Pratique du Shadowing: Garry Tan: Own Your Intelligence - Apprendre l'anglais à l'oral avec la vidéo
Choisissez un mode d'entraînement
Chaque mode ouvre une salle d'entraînement en plein écran.
565 phrases
Si les phrases sont trop courtes ou trop longues, cliquez sur Edit pour les ajuster.
1
So, the internet calls me one of the most AI psychotic people online.
2
So, it's only right that I start my talk with a story about one of the most cancelled men in history.
3
His name was Baruch Spinoza.
4
And in case the philosophy elective wasn't your thing, here are the highlights you need to know.
5
In 1929, a New York rabbi challenged Einstein by telegram.
6
Do you believe in God?
7
Answer in 50 words.
8
Einstein answered in 25.
9
I believe in Spinoza's God who reveals himself in the lawful harmony of the world,
10
not in a God who concerns himself with the fate and doings of mankind.
11
The most famous scientist alive asked the biggest question there is, pointed at Spinoza.
12
But here's what Baruch Spinoza's own community did to him.
13
Amsterdam, July 27th, 1656.
14
Spinoza is 23 years old, a member of a tight-knit Sephardic Jewish community.
15
He stands in a synagogue while the elders excommunicate him with the most violent curse the community ever produced.
16
Cursed be he by day and cursed be he by night.
17
Cursed be he when he lies down and cursed be he
18
when he rises up nobody may speak to him nobody may
19
trade with him nobody may come within four cubits of him nobody may read anything he writes and this ban
20
uniquely among the roughly 40 bands spinoza's community issued that century has no repentance clause.
21
It has never been lifted.
22
It technically is still in force today.
23
Spinoza was 23.
24
His crime was evil opinions, expressing forbidden thoughts.
25
His punishment was complete deletion from the community.
26
Before his community cursed him, they tried to buy him a thousand guilders a year, Serious money.
27
All he had to do was show up at synagogue once in a while and keep his mouth shut.
28
Hear that in founder terms.
29
They've offered him a salary to stop building.
30
He said no. Not for $10,000, he said.
31
He wanted truth, not comfort.
32
Shortly before his excommunication, a fanatic came at him with a knife.
33
The blade tore through his cloak and missed him.
34
He kept that cloak, scar unmended, for the rest of his life.
35
He wanted to remember what ideas cost.
36
So what does the most canceled man of the 17th century do next?
37
He grinds lenses.
38
By day, he makes optical instruments, tools that let human beings see further than their eyes allow.
39
He makes them so well that the best scientists in Europe seek them out.
40
And by night, he writes a book so dangerous, he cannot publish it while he is alive.
41
When he dies at 44, lungs full of glass dust from making other people's lenses, the manuscript is locked in his writing desk.
42
His dying instruction?
43
Ship the desk by canal barge to his publisher in Amsterdam.
44
That manuscript became his posthumous works, which attracted immediate attention across Europe and inspired some of the most important philosophers of the Enlightened.
45
What does Spinoza have to do with startups, you might ask?
46
Well, here is a man, canceled by everyone he knew, offered a salary to stop,
47
nearly killed for shipping, and his response was to build precision tools by day
48
and write the most dangerous book in Europe by night alone, with no permission from anybody.
49
If you're going to start a startup, you could do well to learn from Spinoza.
50
He had a name for the engine that kept him going.
51
Kanatus.
52
It means your striving.
53
The drive in every living thing to keep going and to increase its power to act.
54
Not your resume.
55
Not your title.
56
Not your job.
57
The striving itself.
58
This talk is about the tools that amplify it.
59
So what did he actually say that was worth deleting a man over?
60
The gist of the heresy was that God is not a king on a throne.
61
God is spread through everything that exists.
62
God or nature, he wrote.
63
400 years later, we are making a similar mistake about intelligence.
64
Everyone is waiting for AGI as a singular event, a god in a data center,
65
some announcement, some threshold, some day when the sky changes color.
66
So now I'll say a version of Spinoza's heresy, updated 400 years later.
67
Everyone is watching the sky, and the thing they're watching for is already in the room.
68
It doesn't look like a god.
69
It looks like infrastructure, a terminal window, a folder of markdown files, a job that finishes while you sleep.
70
Spread through everything, which is exactly where Sminosa told you to look.
71
AGI isn't arriving as an event.
72
It's arriving diffused as your agent, running on your context, doing your work.
73
I call it personal AGI, not artificial general intelligence for everyone all at once, general intelligence for one person, you.
74
This was a dream of a great many people.
75
Vannevar Bush called it the memex, a machine that would be an extension of yourself and your brain.
76
And I want to be precise about what I mean, because the words personal AI has already been captured by marketing departments.
77
I do not mean a chatbot you pay $20 a month to.
78
I do not mean a slightly better autocomplete.
79
I do not mean an assistant that knows your calendar and nothing else.
80
That's just a subscription you rent.
81
It's a corporate AGI you don't own.
82
It resets when you close the tab.
83
It knows what everyone else already knows.
