3/31/2026 - Jack Dorsey’s Plan to Replace Corporate Hierarchy with AI Intelligence Layers

Description

Chapters

Transcript

0:00 campy.
0:02 Huh?
0:22 Meow.
0:26 We will be
0:34 happy.
1:13 What is
1:40 Best lead singer ever. You're damn
1:42 right. That champ Shannon, he's a hell
1:44 of a hell of a dog. Load the car and
1:48 write the note.
1:52 Tell Grab your bags and grab your coat.
1:58 Tell the ones that need to know
2:04 we are headed north.
2:12 One foot in and one foot back.
2:17 Well, it don't pay to live like that.
2:23 So, I cut the ties and jump the tracks.
2:28 For never for never to return.
2:37 Brooklyn, Brooklyn, take me in.
2:42 Are you where the shape I'm in?
2:48 My hands they shake, my head it spins.
2:53 Oh, Brooklyn, Brooklyn, take me in.
2:59 When at first I learned to speak,
3:04 I used all my words to fight
3:09 with him and her and you and me.
3:14 Yeah, it was just a waste of time.
3:22 It was such a waste of time.
3:30 That woman, she's got eyes that shine
3:35 like a pair of stolen polished eyes.
3:40 She has to dance. I satisfied.
3:46 I'll see you in the morning time.
3:53 Brooklyn, Brooklyn, take me in.
3:58 Are you aware the shape I'm in?
4:03 My hands they shake my head. It spins.
4:08 Oh, Brooklyn, Brooklyn, take me head.
4:14 Three words that became hard to say.
4:21 I am loving you
4:26 more than I am today.
4:31 Look at the things I do.
4:39 Brooklyn, Brooklyn, take me in.
4:43 Are you wear the shape I'm in?
4:48 My hands they shake. My head it spins.
4:53 Oh, Brooklyn, Brooklyn, Brooklyn, take
4:56 me in.
5:05 Oh, there's a lovely song from a
5:08 brothers. Hello, a brothers. What's
5:10 shaking? What's going down?
5:14 Wolfman Jack.
5:30 Tik Tok has characterized this as
5:33 performing live karaoke with a pop
5:35 tribute.
5:43 Welcome to live karaoke night here at
5:45 the AI learning lab.
5:51 It's dog karaoke.
6:27 Genie in a bottle is the rising star of
6:30 the night. Congratulations
6:35 in a westerly direct show.
6:42 This car is my train.
6:46 I've been driving. I've been wondering
6:51 what it is I'm running from again.
6:56 Feel like an 80y old man.
7:00 Holding on to 29.
7:04 And uphead on that horizon
7:08 is a California line.
7:18 Double hit up trucks car scaring a wide
7:20 load.
7:22 Refab house cut in half.
7:26 Cute little front door and two windows.
7:28 My love
7:30 ain't sure whether to cry or laugh.
7:34 You see, I broke a home up myself once
7:38 I stumbled to that dog.
7:42 I read a note about the dogs li
7:51 anymore.
7:54 Well, I've had enough.
7:57 All this freedom on my road. Sing along,
8:00 Chappie.
8:02 No, it's good with decision.
8:05 At least that what I've been told.
8:16 Hey, you know something? I realized this
8:19 sort of halfass janky ass dog singing
8:23 people singing same 17 songs thing.
8:30 Um,
8:32 when all the AI learning labs out there
8:34 are just AI avatars doing perfect
8:37 [ __ ] here's the latest news of the
8:40 day in artificial intelligence.
8:42 Everything's faster. Everyone's making
8:44 more money and more people are picking
8:47 blueberries. When they're all doing
8:49 that, just perfect.
8:51 I'll just be here being a human being.
8:54 You're like, "Oh yeah, that one's real."
8:56 Listen to the dog and the whatever is
8:59 that singing.
9:14 You're the authentic one. Ha ha. My name
9:17 is Shannon. I'm the authentic one.
9:26 Is his guitar out of tune? I don't think
9:28 it's ever been in tune.
9:40 Hey, if you're on YouTube, do we have
9:41 anyone on YouTube? YouTube or Twitter or
9:43 LinkedIn? Do make a comment for me.
9:46 Like, let me shake make sure my comments
9:47 are working. I know there's nothing to
9:49 really comment on yet. It's like, uh,
9:52 you know, it's night time, right? You
9:54 don't need to wear sunglasses.
9:59 >> Champion Kyle. Keeping it real.
10:01 Authenticity is the the big thing of the
10:04 day.
10:11 You and I here all alone.
10:16 Sunday morning
10:18 at home.
10:21 Skye's blue and a coffee is strong. It's
10:25 true.
10:29 And I open my eyes to a dream realized
10:34 in front of me.
10:38 And I had got a clue what in the world
10:40 is happening to me.
10:46 Think I think I'm happy.
10:49 Like first day summer vacation happy.
10:53 Got to get a little rest and relaxation.
10:56 Happy
10:58 like the choir on Sunday morning singing
11:00 true.
11:31 Yeah.
11:34 Yeah.
11:54 That's a seven and garunkle song.
12:05 Um.
12:11 Uh.
12:21 Uh. Oh, there's no comments coming in.
12:23 Anyone on YouTube? Is anyone on YouTube?
12:26 Can someone from Tik Tok hop over to
12:27 YouTube and post a comment?
12:29 >> There's one. We got someone from Silver
12:31 Fox. Oh,
12:32 >> hang on. Let me reload.
12:34 Oh, wait. I don't want to reload. Hang
12:36 on. I need to go away from it and come
12:38 back. There we go.
12:41 Hey, chief.
12:43 Okay, commenting. Okay, we've commented.
12:46 Comments are working now. I thought
12:47 something was weird. Something was
12:48 weird. It was my It was my window. I had
12:52 a window refresh issue. Yes, Mr. So,
12:57 >> it wasn't weird. It was a regular
13:01 Thank you. Thank you.
13:04 Oh, yes.
13:07 Yes. Yes. Yes.
13:10 It is the 31st
13:14 of March.
13:16 The quarter is complete.
13:56 Whatever time I see you now, get that
13:59 looking.
14:02 Every time I see your mouth, I hear that
14:06 smile.
14:09 The early misty morning light heard the
14:13 engine turning.
14:15 The old foot outside.
14:20 I was thinking about So the new Sunno
14:22 5.5
14:25 is supposed to do good at cloning your
14:28 voice.
14:29 So, I was thinking of doing that. I
14:30 don't know if I want to do it live
14:32 because I don't know, performance
14:34 anxiety. You know, what if I can't get
14:37 it musically up
14:44 and I just hose it and make a model on
14:46 my shitty singing voice.
14:50 That would be not fun, wouldn't it?
