thành tiếp diễn have/has been + -ing — việc bắt đầu từ trước và vẫn đang tiếp tục | we've been working · I've been looking · I've been waiting |
| Thì hiện tại hoàn thành have/has + quá khứ phân từ — việc đã xảy ra nhưng còn liên quan đến hiện tại | have come · They've just raised · They've raised |
| Mệnh đề quan hệ who / which + mệnh đề — thêm thông tin về người hoặc vật | people, which is · benefit which is · adulthood which is |
| Câu bị động be + quá khứ phân từ — nhấn vào việc xảy ra, không phải người làm | is based · is guaranteed |
Phát âm cần chú ý
Người nói dùng 79 dạng rút gọn, ví dụ I'm, we're, don't. Hãy nói theo dạng ngắn đúng như bạn nghe.
- Âm “sh” và “zh”: actionable /ˈæk.ʃə.nə.bəl/, expiration /ˌɛk.spɚˈeɪ̯.ʃən/, realization /ˌɹɪə.laɪˈzeɪ.ʃən/, indication /ɪndɪˈkeɪʃən/, recommendation /ˌɹɛkəmɛnˈdeɪʃən/
- Từ dài — đặt trọng âm cho đúng: actionable /ˈæk.ʃə.nə.bəl/, defensible /dɪˈfɛn.sɪ.bəl/, expiration /ˌɛk.spɚˈeɪ̯.ʃən/, suboptimal /sʌbˈɑptɪməl/, unequivocally /ˌʌnɪˈkwɪvək(ə)li/
Những âm người Việt thường gặp khó:
- Cụm phụ âm cuối — đọc đủ từng âm, đừng nuốt âm cuối: clearance /ˈklɪɚ.əns/, preference /ˈpɹɛf(ə)ɹ(ə)ns/, oftentimes /ˈɔf.ənˌtaɪmz/, implement /ˈɪmpləmənt/
- Âm /tʃ/, /dʒ/ và /ʒ/ — tiếng Việt không có, đừng đọc thành “ch” hay “gi”: widget /ˈwɪd͡ʒ.ɪt/, badge /bæd͡ʒ/
- Âm /æ/ — mở miệng rộng hơn âm “e”: calorie /ˈkæləɹi/, stack /stæk/, actionable /ˈæk.ʃə.nə.bəl/, backspace /ˈbækˌspeɪs/, proactive /pɹoʊˈæk.tɪv/
Cách luyện với video này
- Nghe hết video một lần, chưa cần nói, và ghi lại những từ bạn chưa biết.
- Bắt đầu ở tốc độ 0,75×, nói đuổi từng câu, rồi quay lại tốc độ bình thường khi đã quen.
- Ghi âm giọng mình rồi so với bản gốc, chú ý các từ như calorie, bake, sharpness.
Phương Pháp Shadowing Là Gì?
Shadowing là kỹ thuật học ngôn ngữ có cơ sở khoa học, ban đầu được phát triển cho chương trình đào tạo phiên dịch viên chuyên nghiệp và được phổ biến rộng rãi bởi nhà đa ngôn ngữ học Dr. Alexander Arguelles. Nguyên lý cốt lõi đơn giản nhưng cực kỳ hiệu quả: bạn nghe tiếng Anh của người bản xứ và lặp lại to ngay lập tức — như một "cái bóng" (shadow) đuổi theo người nói với độ trễ chỉ 1–2 giây. Khác với luyện ngữ pháp hay học từ vựng bị động, Shadowing buộc não bộ và cơ miệng phải đồng thời xử lý và tái tạo ngôn ngữ thực tế. Các nghiên cứu khoa học xác nhận phương pháp này cải thiện đáng kể phát âm, ngữ điệu, nhịp điệu, nối âm, kỹ năng nghe và độ lưu loát khi nói — đặc biệt hiệu quả cho người luyện IELTS Speaking và muốn giao tiếp tiếng Anh tự nhiên như người bản ngữ.