84
and when the company behind it pivots, your so-called assistant gets a lobotomy on someone else's schedule.
85
Personal AGI is a different animal.
86
An agent that runs on your infrastructure, reads from a memory you own, executes procedures you wrote, and compounds.
87
The corporate AGI that you don't own gets better only when the company ships something.
88
Your personal AGI gets better every single day you use it because every day it knows more of your life.
89
One of these is a product you consume.
90
The other is an asset you build.
91
Almost nobody in the world has this second thing yet.
92
And everyone in this arena could have it by Monday.
93
And I believe intelligence, intelligence of this kind should be owned, owned by you, not rented.
94
If you go forth and build this for yourself, if 2034 doesn't have to be like 1984.
95
You might ask why this personal AGI is happening only now.
96
Well, I think it's because of what agents can do, and nowhere is it more obvious than encoding agents.
97
In 2013, I was a YC partner building Bookface, our internal social network, at night.
98
I shipped maybe 14 useful lines of code a day, which, if you know the literature on programmer productivity, is dead on median.
99
That was me at full effort.
100
This year, I run YC full-time, same brain, same hours, plus a five o'clock kid pickup.
101
I did the math on my output, and I'm at about 400x what I did in 2013.
102
Now, before the skeptic in row three deflates that number for me, let me deflate that for myself.
103
You don't trust the raw lines of code?
104
Fine.
105
Apply the most pathological verbosity penalty you can stomach and assume the agent writes bloated code.
106
assume half of it is scaffolding.
107
Assume I'm flattering myself, which is always a live possibility.
108
It's still 8x at the absolute floor and 10 times that in the middle of the range.
109
The number is large no matter how you torture it.
110
Now, this is just code, and if you're at the beginning of your career, you're in luck.
111
This applies to design.
112
This applies to product management.
113
This applies to growth.
114
This applies to every part of what you might want to do.
115
The multiplier for coding is not just for coding.
116
It's for every piece of knowledge work.
117
And it's not just me.
118
At YC, we get to watch this at portfolio scale.
119
A year and a half ago in the winter 25 batch, a quarter of the companies had code bases that were 95% AI generated.
120
Those companies use AI agents for everything now, not just code.
121
And that batch is on track to becoming one of the fastest growing, most profitable batches in the history of YC.
122
Now, I know what a correlation is, so let me say it carefully.
123
I cannot prove that the AI generated code and everything else caused the growth.
124
But what I can tell you is that the fastest growing founders we fund are not treating AI as autocomplete.
125
They are treating it as a workforce.
126
There are 2x people and there are 100x people who are using the same clod, same weights, same context window size, same API.
127
But the leverage is not in the weights.
128
It's in what context you give it, how relevant it is, and does it happen at the right step.
129
We'll come back to this.
130
Now, Spinoza has a definition I think about every single week.
131
In ethics, he defines joy as the feeling of your power of acting increasing,
132
which is why the first time an agent does a week of your work in an afternoon, it doesn't feel like a convenience.
133
It feels like joy.
134
And that's not me being poetic.
135
That's the technical term.
136
Your power of acting increased.
137
Your canatus just got bigger.
138
He defined the opposite to, by the way, sadness, the feeling of your power of acting decreasing.
139
If your Sunday nights have a specific heaviness, like your ability to influence the world is receding, that you feel like you're quiet quitting,
140
then this is what you feel and hold that thought
141
because we're coming back to this in the second half of this talk and it gets political.
142
So here's the equation for the next decade of your life, a frontier model, which is rented and a commodity and getting cheaper by the quarter,
143
plus your context, which is owned by you and unique, and ideally nobody else on this earth has it,
144
plus a harness that wires them together.
145
That harness might be open claw, Hermes agent, clawed code, or codex.
146
Add that up, and that gives you an agent that acts like a very fast version of you.
147
Model quality is rented, but your brain is owned ideally by you.
148
Marshall McLuhan said that technology is an extension of man.
149
Steve Jobs called a computer a bicycle for the mind.
150
And if you have what I'm describing here, then you have a self-driving rocket.
151
Paul Graham taught every founder in this building two things, make something people want and do things that don't scale.
152
Both still govern everything.
153
What's new is the multiplier on the second one.
154
Agents are how one founder now does unscalable things at scale.
155
The advice didn't change, but the physics of all startups and of what you can do did.
156
Spinoza ground lenses, instruments that let people see past the limits of their eyes.
157
I want to spend the next 15 minutes showing you what grinding lenses for the mind looks like.
158
This is the machinery I actually run my life on and every concept travels to whatever stack you use.
159
Let's start with working memory because it explains everything.
160
You and I as human beings hold about seven things in our head at once.
161
Seven plus or minus two.
162
It's the most famous paper in cognitive psychology.
163
It's why local phone numbers are seven digits and why you forget the eighth item on a grocery list.
164
That is the entire working memory of a human being and every institution humanity has ever built.
165
Every checklist, every org chart, every filing cabinet, every stand-up meeting is a prosthetic for that limit.
166
An AI agent, though, holds a million tokens.
167
That's about a thousand pages.