14:54 Scott Parker, hey there. Can't wait to
14:56 tell you how participating early on just
14:59 getting ready to change my life. Oh
15:02 yeah.
15:04 Scott Shioone Parker is a longtime
15:07 watcher and a rising star and you're
15:09 live.
15:11 If you're watching Kyle and Chappie here
15:13 Chappie here in the AI learning lab,
15:15 what are you doing with your life that's
15:16 more worthwhile?
15:21 Well, here's the thing.
15:24 If you if you come here expecting to get
15:26 anything from me other than maybe, you
15:28 know, a smarmy ass joke once in a while,
15:32 um,
15:35 that ain't what this is about. What this
15:37 is about is being in the conversation
15:39 with the people that come here
15:42 >> and understanding that um,
15:46 there's a lot of stuff to figure out.
15:51 We're We're
15:54 This next quarter is going to be pretty
15:56 intense. The first quarter was pretty
15:58 [ __ ] intense, right? The first three
16:00 three months of the year, we got We got
16:03 Opus 4.6 in December, which I know is
16:06 not this year. Shut up. I I I know how
16:09 to do months,
16:11 but we got Opus 4.6 in in December,
16:14 where all the programmers like, "Ah,
16:16 [ __ ] All right. It's good enough now to
16:19 do my job.
16:21 God damn it. I used to like coding. Now
16:24 I'm just babysitting agents. This sucks.
16:28 So that happened in December. Open Claw
16:30 comes out in January and it's like,
16:32 "Hey, you want your own little baby AI
16:35 that's going to hang out with you and do
16:37 shit?"
16:39 Everyone lost their [ __ ] mind. The
16:42 the sales of Mac minis went through the
16:44 roof.
16:47 Um,
16:53 you know, Seed Dance 2.0 came out, the
16:55 the Hollywood thing. Oh, did you see
16:57 that? Hollywood was like, "We're going
16:59 to sue you for copy." Hey, could we get
17:02 access to your little model there? We'll
17:05 we'll give you $2 million.
17:09 Oh, you'll take it? Okay, let's go.
17:11 That's awesome.
17:14 [ __ ] [ __ ] Hollywood. Everyone's up
17:17 in arms still. They're like, "Oh, we
17:18 could get a movie, too.
17:23 Joy Party's here." Joy Pertie's like,
17:25 "Yeah, I'm making movies. Anybody need
17:27 an AI filmmaker?" Hollywood's like, "Uh,
17:29 yeah, we do. We got the seed dance thing
17:32 here. Do you could you turn that into a
17:35 Hollywood blockbuster? We're not going
17:37 to tell Russell Crowe
17:40 if that's good with you."
17:42 No, we've already got his likeness.
17:44 We've got that from some other movies.
17:46 We're just put it Joy, if you could just
17:49 We're going to put you in the back room.
17:51 There's a What's it You There's still
17:54 some brooms in it, but if you could work
17:57 back there and just come come in the
17:59 back door with it. Don't drive a fancy
18:02 car. Okay.
18:05 In the basement, Joy Party. We We got We
18:08 We keep people like you in the basement.
18:11 We're calling it Hollywood 2.0.
18:17 I'll definitely come back soon. Got a
18:19 couple of things in soul procurement I
18:22 need to testify. I'll testify soon.
18:24 Okay, good. Beautiful.
18:29 There was some really interesting news
18:31 today. We'll get to I'm going to I'm
18:32 going to just blast through X tonight
18:34 because there's a bunch of a bunch of
18:35 really interesting stuff happened today.
19:26 The boxer.
19:34 Sitting in sitting in railway station
19:36 now taking a ticket to my destination.
19:54 Hey, how do I know if this channel's
19:56 real? Play the guitar.
20:02 Okay, it's real.
20:11 Chimpy,
20:13 you want to sing little A minor?
20:34 There was a single card
20:38 David played and it plays the Lord.
20:43 You don't really care for music, do you?
20:51 goes like this. The fourth fifth
20:56 round of fall lives
21:01 king.
21:03 Hallelujah.
21:08 Hallelujah.
21:12 Hallelujah.
21:17 Hallelujah.
21:22 Hello.
21:26 Yeah.
21:34 All right, let's get this show on the
21:35 road. Enough of that singing them sappy
21:38 songs. They're making me all sad. You
21:39 put me to sleep. Wake up, people.
21:45 We got some AI to do.
21:49 Good lord. All right,
21:52 friends pap. Hello friends. Just got in
21:56 from the AI documentary. It came out
21:58 Thursday. Highly recommend. Very nice.
22:01 Here we get education, community, group
22:03 learning, great conversation,
22:05 relationship building, creativity.
22:07 Exactly. That's why you come here.
22:10 I am just the I am just the carrier of
22:13 the community.
22:16 That's it. I show up. I show up. You
22:19 show up. That's what we do.
22:21 Lori Kay is a longtime watcher and a
22:24 rising star of the AI learning lab.
22:26 Congratulations, Lori. You're a rising
22:29 star. Do I get something? No.
22:37 I know, champ. It's exhausting.
22:40 I know. I'm tired of it, too.
22:45 You done?
22:48 What do you want? Cheese.
22:50 I said cheese. He wagged his tail.
22:54 Invites to AI learning lab. If you see a
22:57 guy playing guitar singing with a dog,
23:02 yeah, just scroll Tik Tok until you see
23:04 a guy singing with a dog.
23:09 All right. Um, let's get let's get
23:11 going. Let's get going. Let's get this
23:13 show on the road. I tell you what. I
23:15 tell you what. I TELL YOU WHAT. OH YEAH.
23:19 MHM.
23:22 I tell you what. I tell you what.
23:26 We'll put it. We're going to share a tab
23:28 in case there's some sort of audio we
23:30 want to listen to. All right.
23:33 So, we're on the Twitter.
23:37 It's technically called X. Uh Elon Musk.
23:40 Uh Elon Musk purchased Twitter uh for
23:43 $42 billion, of course, in a famous
23:45 transaction that he didn't want to
23:46 actually complete, but uh uh it was a
23:49 was a a failed negotiation. Anyway, $42
23:53 billion. It's now worth about 20 $18
23:56 billion, something in that neighborhood.
23:57 It's so he parlayed the 42 billion into
24:00 the 18 billion. That's it was a parlay.
24:02 It was what they call a reverse parlay,
24:04 a negative parlay. And uh and then but
24:06 but then what he what he wanted to do
24:09 was uh he didn't like the bird. Um I I
24:13 think the bird reminded him of his
24:15 childhood. I don't I don't know what it
24:16 was. And he liked the letter X uh
24:18 because of airplanes and so he named it
24:21 X. And so that but it used to be called
24:23 Twitter.