168
Three Harry Potter books sitting open on its head all at once,
169
and it can find a needle in any of them and synthesize across all three in seconds.
170
Three Harry Potter books versus seven digits.
171
You could argue that's not quite AGI yet, but it is already a different operating regime.
172
And almost everyone on Earth is still running their life on an org chart
173
and a way of doing things designed for the seven-digit brain.
174
Run that number in the other direction.
175
A thousand pages is a lot, but it is also very little.
176
Your life is not three books.
177
Your life is a library.
178
Every email you ever sent, every meeting, every decision, and every reason behind it, every conversation with every person you know.
179
The question that determines whether your agent is a genius or a goldfish is this.
180
Who decides or what decides which three books are open on the desk.
181
And that's what a brain is.
182
That's what G-brain is meant to be.
183
The library plus the librarian.
184
I've been building G-brain in the open.
185
My personal open claw has a Karpathy-style knowledge wiki with about 220,000 markdown pages.
186
25 years of my life, diarized.
187
Every email, every meeting, my notes, my photos, my drafts, the things I got wrong.
188
Compiled mostly by agents, curated by agents, searched for by agents, but I never re-ask a question I already answered.
189
And the lived experience in the system is the point.
190
A founder emails me about a crisis before I finish reading the email.
191
My agent has already pulled every prior conversation I've had with that founder, three portfolio companies that hit the same wall, and what actually worked for them.
192
When my agent does anything, it does knowing everything I know.
193
And that's the difference between an assistant and a colleague.
194
Let me walk you through an actual day because that matters more than an architecture diagram.
195
While I slept last night, my agent processed my inbox.
196
Not sorted it, processed it.
197
It knows which emails are from founders in trouble, which are from people trying to sell me something, and which are from the 17 mailing lists I never quite unsubscribe from.
198
The ones that matter are triaged with context pulled from the library, who this person is, my whole history with them, what they're really asking under what they wrote,
199
and what that might mean for me.
200
I wake up to a briefing, not a pile of emails.
201
Before every meeting, a prep doc, who I'm meeting, what we said last time, what changed since, and what I should ask.
202
Research I was curious about at midnight is finished by morning, and when something interesting happens in the world, my agent has usually read it, cross-referenced against what I care about,
203
and filed it before I've had coffee.
204
On top of this library sits my agentic coding framework, GStack, 123,000 stars now, which put it in the top 100 open source projects in the history of GitHub.
205
and what's actually in the punchline of this full architecture, it's mostly skill files, plus a browser that the agents can drive,
206
pages of English, and a way to act on the world.
207
Markdown, not magic.
208
Fat skills, thin harness.
209
Let me show you what a skill file is
210
because I keep saying this phrase and I want you to see how unmagical it is.
211
Here's a real one, lightly redact is.
212
It says, when a meeting recording lands from Circleback, transcribe it with speaker labels, pull out the commitment made, who made it, and the deadline.
213
Cross-check every person named against the library and link their pages.
214
File the summary here, full transcript there.
215
If anything contradicts something we already believe, flag it.
216
Don't overwrite it.
217
That's it.
218
That's a skill.
219
It's a page of English.
220
A smart intern, anyone really, who could read, could follow it.
221
And that's the test, actually.
222
If a smart intern could follow it, an agent can run it.
223
Which means, actually, a kind of profound thing.
224
I know I caught a lot of flack for talking about this, but I think it's more true than ever, especially now.
225
Markdown is actually code.
226
If you can write clear instructions in English, you're a programmer.
227
The compiler is a language model.
228
And that's why it's not just for engineers anymore.
229
At YC, our media people, events staff, finance team, people who never open a terminal in their lives, are building skill files and scheduled jobs.
230
One of our finance folks compiled about 100 Excel workbooks into a single app she built with an internal agent.
231
She's not a programmer.
232
She's a manager of agents now.
233
Everyone is about to be.
234
The most important question to ask here is, where is the computation happening?
235
And there are exactly two answers, and confusing them causes every agent failure I've ever seen.
236
Some computation belongs in latent space.
237
Taste, judgment, reading what a human actually wants from a vague request, that lives in the model, and you steer it with a markdown file.
238
And then some computation belongs in deterministic space, the arithmetic, the SQL query,
239
For instance, the seating chart for what sessions you're going to go to today for your breakouts.
240
All of that needs to be stored in a SQL database used by the markdown files.
241
Being smart about this goes a long way.
242
Ask an agent or human to seat five people around a table.
243
That's easy.
244
Do it in latent space.
245
Ask it to make custom schedules for 6,000 people in an arena like we just did for you
246
and your latent space agent needs to write some code to keep track of it.
247
Your experience at this conference had to be markdown files calling code in exactly this way.
248
And you couldn't do it without the code.
249
The model fails where we fail.
250
The fix is having the model compute the way humans compute.
251
The latent and the deterministic markdown files calling databases and scripts.
252
Simple, but it's what everything is actually built on.
253
And I'll give you one more receipt my favorite one because you're sitting inside it right now.
254
Five days ago, I decided this talk needed Spinoza, one of my favorite philosophers, especially because of how canceled he got.
255
So my agent went and acquired three of the best biographies about the man, books by Nadler, Goldstein, and Stewart, about 1,500 pages.
256
It read all three.
257
It built me a synthesis, a dated chronology of his life, every place the three biographers disagree with each other, and the best verbatim quotes with chapter citations.
258
And because it knows what I need, the 10 most tellable moments of his life ranked with delivery notes.
259
The knife attack, the bribe, the desk.
260
Every beat of our opening that might have given you some chills 20 minutes ago came out of that overnight run.
261
1500 pages became a stage ready story that I could edit.
262
I call it It's a compendium skill and I use it daily.
263
It's a personal skill that is a mega, mega version of deep research, only deeper than anything the corporate AI products will give you.
264
The spine of this talk you're watching was inspired by the machine we're describing now.
265
And if you're wondering where mine actually started, it was not 220,000 pages.
266
It was a folder.
267
It was a few markdown files about the companies I was working with and the people I kept emailing.
268
And the library got big the same way anything gets big.
269
A little every day compounding with agents doing the filing.
270
Nobody builds the warehouse first.
271
First you build one shelf.
272
When you sit down with an agent tonight, you're not coding.
273
You're managing a workforce made of markdown.
274
A skill file is an employee.
275
It has one capability, one job written down clearly enough that someone new could execute it.
276
A resolver is an org chart.
277
A task comes in and it decides which markdown file or who handles it.
278
Which means that before you ever incorporate anything, before you have a co-founder or a logo or a deck, you can already be running an organization.
279
An organization of one plus your agents.
280
You are the founder and the entire management layer of you incorporated
281
and the headcount under you is now whatever you decide it is.
282
This already produces companies that break the old math.
283
Emergent out of our summer 24 batch went from public launch to nine figures of revenue in eight months.
284
When they crossed $15 million in annualized revenue, they were 15 people.
285
Retail, Winter 24, hit $60 million annualized with about 40.
286
That revenue per person did not exist before.
287
Not in software, not in oil, not in railroads.
288
And these aren't freaks of nature.
289
They're the first companies built natively on the new physics, and every one of them started as one or two people wired the way I just described.
290
Now, picture our batch room in the dog patch.
291
Hundreds of founders.
292
Every single day, each one of them doing what used to be a person's entire year of work.
293
That is not the future.
294
That is the bar right now with this batch.
295
If you're not doing it, your competitor is, and they will eat your lunch politely and thank you for it.
296
It also changes what software even is.
297
Software doesn't have to be precious anymore.
298
You can build exactly the tool you need for the audience of one in a weekend.
299
The old advice was scratch your own itch and hope it's the market.
300
The new version is much better.
301
Scratch your own itch because scratching itches is nearly free.
302
And some of your tools for one will turn out to be entire companies.
303
You'll know because other people start begging for them.
304
And one honest caveat before the how-to because you catch me out in any way.
305
A brain nobody curates is a garbage dump with great search.
306
Retrieval will surface a stale fact with total confidence.
307
A bad skill file encodes a bad process forever.
308
So the primitive is memory plus hygiene, provenance on every fact, contradiction checks when new information collides with old,
309
and a librarian whose actual job is pruning.
310
Treat the brain like production infrastructure and it compounds.
311
Treat it like a dumping ground and you get a very confident agent that is wrong in ways nobody can trace.
312
Everything so far is philosophy and receipts.
313
So let's get into some how-to.
314
If you do what I describe in the next six minutes, you'll be ahead of 99% of people who watched this talk and just nodded.
315
Step one tonight, pick a harness and run an agent on your own machine machine.
316
I use OpenClaw and Hermes Agent with Gbrain.
317
A hosted version of this is at gbrain.io.
318
It's free.
319
Gbrain itself is free and open source.
320
I always recommend the Ferrari, but I'll be honest, the Honda is really good too.
321
Codex, Claude Code, whatever.
322
Any of them will do 99% of this, and the upside of not Ferrari is
323
that it will also get you to your destination with a
324
little less of less getting out to fix it on the side of the road.
325
The concepts are the point, not any given repo or product.
326
The intelligence is on tap and there are many paths.
327
Step two this weekend, start your library, not a grand archive.
328
One folder of markdown files, export your notes, export your email if you can.
329
Write one page about each project you're working on, each person you work with, and on those pages write the things you actually know,
330
what you're building together, what they care about what you owe them, what they said last time.
331
That's stuff no model on earth has because it only exists in your head.
332
And your head, as we established, only holds seven things.
333
The first time an agent answers a question using your context instead of the internets, you'll feel the click and you won't go back.
334
You are all sitting on five, ten years of your own history in one inbox or another.
335
That's your moat just lying there, unindexed, doing nothing.
336
The only gate between you and this entire architecture is probably 24 hours.
337
Step three, write your first skill file.
338
Picking it is easy.
339
You know, what's the task you do every single week that you hate the most?
340
Might be expense reports, meeting notes, the weekly status update, competitor research.