24:28 Uh what's going on over here? I don't
24:30 have enough
24:33 black bar on the bottom for the Twitter
24:35 people. Hang on people, I'll be with
24:38 you. Just everybody calm down.
24:42 Everybody calm down.
24:46 Fantastic. There you go. There you go,
24:49 Twitter people. That's a little more
24:50 better. You can see it.
25:01 Okay.
25:04 So, one of the things I wanted to show
25:06 you,
25:08 was it this one? Yes.
25:12 So,
25:14 apparently,
25:17 um, Claude Codes source code was leaked
25:22 yesterday.
25:24 Listen to this timeline. So this is
25:26 shrudy
25:28 uh on on uh on X. Anthropic leaked
25:32 512,000 lines of claude code source code
25:35 yesterday. What happened in the next
25:37 next 12 hours is absolutely wild.
25:42 4:00 a.m. Enthropic pushes an update to
25:44 npm. Inside the package, their entire
25:48 codebase, a 60 megabyte debugging file
25:52 accidentally bundled in. 23 minutes
25:55 later, researcher uh Chow ofen Shu spots
25:59 it, downloads the zip,
26:03 posts it on X. Within six hours, 3
26:07 million views. So by 10:00 a.m. there's
26:09 three million views of the uh of the
26:12 downloaded source code of Cloud Code. By
26:16 the time Anthropics team woke up, the
26:18 code was forked 41,000 times across
26:21 GitHub. Anthropics started firing DMCA
26:24 takedowns. Too late. A Korean developer
26:28 developer named Sigrid Jyn woke up with
26:31 his phone exploding. He's Claude Code's
26:33 biggest power user. Wall Street Journal
26:36 reported he burned through 25 billion
26:38 tokens last year. He read the leaked
26:40 code. He wrote he he rewrote the entire
26:44 thing in Python in eight hours. His repo
26:47 hit 30,000 stars, faster than any GitHub
26:50 project in history. Then he rewrote it
26:52 again in Rust. That version now has
26:55 49,000 stars. Someone mirrored it to a
26:58 decentralized platform with one message
27:00 will never be taken down. The code is
27:03 permanent. Enthropic cannot get it back.
27:09 Here's the part I can't stop thinking
27:11 about. Anthropic built in something
27:13 called undercover mode. Its only job,
27:16 prevent Claude from accidentally leaking
27:18 internal secrets. They shipped an entire
27:20 anti-ak system in their own product,
27:23 then leaked their own source code in a
27:26 map map file.
27:29 Uh oh, someone's unemployed. And in
27:31 fact, the the guy that leaked it posted
27:34 his uh I no longer work at Anthropic. Uh
27:38 here, Kevin Notton Jr., Kevin Notton Jr.
27:42 wrote, "I was fired from Anthropic
27:44 today.
27:46 I was the engineer responsible for
27:48 shipping the latest dev cloud code npm
27:51 package. Wanting to improve the
27:52 debugging experience for the team, I
27:55 decided to include some source maps in
27:57 the release. This resulted in the entire
27:59 internal codebase being publicly exposed
28:02 including thousands of files and every
28:05 agent command all system prompts the
28:08 complete query engine undercover mode
28:10 bypass permissions mode and our internal
28:13 telemetry configuration.
28:15 So so anthropic so anthropic is more
28:18 open source than open AAI right now. I
28:21 take full responsibility. I genuinely
28:23 believe the safeguards Claude code had
28:25 built in for me would be adequate and it
28:27 was a serious mis miscalculation on my
28:29 part. My actions have unintentionally
28:32 open sourced major parts of Claude's
28:34 architecture well ahead of schedule
28:37 although well ahead of schedule means
28:38 that they were planned for for being
28:40 open sourced. It sounds like I apologize
28:42 to the team and Claude. And then if you
28:44 go to his um his thing, it says recently
28:48 exanthropic looking for work.
28:53 Poor dude. Oopsie. Exactly. Why would
28:56 you share that with the world? Not great
28:59 for future hiring opportunities. Well, I
29:01 don't think he I don't think he intended
29:03 for it to get out. He But I don't I
29:05 don't know. I mean, clearly he clearly
29:08 it was a miscalculation. And here's the
29:10 deal, man. Developers are [ __ ]
29:11 brutal. If you if you make one mistake,
29:15 they will find you out,
29:19 you know. Crazy. It's crazy. I'll tell
29:23 you. It's crazy. I'll tell you. It's
29:25 crazy.
29:28 All right, let's see. A simpler tech. If
29:31 you could figure out HTML, you could
29:33 play
29:38 the href.
29:43 was the soul in ovation
29:48 that changed the world.
29:52 AI
29:54 is like a thousand
29:58 hrefs
29:59 a day.
30:11 Oh, you can't see this. Share this tab
30:13 instead. Okay.
30:17 Oh, here's another one. Okay. All right.
30:20 This was the other one I wanted to talk
30:21 about. Um,
30:27 so we y'all need a homebased data center
30:30 equivalent to Anthropics where we can
30:31 have our own cloud code at home. Well,
30:33 you're not I mean the the thing about
30:36 Claude Code's source being uh put out
30:39 there is that people are going to be
30:40 able to take open source models and make
30:42 them really good because Claude Code was
30:44 the thing that was it was the wrapper
30:47 around their models that was the good
30:49 part and that part's now been leaked.
30:53 It's amazing.
30:56 Okay.
31:02 So why use anthropic now? Well, that's
31:04 what we'll see. We'll see what the I
31:06 mean, I just have a feeling there's
31:09 going to be a whole bunch of Claude code
31:10 clones now that are based on Claude
31:12 code. And then what what will be cool is
31:15 there will be some developers that make
31:16 it better. So this will lead to
31:18 innovation, but crazy. Okay, Jack
31:20 Dorsey. Let's read the Jack Jack Dorsey
31:23 post.
31:24 It's long. You want to go in with me?
31:28 I'm not going to read that whole [ __ ]
31:29 thing, but we're going to read we're
31:30 going to read a lot of it.
31:33 So, Jack Dorsey, if you don't remember,
31:38 he um
31:40 he founded Twitter. He founded the
31:42 Twitter back when it had a little birdie
31:44 and it was blue. The only thing that's
31:47 blue that remains is the check mark.
31:50 Jack Dorsey founded Twitter and his most
31:55 recent company is called Block. I don't
31:57 even know what Block does, but Block has
31:59 10,000 employees.
32:01 No, Block had 10,000 employees. And
32:04 about two weeks ago, Jack Dorsey
32:08 fired
32:10 4,000 of his 10,000 employees and said,
32:13 "Here's the thing. This is not because
32:16 we're hurting for money. We are
32:17 profitable and we are growing.