341
Explain it to your agent.
342
What do you want to do in plain English, the way you'd explain to a smart friend on their first day of a job, and then let it get it wrong.
343
If it gets it wrong, correct it.
344
Every rule, every exception, every O, and also put it in there, and it'll fix it.
345
That page is now an employee.
346
Run it.
347
Step four, wire it up to be a recurring job.
348
Maybe it's the job you just created in step three.
349
Every morning at seven, do this.
350
Every Friday, summarize that.
351
The first time you wake up to work that finished while you slept, something shifts in your head permanently.
352
That's the day that the day stops being the unit of work for you.
353
It becomes what you can imagine, and it should be driven by what your goals are and what you want to create in the world.
354
Step five, this is the discipline that separates the compounders from the dabblers.
355
Never do one-off work.
356
Most people run one operation with one agent and then throw the context away.
357
They close the window, that's it.
358
Don't.
359
At the end of every task, ask the agent to skillify what it did.
360
Skillify is a special skill you can find in Gbrain.
361
You can point it at that repo and say, extract Skillify, learn how to do it.
362
Turn it into a markdown file you can use and reuse forever.
363
I'll say it the way I say it at YC.
364
If you have to ask for something twice, you failed.
365
The person who captures what they learn gets smarter every single day.
366
The person who wakes up every morning with amnesia, well, that's a waste of your time.
367
And it sort of doesn't matter how good the model gets if you can't turn it into real memory.
368
Do those five things and I can tell you what your next 90 days look like.
369
Week one, honestly, it's a toy.
370
The library's thin.
371
The skills are clumsy.
372
You're fixing more than you're saving.
373
Week four, the flywheel catches.
374
The agent starts answering with your context.
375
The morning job produces something you actually read.
376
and you write your third and fourth skill because the first two worked.
377
Week 12, you have a library that answers before you finish asking.
378
A dozen skill files running the parts of your week you used to dread
379
and one or two tools that other people keep asking to borrow, which in this room is called a startup.
380
The curve is the same curve as any compounding thing you've ever seen.
381
Flat, flat flat then not most people who try this will quit this in week two
382
which is precisely why the ones who don't feel like they're
383
cheating by week 12 now i need to tell you the part
384
that isn't fun because everything i taught you just cuts both ways i told you spinoza's definition of sadness earlier.
385
The feeling of your power acting, power of acting, decreasing.
386
And I said it gets political.
387
This is where.
388
A skill file is not a document.
389
It's a piece of your cognition, how you do the thing.
390
Extracted from your head, written down, and executable.
391
Every skill you teach an agent is you, externalized.
392
And the exact same file is two opposite futures, depending on one variable who controls it.
393
Take a fictional example of a support engineer.
394
Let's call her Maya.
395
Over two years, Maya teaches her agents 40 skills.
396
How to triage a P0 at 2 in the morning.
397
How to de-escalate the customer who's about to churn.
398
How to write a post-mortem that actually prevents the next incident.
399
40 files.
400
That's her judgment, the thing that took her two years to build, sitting on a disk.
401
Version 1, those files live in Maya's repo.
402
She changes jobs, they go with her.
403
Day one at a new company, she's operating with years of compounded judgment on tap.
404
Every year she works, she compounds.
405
That's ownership.
406
And if she wanted to start a company that does this, it's her expertise.
407
And it turns out she can.
408
Entire startups these days will be Markdown files.
409
Version two.
410
Those files live in the company's repo.
411
Under the company's IT policy, Maya leaves with nothing.
412
The company keeps running her judgment without her.
413
40 files executing forever and her name isn't even in the commit history.
414
She didn't have a career.
415
She had an extraction.
416
Same files, same Maya, one variable.
417
So this is the doctrine and I want you to be able to repeat it tomorrow.
418
I believe skill files are yours.
419
Own your skills because if you don't, your job becomes a skill file.
420
And this happened before.
421
Craftsmen own own their tools.
422
That's what made them free.
423
The factory broke that.
424
The loom belonged to the mill.
425
The knowledge workers assumed we were safe because our tools lived in our heads where nobody could confiscate them.
426
Skill files end that.
427
For the first time in history, your cognition can be extracted, stored, versioned, and owned.
428
The only question is, by whom?
429
Do you remember the thousand guilders?
430
That offer never went away.
431
It got rebranded.
432
Every comfortable arrangement where your judgment compounds in someone else's repo is a thousand guilders a year to show up, keep quiet, and stop building your own thing.
433
And that's why you should start a startup, because this is how you can actually make those skill files work for you.
434
Spinoza faced the upgraded version two.
435
In 1673, Heidelberg offered the cursed heretic a full professorship,
436
salary, legitimacy, a chair, and,
437
quote, freedom of philosophizing, provided he not disturbed the established religion.
438
His answer was, I do not know what the limits of that freedom of philosophizing might have to be.
439
He read the terms of service and he declined the acquisition.
440
He had a phrase for what he was protecting, under your own power, as opposed to under someone else's.
441
Your power of acting exists either way.
442
The political question in 1673 and in 2026 is who commands it.
443
Personal AI is about controlling your own cognitive abilities and protecting yourself.
444
That's the whole thesis of this talk in one sentence.
445
Personal AGI is how you stay under your own power in the age of agents.
446
So keep your brain
447
and your skills in a repo you control from day one before any platform or any acquirer has an opinion about it.
448
When Spinoza died, they inventoried the room.
449
Two pairs of pants, seven shirts, a lens lathe, 160 books.
450
And the ethics locked in a desk.
451
He owned almost nothing and nobody ever controlled his skill files.
452
The desk drawer was his repo.
453
Own yours like he owned his.
454
Now three objections and I can hear them from up here so let's just do them.
455
Objection one, the models are improving so fast that all this harness stuff will be obsolete.
456
Just wait for the next release.
457
This is the bitter lesson crowd, and I love them, but notice what actually happens in every model release.
458
The better the models get, the more the differentiator moves to context.
459
When everyone's engine is a thousand horsepower, the race is won on the driver and the map.
460
The weights are everyone's.
461
The library is yours.
462
At least I hope it is.
463
A better model makes your library worth more because a smarter reader extracts more from the same books.
464
I'm rooting for the labs as hard as anyone in this building, but every release they ship is a free upgrade to a workforce I already own
465
and a workforce I want you to own.
466
Objection two, is this just rag?
467
Sure, and Postgres is just bee trees.
468
Retrieval is the primitive, not the product.
469
The hard part is everything around it.
470
What gets written down in the first place, how it gets enriched and linked, What gets promoted to hot memory versus filed as cold reference?
471
Who arbitrates when two facts disagree?
472
Retrieval is easy.
473
Being worth retrieving from is the product.
474
Objection three, and it's the one that deserves the most respect.
475
You put your entire life in one system, your email, your meetings, your kids' schedules.
476
What happens when it leaks?
477
My answer is the same answer as the whole talk.
478
That's exactly why it has to be yours.
479
My brain runs on my own infra, in my own repo, under my own keys.
480
Compare that to the default, which is not privacy.
481
The default is your life is already scattered across 10 clouds owned by companies whose incentives are not yours, searchable by everyone except you.
482
I didn't create the risk by consolidating my context.
483
I took custody of it.
484
Custody is the security model.
485
And if you don't trust yourself to hold the keys, I promise you the answer isn't trusting someone else's terms of service more.
486
So why did I open source all of it?
487
The harness, the brain architecture, the skills, the whole personal operating system.
488
People ask me this because they seem like they think there must be a catch.
489
Well, the answer is because I can.
490
Because being at YC for me means I don't have to monetize my own infrastructure.
491
But because I can is also the answer to the wrong question.
492
The real question is why anyone should.
493
And the answer is that I believe tools of the powerful should be given away.
494
Every era has a private technology of leverage, a thing the powerful have and everyone else doesn't.
495
For a long time, it was literacy.
496
Then it was capital.
497
Right now, today, it's this, the harness, the library, the workforce made of Markdown.
498
The people who have it are quietly operating at a different scale than the people who don't.
499
And the gap is widening every month.
500
And that's what this whole conference is about.
501
to give you the power to be able to do it for yourself.
502
When something like that, that powerful stays private, you get a priesthood.
503
When it gets given away, you get a renaissance.
504
I know which one I want to live in, which means I get to do the thing that I actually believe in.
505
And I'll give it to you as a creed because it's the closest thing I have to one.
506
Say the things other people won't.
507
Fund the people other people won't.
508
Build the buildings other people won't.
509
Write and give away the code that other people won't.
510
Leave behind the institutions that other people won't.
511
And when you build in the open, you should know what's coming because Spinoza's story has one more chapter.
512
November 1676, Gottfried Liebnis.
513
The most glittering genius in Europe, silk stockings, a calculating machine in his luggage,
514
travels to The Hague to spend three days in an attic with the most hated man on the continent.
515
And then he spends the next 40 years lying about it.
516
Publicly, the visit was a few hours in passing.
517
Privately, his notes are crammed with obsessive commentary on Spinoza.
518
I love a small version of this weekly.
519
I say agents write most of my code now and the dunks arrive by lunch.
520
Then I look at what the loudest dunkers are actually shipping and it's agents all the way down.
521
So learn the pattern now because building in public guarantees you'll need it.
522
First they quote tweet you, then they git clone you.
523
The dunks are just the adoption curve announcing itself.
524
And I want to show you what this architecture looks like when it's pointed at the only thing that really matters.
525
I have a friend whose son has a rare form of epilepsy.
526
No lab, no grant, no permission.
527
He just went and you can just do things.
528
He built a repo of 80,000 markdown files, a brain for one small boy,
529
and pushed himself to the absolute edge of what humanity knows about his son's exact condition.
530
Every specialist visit, every paper, every seizure log, every drug interaction.
531
indexed and cross-linked and ready so that when a new doctor has an idea, he knows in minutes whether it's already been tried.
532
A father, a laptop, and a library.