32:20 We could have kept you on. We're
32:22 choosing not to because
32:25 our employees armed with intelligent AI
32:28 working in smaller flatter teams are
32:32 changing how we start and run companies.
32:34 That was one of the lines within
32:37 within his document firing those 4,000
32:40 people laying them off. And so today he
32:43 put out this thing. I think it was
32:45 today. Was it today? Jack, where's the
32:48 date? There's no date here.
32:52 article article 10 hours ago
32:57 at Sequoia we see the speed we see that
32:59 speed is the best predictor of startup
33:02 success most companies are focused on AI
33:05 AI as a productivity enhancer now I want
33:07 to show you something I want to show you
33:09 something
33:11 this oh you can't see it I got to change
33:14 how I'm sharing everybody calm down
33:17 everybody calm down. All right. Okay,
33:21 Kyle, we'll calm down. We're calm now.
33:23 Okay. All right. Oh, yeah. The seven
33:26 economies. You talked about that before.
33:29 Okay. Look, here's
33:31 here's Jack Dury. What does he say?
33:36 At Sequoia, we see speed as the best
33:39 predictor of startup success. Most
33:41 companies are focused on AI as a
33:44 productivity enhancer. See that right
33:46 there in the in the in the seven
33:49 economies. That's a economy. Number
33:51 three
33:53 is the efficiency economy. AI is cool.
33:56 It makes our business more efficient.
33:58 I'm going to take AI and I'm going to
34:00 use it to take what we currently do and
34:02 make it more efficient. Okay, that's his
34:05 his second sentence.
34:08 Few are focused on the potential of how
34:10 AI of AI to change how we work together.
34:14 few, right?
34:17 So, the transition economy are companies
34:20 that do see that possibility of AI to to
34:24 innovate things, but they're basically
34:26 going to keep things as they are.
34:28 But he says, "Block is showing is
34:31 showing what it looks like to
34:32 fundamentally rethink organization
34:34 design, ultimately harnessing AI to
34:37 increase speed as a compounding
34:39 competitive advantage. Block is showing
34:42 what it looks like to fundamentally
34:44 rethink organ organization design.
34:47 That's economy five, the aggressor
34:50 economy. Burn the org chart. Let's
34:52 reinvent ourselves. So this is his
34:55 manifesto on doing that.
34:58 2,000 years before the first corporate
35:00 org chart, the Roman army solved a
35:02 problem a very large organization still
35:04 faces. How do you coordinate thousands
35:05 of people across vast distances with
35:07 limited communication?
35:10 Their answer was a nested hierarchy
35:12 within a const a consistent span of
35:14 control at every level. The smallest
35:16 unit was the uh conternium
35:21 conternium
35:23 eight soldiers who shared a tent
35:25 equipment and a mule. Remember when I
35:27 talked about groups of six to eight
35:29 people, cohorts within the AI salon?
35:33 I talked about we should we should find
35:36 ways to create persistent groups of six
35:38 to eight people collaborating and
35:41 working together. We should experiment
35:42 with that in here. Maybe we'll kick that
35:44 off. So in about well so so here's the
35:47 thing. April 30th marks the 3r year
35:50 anniversary of AI Learning Lab going
35:52 live on Tik Tok. And for the past year
35:55 it's been live on Tik Tok and YouTube.
35:58 We're going to be shifting this inside
36:00 the AI salon because I want to get back
36:03 to the AI salon being this this really
36:06 vibrant collaborative place. Maybe we
36:08 play around with this structure in our
36:10 community. There's something here.
36:12 Anyway,
36:14 eight soldiers who shared a tent,
36:15 equipment, a mule led by a danis. I
36:19 don't know what a danis is. 10 uh
36:21 contraburnia formed a century of 80 men
36:25 under a centurion. Six centuries made a
36:28 cohort. 10 cohorts made a legion of
36:30 roughly 5,000. At each layer, a named
36:33 commander held defined authority,
36:36 aggregated information from below, and
36:38 relayed decisions from above. The
36:40 structure 8 to 80 to 480 to 5,000 was an
36:44 information routing protocol built on a
36:47 simple human limitation. A leader can
36:49 effectively manage somewhere between
36:51 three and eight people. Romans
36:54 discovered this through centuries of
36:56 warfare. Even today, the US Army's
36:58 hierarchical chain follows a similar
37:00 pattern, what we now call span of
37:03 control, and it remains a governing
37:05 constraint of every large organization
37:06 on Earth. The next big change came from
37:09 Prussia. After Napoleon's army destroyed
37:11 the Prussian forces at the battle of
37:13 Jenna in 1806, a group of reformers led
37:16 by Scarnhorst and
37:19 Jezanau Jizenu Jizanu
37:22 Nissa Nissan now I don't know rebuilt
37:26 military around an uncomfortable truth.
37:28 You cannot depend on individual genius
37:30 at the top. You need a system. He
37:33 created the general staff, a dedicated
37:36 class of trained officers who job whose
37:38 job was not to fight but to plan
37:40 operation operations process
37:43 information, coordinate across units.
37:45 Scarnhorst uh intended these staff
37:48 officers to support incompetent generals
37:51 providing the talents that might
37:52 otherwise be wanting among leaders and
37:54 commanders. This was middle management
37:57 before the term existed. professionals
37:59 whose purpose was to route information,
38:02 precomputee decisions, maintain
38:04 alignment across complex organizations.
38:06 The military also formalized the distin
38:09 distinction between line and staff
38:11 functions. Line advances the core
38:14 mission. Staff provides specialized
38:17 support. Every corporation still uses
38:19 this vocabulary today.
38:22 Military hierarchy. Anyone saying
38:24 anything? So, we all need a homebased
38:26 data set. Okay. The sort of thing that
38:29 prevents elite capture many people were
38:31 afraid of.
38:34 All right.
38:38 Roman Empire.
38:42 That small group dynamic sounds
38:43 interesting, Kyle, doesn't it, Cam? So,
38:45 so the basic idea was thinking about
38:49 this is this is inspired by um in in
38:52 content evolution. I run this group
38:54 called Collab and it's it's evolved into
38:58 being about eight of us that show up on
38:59 a regular basis, seven or eight of us.
39:02 And like we've been working together now
39:04 for three years and like we know each
39:06 other. We know each other's businesses
39:08 are, but we don't we don't collaborate
39:09 in our businesses necessarily, but when
39:12 we come together, we can talk about
39:14 stuff and get stuff done and pay
39:15 attention to the news and and prototype
39:18 stuff if we need to. And I just thought,
39:21 wouldn't that be a fascinating? And we
39:23 we kind of have that informally in the
39:25 salon right now. We've got little groups
39:27 of of, you know, irregulars and things
39:30 like that that that pile up and and
39:32 travel around in packs. What if we
39:35 formalized that as as a as a kind of
39:37 unit that you could sign up for a cohort
39:40 and and just meet with them on a regular
39:42 basis? I don't know. I think there might
39:44 be something there. Anyway, military
39:46 hierarchy entered the business world
39:48 through the American railroads in the
39:49 1840s and50s. The US Army lent um West
39:53 Point trained engineers to private
39:55 railroad companies. These officers
39:56 brought military organizational thinking
39:58 with them. Staff and line hierarchies,
40:01 divisional structure, bureaucratic
40:03 systems of reporting and control. All of
40:05 it was developed by the military before
40:07 the re railroads adopted it. In the
40:09 1850s,
40:11 um, Daniel McCllum of the New York and
40:13 Eerie Railroad created the world's first
40:15 organizational chart. Let's go find it.
40:19 Daniel McCllum, world's first
40:20 organizational chart.
40:23 I'm sure this exists somewhere.
40:28 Images.
40:31 There it is. Look at that.
40:38 What's happening?
40:43 Oh, funny. I set a timer on my watch. It
40:45 was buzzing.
40:49 Dr. J, there's something there. Okay.
40:52 With the with the small groups, right?
40:54 Isn't there something cool there?
40:55 Introducing the organization chart. Holy
40:57 [ __ ] look at this thing.
41:02 Can I
41:04 copy image?
41:10 Holy [ __ ] this is cool.
41:17 Oh, you can't read it. It's too small.
41:19 That's too bad. New York Eerie Railroad
41:22 representing a plan of organization.
41:28 Wow.
41:30 That's crazy.
41:32 There's the first org chart,
41:36 huh? Who knew
41:45 the informal management? Wait, let's
41:47 see. Uh, the informal management styles
41:51 that worked for smaller railroads were
41:52 failing. Train collisions were killing
41:54 people. Macllum's chart formalized the
41:57 same hierarchical logic. Romans had used
41:59 layers of authority, defined reporting
42:02 lines, structured information flow. It
42:04 became the blueprint for the modern
42:06 corporation. YouTube comment. Um, I
42:09 worked for one for 10 years. That looks
42:12 like a railroad organization chart. All
42:14 right. That's wild. BNSF railroad.
42:17 That's that's wild, dude. Amazing. Um,
42:20 okay. Let's see. Frederick Taylor 1856
42:24 to 1915 often called the father of
42:27 scientific management optimized what had
42:29 become what what happened within that
42:32 hierarchy. Taylor broke work into
42:34 specialized tasks. So this begins okay
42:37 so 18 so this is probably he was born in
42:41 1856 so probably around 1900
42:45 Taylor breaks work into specialized
42:47 tasks
42:49 tasks
42:51 right end of the 1800s early 1900s
42:57 specialized tasks
43:00 we train for skills you're going to do
43:03 your tasks tailorism here we Go.
43:07 This produced functional pyramid
43:09 organization, a structure structure
43:10 optimized for efficiency.
43:14 Oh, wait. Taylor broke work into
43:16 specialized tasks, assigned them to
43:18 trained experts, and managed through
43:21 measurement rather than intuition.
43:24 Right. So, KPIs. What are your KPIs,
43:27 people?
43:29 Key performance indicators. Huh? This
43:31 produced a functional pyramid
43:33 organization, a structure optimized for
43:35 efficiency within the information
43:37 routing system that the military had
43:39 pioneered and the railroads had
43:41 commercialized. The first rail stress
43:44 test of functional hierarchy came during
43:46 World War II. The Manhattan project
43:49 required physicists, chemists,
43:51 engineers, metallurgists, and military
43:54 officers to work across disciplinary
43:55 boundaries toward a single objective
43:57 under extreme secrecy and time pressure.
44:01 Robert Oppenheimer organized Los Alamos
44:03 into functional divisions, but insisted
44:06 on open collaboration across them.
44:09 Resisting the military's instinct to
44:10 compartmentalize, and NASA
44:12 compartmentalizes.
44:14 When the implosion problem became
44:16 critical in 1944, he reorganized the lab
44:19 around it, created crossf functional
44:21 teams unlike anything in corporate
44:23 America at the time. It worked, but it
44:25 was a wartime ex exception led by a
44:29 singular figure. this kind of question.
44:33 Wait, the question the post-war business
44:35 world faced was whether that kind of
44:37 crossf functional coordination could be
44:39 made routine.
44:41 With the growth in globalization of
44:43 companies after World War II, the scale
44:46 limitations of functional design became
44:49 acute. In 1959, McKenzie's Gilbert Cle
44:53 and Alfred Dipio published creating a
44:56 world enterprise in the Harvard Business
44:58 Review, providing an intellectual
45:00 framework for a matrix organization that
45:03 combined functional specialties with
45:05 divisional units.
45:08 Fascinating wild stuff because it's all
45:11 about to blow up.
45:14 It's all about to blow up, which is what
45:16 this article is about. Under the
45:19 leadership of Marvin Bower, McKenzie
45:20 helped companies like Shell and GE
45:22 implement these principles, balancing
45:24 central standards with local agility.
45:27 This became the professional or modern
45:30 corporation that propelled the post-war
45:33 global economy.
45:35 Over time, other frameworks uh came in
45:38 to address complexity, rigidity,
45:40 bureaucracy. The McKenzie 7S framework
45:42 developed in the 70s by Tom Peters and
45:45 Peter Robert Waterman distinguish the
45:48 hard S's strategy structure systems and
45:51 the soft S's shared values skills staff
45:54 and style. The core idea was structural
45:57 elements alone were insufficient.
45:59 Organizational efficiencies required
46:02 alignment across cultural traits and
46:04 human factors blah blah blah. Okay.
46:06 Something coming from Brandon.
46:15 The reading something learning lab.
46:18 Oh, the railroad learning lab.
46:23 Okay.
46:24 In more recent decades, technology
46:26 companies have experimented aggressively
46:28 with organizational structure. Spotify
46:31 popularized crossf functional squads
46:33 with short sprint cycles. Zappos
46:35 attempted hocrisy eliminating ma
46:39 management t titles entirely.
46:42 Um we faced this when we did agency.com
46:44 we tried all sorts of different
46:46 structures. Where we ended up in the end
46:48 was crossf functional 25 to 30 person
46:52 teams that had basically were like
46:55 little mini agencies. So we had um we
46:59 had like a VP that was we had VPs for
47:02 head of the disciplines and then within
47:05 a team you would have project
47:06 management, account management, creative
47:10 um technologists and site builders. So
47:12 front end and back end basically and
47:14 they would all sit in you know
47:17 physically colllocated units um with
47:20 groups of four or five sitting together.