533
That is personal AGI, not a benchmark, not a demo.
534
The entire architecture I've described tonight, the library, the librarian, the right three books open at the right moment,
535
aimed at the one thing one man loves the most in the world.
536
Nobody was coming to build that for him, so he built it.
537
And nobody is coming to build yours for you.
538
That's the good news.
539
Everything you were told you needed, the team, the funding, the permission, the credential, was a workaround for the fact
540
that one person could hold seven things in their head and work 16 hours a day.
541
That fact just expired.
542
you can fly now not metaphorically mechanically every problem where you
543
thought I wish I had this person I wish I could hire this person
544
but I can't get them you can every archive too big
545
to read every data set too gnarly to clean every ocean
546
you were told not to boil we can boil the ocean now I have a sentence I live by
547
and I want to leave it with you.
548
It's all made up, but you get to make it up.
549
Every institution in the world, including the one that read a curse over a 23-year-old in 1656, was made up by people no smarter than you.
550
The difference between you and every generation of founders before you is
551
that they had to recruit dozens of believers before they could build anything at all.
552
You need a laptop and a few years of your own history you're already sitting on.
553
There are about 7,000 people at this whole event, 7,000 kanadises, 7,000 strivings.
554
For most of history, almost all of that striving never got an audience.
555
It died waiting for funding, waiting for headcount waiting for permission, waiting for someone else to believe first.
556
The machinery I showed you tonight is the first technology I've ever seen that lets the striving go straight to work.
557
One person, no intermediaries, no permission.
558
I genuinely do not think the world understands yet what 7,000 people with
559
that kind of leverage walk out of a building and do.
560
Spinoza closed the ethics, the book that had to be smuggled out in a desk with nine words.
561
All things excellent are as difficult as they are rare.
562
The difficulty just collapsed.
563
The rarity is now up to you.
564
Go and build.
565
Thank you.
Télécharger l'application
Tout pour parler couramment
Prononciation IAChaque phrase notée
Phonétique (IPA)Maîtrisez chaque son
VocabulaireEnrichissez votre lexique
Jeu de vocabulaireApprenez en jouant
Même chaîne
B22
Dot Plots: How to Actually See What Your Users Are Doing
Daily ConversationWorking Adults
B25
How Pitching Investors is Different Than Pitching Customers - Michael Seibel
Business EnglishWorking Adults
C115
Alexandr Wang on what matters in an age of abundant intelligence
Daily ConversationUniversity StudentsWorking Adults
B24
Jensen Huang: The Mindset That Built NVIDIA
Daily ConversationUniversity StudentsWorking Adults
C1Alexandr Wang: “This is a Once-in-a-Civilization Opportunity”
Daily ConversationUniversity StudentsWorking Adults
B2How to Get Your First 10 Customers
Business EnglishWorking Adults
C1How to Build an AI-Native Services Company
Business EnglishUniversity StudentsWorking Adults
FOR YOU
Vidéo recommandée
C11
150 phrases
Anthropic just proved AI isn't getting better
Daily ConversationWorking Adults
C11
184 phrases
Chills Will Run Down Your Spine When You Hear this Story - Eric Weinstein
Daily ConversationWorking Adults
C12
174 phrases
This 1 Book Has Produced More Geniuses Than Any University in History
Daily ConversationUniversity StudentsWorking Adults
C11
654 phrases
Stuart & Cliffe Knechtle | Do Agnostics Just Want To Have Control? | Give Me an Answer
Daily ConversationUniversity StudentsWorking Adults
C1662 phrases
Daniel Ek: Life After Spotify, Broken Healthcare Incentives, Catching Disease Early & AI's Potential
Daily ConversationWorking Adults
![You’ll stop using ChatGPT after listening to this | Jonathan Pageau [ARC 2026]](https://img.youtube.com/vi/yZUuKzDQSsI/mqdefault.jpg)
C13
173 phrases
You’ll stop using ChatGPT after listening to this | Jonathan Pageau [ARC 2026]
Daily ConversationWorking Adults
C11
147 phrases
Way To Eradicate All Debt | Mr Robot
Daily ConversationWorking Adults
C11
140 phrases
Stop Playing Small. Go All In.
Daily ConversationUniversity StudentsWorking Adults
C13
130 phrases
On Suicide
Daily ConversationUniversity StudentsWorking Adults
B21
61 phrases
This is a crazy joke #comedy #relatable #shorts
Daily ConversationWorking Adults
B21
433 phrases
Sacha Baron Cohen on Why Democracy Isn't Compatible with Social Media | Pivot Conversations
Daily ConversationWorking Adults
C17
175 phrases
The Lost Art Of The Intellectual Life
Daily ConversationUniversity StudentsWorking Adults
Vocabulaire et conseils d’expression pour cette leçon
Cette vidéo contient 565 phrases et 6473 mots à répéter en shadowing. La partie parlée dure 41:57. Le locuteur parle à un rythme naturel d’environ 154 mots par minute, proche d’une conversation courante. Seuls 81 % des mots font partie des 3 000 mots les plus courants en anglais, le vocabulaire est donc exigeant.