47:22 So and it it ended up working really
47:24 really well. Champ, what do you want?
47:28 You want cheese? You want mama? What do
47:31 you want? All right, hold, please.
48:00 Ow, my knees hurt. Okay.
48:04 Valve operated with a flat structure nor
48:06 no formal hierarchy. Each of these
48:08 experiments revealed that something
48:09 about the limitations of traditional
48:11 hierarchy, but none solved the
48:12 underlying problem. Spotify moved back
48:15 toward conventional management as it
48:17 scaled. Zappos saw significant
48:20 attrition.
48:22 Valve's model proved difficult to scale
48:24 beyond a few hundred people. As
48:25 organizations grew into thousands, they
48:28 revert to hierarchical coordination.
48:30 Okay, so now we're moving on. The
48:34 constraint is the same one the Romans
48:36 faced in the Marine Corps rediscovered
48:38 in World War II. Narrowing span of
48:40 control means adding layers of command,
48:42 but more layers means slower information
48:44 flow. 2,000 years of organizational
48:47 innovation has been an attempt to work
48:49 around this trade-off without breaking
48:51 it. Okay, here's the setup. This is the
48:54 setup.
48:56 2,000 years of organizational innovation
49:00 has been an attempt to work around this
49:02 trade-off. The trade-off being
49:06 as you narrow the span in a large
49:08 organization, you increase layers and
49:10 you slow down communication.
49:12 At block, we're questioning the
49:14 underlying assumption that organizations
49:17 have to be hierarchically organized with
49:20 humans as the coordinating mechanism.
49:22 Instead, we intend to replace what the
49:25 hierarchy does. Most companies using AI
49:29 today are giving everyone a co-pilot,
49:31 which makes the existing structure work
49:33 slightly better without changing it.
49:37 Economy three,
49:40 the efficiency economy.
49:43 We're going to take what we currently do
49:44 and make it slightly better. This is
49:46 what Jack's talking about. He's in
49:47 economy five. He's he's talking about
49:49 being in economy five. He's telling us
49:52 how to be an economy 5.
49:54 Okay.
49:58 Most companies today are using AI are
50:01 giving everyone a co-pilot, which makes
50:02 the existing structure work slightly
50:04 better without changing it. We're after
50:06 something different. A company built as
50:09 an intelligence or a mini AGI. We're not
50:13 the first one to move beyond traditional
50:15 hierarchy. Her hires
50:18 renden hay model platform organizations
50:21 datadriven management. These are real
50:23 attempts at the same problem. What they
50:25 lacked was a technology capability of
50:28 actually performing the coordination
50:30 functions. So basically what he's going
50:32 to say is he's going to turn loose small
50:35 pods of teams and let the AI organize it
50:37 and track it and collect speed up make
50:42 instant the communication challenge.
50:45 This is wild.
50:48 Um what they lacked was a technology
50:51 capable of actually performing the
50:53 coordination functions that hierarchy
50:55 exists to provide. AI is that
50:57 technology.
50:59 For the first time, a system can
51:01 maintain a continuously updated model as
51:05 an entire of an entire business and use
51:08 it to coordinate work in ways that
51:10 previously required humans relaying
51:13 information through layers of
51:15 management.
51:17 There's going to be no middle management
51:19 in his his organization. And there's
51:22 going to be probably a seuite
51:25 and then pods
51:27 and AI will just manage the pods. Go do
51:29 this [ __ ] I need this [ __ ] done to for
51:32 what these guys want. You all go do my
51:35 [ __ ]
51:36 It's amazing.
51:38 Um, for this to work, a company needs
51:41 two things. a kind of world model of its
51:43 own operations
51:45 and a customer signal rich a customer
51:48 signal rich enough to make that model
51:50 useful. Block is remote first.
51:53 Everything we do creates artifacts,
51:55 decisions, discussions, code, designs,
51:59 plans, problems, and progress all exist
52:02 as recorded actions. It's the raw
52:04 material for a company world model. In a
52:07 traditional company, a manager's job is
52:09 to know what's happening across their
52:11 team and relay that context up and down
52:13 the chain.
52:15 In a remote first company where work is
52:17 already machine readable, AI can build
52:20 and maintain that picture continuously.
52:23 What's being built, what's being
52:24 blocked, where the resources are
52:26 allocated, what's working and what
52:27 isn't. That's the information the
52:30 hierarchy used to carry. The company
52:33 world model carries it instead. This is
52:36 [ __ ] brilliant.
52:38 AI is the tech. Bingo. Yeah, exactly.
52:42 This is brilliant.
52:45 Uh but the capability of the system is
52:48 only as good as the quality of the
52:50 customer signal. Okay. So the first
52:52 thing he's saying is
52:55 if all of your
52:59 business knowledge is locked in middle
53:03 management's heads,
53:05 you're [ __ ]
53:07 If you're a remote company,
53:10 you've already got it digitally. So the
53:12 AI can deal deal with the digital stuff.
53:14 It can it can read the story. But now
53:16 he's saying, "But the capability of the
53:18 system is only as good as the quality of
53:20 the customer signal feeding it." And
53:23 money is the most honest signal in the
53:25 world. People lie on surveys. They
53:27 ignore ads. They abandon carts. But when
53:29 they spend, save uh send, borrow, or
53:33 repay. That's the truth. Every
53:36 transaction is a fact about someone's
53:38 life. Block sees both sides of millions
53:40 of these transactions every day. a buyer
53:43 through Cash App and a seller through
53:45 Square plus the operational data from
53:48 running the merchants's business that
53:49 gives the customer world model something
53:51 rare. A per customer per merchant
53:54 understanding of the financial reality
53:56 built from honest signal that compounds.
53:59 The richer the signal, the better the
54:01 model. The better the model, the more
54:03 the transactions. The more the
54:04 transactions, the richer the signal.
54:07 Jack Dorsey's [ __ ] smart dude.
54:12 So cool.
54:14 Together, the world model and the
54:16 customer world, the the company world
54:18 model and the customer world model form
54:20 the foundation for a different kind of
54:23 company. Instead of product teams
54:25 building predetermined road maps, you
54:28 build four things. This is so cool.
54:30 First, capabilities. The atomic
54:33 financial primitives, payments, lending,
54:35 card issuing, banking, buy now pay
54:38 later, payroll, and so on. These are not
54:40 products. They're building blocks that
54:42 are hard to acquire and maintain and
54:44 they have network effects and regul
54:46 regulatory permission. They have no no
54:48 UIs of their own. They have reliability,
54:51 compliance and performance target
54:53 targets. Second, a world model. This has
54:56 two sides. The company world model
55:00 is how the company understands itself in
55:02 its own operations, performance and
55:04 priorities. replacing the information
55:07 that used to flow through layers of
55:09 management. The customer world model is
55:11 the per per customer per merchant per
55:14 market representation built from prop
55:18 proprietary transaction data. It starts
55:21 with the raw transaction data today and
55:23 evolves towards a a full causal and
55:25 predictive model over time. It's
55:28 amazing. So basically, you've got the
55:30 company world model that understands the
55:32 AI understands here's all the [ __ ] I can
55:35 do. I've got these financial
55:39 infrastructure that I can deal with
55:41 money. I've got these skill pods that do
55:46 [ __ ] And then I've got customer
55:49 transactions down to the transaction
55:51 level and up to the merchant level
55:55 that I can track and say, "Oh, that
55:58 thing we did over here made those
56:00 numbers go higher over there. Do more of
56:02 that. In fact, let's take three of these
56:05 pods. You three pods. You're not doing
56:07 this anymore. Now you're doing that."
56:09 And go do go do.
56:13 It's crazy.
56:15 Third, an intelligence layer. This is
56:17 what comp composes capabilities into
56:20 solutions for specific customers at
56:22 specific moments and delivers them
56:24 proactively.
56:26 A restaurant's cash flow is tightening
56:28 ahead of seasonal dip. The model has
56:30 seen before. The intelligence layer
56:33 composes short-term loan from the
56:34 lending capability, adjust the repayment
56:37 schedule using the payments capability,
56:40 surfaces it to the merchant before they
56:42 even think to look for financing. A cash
56:45 app's user spending pattern shifts in
56:48 the way the model associates with a move
56:51 to a new city. The intelligence layer
56:53 composes a new direct deposit setup, a
56:55 cash app card with boosted categories
56:58 for the new neighborhood, a savings goal
57:00 calibrated to their updated income. No
57:03 product manager decided to build either
57:05 solution. The capabilities existed. The
57:07 intelligence layer recognized the moment
57:09 and composed them. So this is he's
57:12 essentially talking about a company
57:14 that's building custom dynamic solutions
57:17 per customer. [ __ ] insane.
57:22 If you don't think the world's about to
57:23 [ __ ] change, think about
57:27 all of the systems
57:29 that are hardcoded right now to do
57:32 everything he just talked about where
57:35 you either have to build those systems
57:36 or you interface with other players that
57:40 take a a cut of every transaction.
57:43 He's talking about dynamically building
57:45 those things per customer. It's insane.
57:49 Fourth, interfaces, hardware and
57:50 software. Square cash app, Afterpay,
57:52 Title, Bit Key, Proto. These are all
57:54 delivery surfaces which the intelligence
57:57 layer delivers composed solutions. They
57:59 are important, but they are not where
58:01 the value is created. The value is in
58:03 the model and the intelligence. When the
58:05 intelligence layer tries to compose a
58:07 solution and can't because the
58:08 capability doesn't exist, that failure
58:11 signal is a future road map. The
58:14 traditional roadmap where product
58:16 managers hypothesize about to what to
58:19 build next in any company's ultim is is
58:21 any company's ultimate limiting factor.
58:24 In this model, customer reality
58:27 generates the backlog directly.
58:30 Customer wanted something, we couldn't
58:32 deliver it. Dynamically design that
58:35 feature, deliver it to the customer.
58:39 That's [ __ ] crazy.
58:44 If this is what the company builds then
58:46 the question becomes what do people do?
58:49 Good. The org structure follows from
58:51 this and it inverts the traditional
58:54 picture. In a conventional c company the
58:56 intelligence is spread throughout the
58:58 people and the hierarchy routes it in
59:01 this model. The intelligence lives in
59:03 the system. The people are on the edge.
59:06 The edge is where the action is. The
59:08 edge is where the intelligence makes
59:10 contact with reality. People reach into
59:12 places the model can't yet go.
59:16 They sense things the model can't
59:17 perceive. Intuition, opinionated
59:20 direction, cultural context, trust
59:22 dynamics, the feeling in a room.
59:26 So the intelligence, the AI intelligence
59:29 is sitting at the center figuring all
59:30 this [ __ ] out and then it's know it
59:32 knows it has this interface of people
59:34 out into the world and with the other
59:36 people in the organization. Lord digital
59:38 gods. The seven economies include my
59:40 pillow makes me sleepy.
59:49 This is really cool. They sense and if I
59:52 if I'm boring you, sorry about that.
59:54 That's okay. I mean, sometimes it's
59:56 boring here.
59:59 Um, they make calls the model shouldn't
1:00:01 make on its own, especially ethical
1:00:03 decisions, novel situations, and high
1:00:05 stakes moments where the cost of being
1:00:06 wrong is existential. A world model
1:00:09 can't touch the world. It's just a
1:00:11 database. But the edge doesn't need
1:00:13 layers of management to coordinate it.
1:00:15 The world model gives every person at
1:00:17 the edge the context they need to act
1:00:20 without waiting for information to
1:00:22 travel up and down the chain of command.
1:00:25 This is [ __ ] brilliant. This is I I I
1:00:29 had a feeling this was coming from what
1:00:30 Jack wrote when he laid off 4,000
1:00:32 people, which is why I created this
1:00:35 economy five, the aggressor economy. I
1:00:37 basically made this surmising this is
1:00:40 what Jack was about to do. And he did it
1:00:43 two weeks later after he laid off 4,000
1:00:45 people. So he obviously had this in in
1:00:48 the in the can.
1:00:51 Um, in practice, this means we normalize
1:00:54 down to three roles. Oh, here we go.
1:01:00 You want to know your job of the future?
1:01:03 Hello, my friends. Hello, Mimi in the
1:01:05 making.
1:01:13 Okay.
1:01:16 Three roles. Three human roles.
1:01:20 Three. Three
1:01:23 individual contributors who build and
1:01:25 operate capabilities. The model, the
1:01:27 intelligence layer, the interfaces. They
1:01:29 are deep specialists and experts in a
1:01:31 specific layer of our system. The world
1:01:33 model provides the context that a
1:01:36 manager used to provide. So individual
1:01:38 contributors can make decisions about
1:01:40 their layer without waiting to be told
1:01:42 what to do.
1:01:43 DRI directly responsible individuals who
1:01:46 own specific crosscutting problems or
1:01:49 opportunities and customer outcomes. A
1:01:52 DRRI might own the problem of merchant
1:01:55 churn in a specific segment for 90 days
1:01:58 with the full authority to pull
1:02:00 resources from the world model team, the
1:02:03 lending capability team, the interface
1:02:05 team as needed. Dr. DRRI may persist on
1:02:09 certain problems or move elsewhere to
1:02:12 solve new ones. And then finally, player
1:02:14 coaches who combine building with
1:02:16 developing people. This is brilliant.