Vocabulaire clé de cette vidéo
15 mots moins courants de la vidéo, avec leur prononciation et leur sens :
- skill /skɪl/ (nom) — habileté, compétence. A capacity to do something well; a technique, an ability, usually acquired or learned, as opposed to abilities that are regarded as innate.
- founder /ˈfaʊ̯n.dəː/ (nom) — fondateur, fondatrice. One who founds or establishes (a company, project, organisation, state, etc.).
- curse /kɝs/ (nom) — mauvais sort, malédiction. A supernatural detriment or hindrance; a bane.
- desk /dɛsk/ (nom) — bureau, pupitre. A table, frame, or case, in past centuries usually with a sloping top but now usually with a flat top, for the use of writers and readers. It often has a…
- harness /ˈhɑɹ.nəs/ (verbe) — harnacher. To place a harness on something; to tie up or restrain.
- architecture /ˈɑː.kɪˌtɛk.t͡ʃə/ (nom) — architecture. The art and science of designing and managing the construction of buildings and other structures, particularly if they are well proportioned and decorated.
- judgment /ˈd͡ʒʌd͡ʒ.mənt/ (nom) — jugement. The act of judging.
- workforce /ˈwɝk.foɹs/ (nom) — main-d'œuvre, personnel. All the workers employed by a specific organization or state, or on a specific project.
- batch /bæt͡ʃ/ (nom) — fournée. The quantity of bread or other baked goods baked at one time.
- chart /t͡ʃɑɹt/ (nom) — carte. A map illustrating the geography of a specific phenomenon.
- infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/ (nom) — infrastructure. An underlying base or foundation for a building, organization, or system.
- cancel /ˈkæn.sl̩/ (verbe) — annuler, résilier. To cross out something with lines etc.
- compound /ˈkɑmpaʊnd/ (nom) — composé. Anything made by combining several things.
- lens /lɛnz/ (nom) — lentille. An object, usually made of glass, that focuses or defocuses the light that passes through it.
- arrive /əˈɹaɪv/ (verbe) — arriver. To reach; to get to a certain place.
Les verbes à particule que vous entendrez
- do in (verbe) — tuer, zigouiller. To kill or end; to defeat.
- give away (verbe) — offrir, donner. To make a gift of (something).
- make up /ˌmeɪk ˈʌp/ (verbe) — constituer, composer. To constitute, to compose.
- write down (verbe) — noter, écrire. To produce or set (something) down in writing; to record something.
- ask for (verbe) — demander. To request; to order.
Prononciation à surveiller
Le locuteur utilise 83 contractions et formes réduites, comme you're, don't, doesn't. Prononcez-les sous leur forme courte, telles que vous les entendez.
- Les sons « th »: ethics /ˈɛθ.ɪks/, thin /ˈθɪn/, mathematics /mæθ(.ə)ˈmæt.ɪks/
- Les sons « sh » et « zh »: objection /əbˈd͡ʒɛkʃən/, extension /ɪkˈstɛnʃən/, institution /ˌɪn.stɪˈtʃuː.ʃən/, machinery /məˈʃiːnəɹi/, shortly /ˈʃɔɹtli/
- Mots longs — placez bien l’accent: architecture /ˈɑː.kɪˌtɛk.t͡ʃə/, infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, librarian /laɪˈbɹɛə̯ɹi.ən/, institution /ˌɪn.stɪˈtʃuː.ʃən/, humanity /hjuˈmænɪti/
Comment s’entraîner avec cette vidéo
- Écoutez la vidéo en entier une fois sans parler et notez les mots que vous ne connaissez pas.
- Commencez à la vitesse 0,75×, répétez phrase par phrase, puis revenez à la vitesse normale quand cela devient facile.
- Enregistrez-vous et comparez avec l’original, en faisant attention à des mots comme skill, founder, curse.
La grammaire de cette vidéo
Les structures que le locuteur utilise le plus, avec les mots exacts de la vidéo :
| Structure | Dans la vidéo |
|---|---|
| « Used to » used to + verbe — une habitude ou un état passé qui n’est plus vrai | used to be · used to dread |
| Present perfect have/has + participe passé — une action passée qui compte encore maintenant | has never been lifted · They've offered · has already been captured |
| Voix passive be + participe passé — l’accent est mis sur ce qui arrive, pas sur qui le fait | been lifted · is locked · is spread |
| Propositions relatives who / which + proposition — une précision sur une personne ou une chose | everything, which is · myself, which is · people who are |
Qu'est-ce que la technique du Shadowing ?
Le Shadowing est une technique d'apprentissage des langues fondée sur la science, développée à l'origine pour la formation des interprètes professionnels. Le principe est simple mais puissant : vous écoutez de l'anglais natif et le répétez immédiatement à voix haute — comme une ombre suivant le locuteur avec un décalage de 1 à 2 secondes. Les recherches montrent une amélioration significative de la précision de la prononciation, de l'intonation, du rythme, des liaisons, de la compréhension orale et de la fluidité.