1:02:19 This is brilliant.
1:02:21 So people do [ __ ] People figure out
1:02:25 what the customer needs and make sure
1:02:27 that those resources are are are well
1:02:29 allocated. And then player coaches that
1:02:31 make sure people are developing.
1:02:34 They replace the traditional manager
1:02:36 whose primary job was information
1:02:37 routing. The player coach still writes
1:02:39 code. A player coach still writes code
1:02:42 or builds model or designs interfaces.
1:02:44 They also invest in the growth of people
1:02:46 around them. They don't spend their days
1:02:47 in status meetings, alignment sessions,
1:02:50 and priority negotiations. The world
1:02:52 model handles alignment. The DRRi DRRI
1:02:56 structure handles strategy and priority.
1:02:59 The player coach handles craft and
1:03:01 people.
1:03:03 It's really good.
1:03:05 There is no need for a permanent middle
1:03:08 management layer. Everything else the
1:03:10 whole whole old hierarchy did the system
1:03:14 coordinates or oh everything else the
1:03:17 old hierarchy did the system coordinates
1:03:20 and everyone is empowered with a role
1:03:22 that's much closer to the work and the
1:03:24 customer. Block is in the early stages
1:03:26 of this transition. It will be a
1:03:28 difficult one. Yeah, this is this is
1:03:31 brutal. This is brutal
1:03:34 and parts of it will likely break before
1:03:36 they work. We're writing about it now
1:03:38 because we believe every company will
1:03:40 eventually need to confront the same
1:03:41 question we did. What does your company
1:03:43 understand that is genuinely hard to
1:03:46 understand? And is it that
1:03:47 understanding? Is that understanding
1:03:49 getting deeper every day? If the answer
1:03:51 is nothing, AI is just a cost
1:03:53 optimization story. You cut your
1:03:55 headcount, improve the margins for a few
1:03:58 quarters, and eventually get absorbed by
1:03:59 something smarter. If the answer is
1:04:02 deep, AI doesn't augment your company.
1:04:05 It reveals what your company actually
1:04:06 is. Block's answer is the economic
1:04:09 graph. Millions of merchants and
1:04:10 customers, both sides of every
1:04:12 transaction, financial financial
1:04:14 behavior observed in real time.
1:04:17 That understanding compounds every
1:04:19 second of the system, every second the
1:04:21 system operates. We believe the pattern
1:04:23 behind this, a company organized as an
1:04:25 intelligence rather than a hierarchy,
1:04:29 is significant enough it will reshape
1:04:31 how companies of all kinds operate over
1:04:33 the coming years. Block is far enough
1:04:35 along to show the ideas more than
1:04:37 theory. Though we welcome debate and
1:04:39 feedback,
1:04:42 we believe
1:04:45 the pattern behind this, a company
1:04:48 organized as intelligence rather than a
1:04:50 hierarchy, is significant enough that it
1:04:53 will reshape how companies of all kinds
1:04:56 operate over the coming years.
1:04:59 That's all we need to read. There's the
1:05:02 punch line. Oh, that was it. Companies
1:05:05 move faster or slow based on information
1:05:07 flow. Hierarchy and middle management
1:05:09 impede information flow. Yes.
1:05:13 Go watch office space.
1:05:17 Yeah. Yeah. Peter. Yeah. Hey, listen.
1:05:20 We're going to need you to go on ahead
1:05:22 and come on in on Saturday. Yeah. Well,
1:05:25 did you did you get the memo about the
1:05:26 TPS reports?
1:05:29 Right.
1:05:32 Hierarchy and middle management impede
1:05:34 information flow for 2,000 years from
1:05:36 the Roman Ku Bernium to today's global
1:05:40 enterprises. We have no real
1:05:42 alternative. Eight soldiers sharing a
1:05:44 tent needed decanus. 80 men need a
1:05:47 centurion. 5,000 needed a legate. The
1:05:50 question was never whether you needed
1:05:52 layers. The question was whether humans
1:05:54 were the only option for what those
1:05:56 layers do. They aren't anymore. Block is
1:05:58 building what comes next. Called this
1:06:01 two weeks ago. Made a chart about it.
1:06:05 Here you go. He's talking about economy
1:06:07 three. He's building economy five.
1:06:11 People in economy four are going to be
1:06:14 pulled toward economy five, but it's
1:06:16 going to be hard. So, they might resist
1:06:18 it. And then what's going to happen is
1:06:19 economies five and six are going to eat
1:06:21 their lunch.
1:06:23 My stapler. That's my stap. Yeah. My my
1:06:26 stapler. My st That's my stapler.
1:06:29 I if I I've done the counting, there are
1:06:32 not enough pieces of cake. Get that was
1:06:36 the That was the last piece piece of
1:06:38 cake. That was the last piece of cake.
1:06:46 [ __ ] crazy people.
1:06:51 From hierarchy to intelligence.
1:06:57 quote I'm going to say this
1:07:02 is
1:07:04 very important
1:07:08 if you are in
1:07:11 business
1:07:14 read this
1:07:18 Jack is building an
1:07:23 economy
1:07:27 company.
1:07:30 In fact,
1:07:34 his
1:07:37 post about laying off 4,000
1:07:44 employees
1:07:47 inspired
1:07:50 my thinking on
1:07:53 the seven economies.
1:08:00 Most companies right now are economies
1:08:07 two, three or four.
1:08:19 you are likely
1:08:22 in an economy 3 company
1:08:27 which Jack
1:08:29 talks about here.
1:08:39 Do not
1:08:42 sleep on this.
1:08:46 And then I'm gonna add my little
1:08:47 picture.
1:08:49 I'm gonna add my picture of the seven
1:08:51 economies. And we're going to send this
1:08:52 out in the world.
1:08:54 I'll tell you what. I'll tell you
1:08:56 something. There, Sarah. Hey, Sarah.
1:09:01 Saturated
1:09:04 name contains saturated.
1:09:06 There we go. Saturated.
1:09:09 There you go.
1:09:12 Fantastic.
1:09:53 post.
1:09:56 All right. All right. All righty. That's
1:10:00 fantastic.
1:10:02 Yeah, that's pretty good. That's pretty
1:10:04 hot. All right. Missed the last five
1:10:07 minutes. Is someone making shakuderie
1:10:10 board
1:10:12 of the seven economies map? I hope so.
1:10:15 That'd be cool, wouldn't it? Um,
1:10:19 that's kind of wild. All right, let me
1:10:21 let me come back to Earth here for a
1:10:22 second and check in with you all. I just
1:10:24 put DRRi on my resume. There you go.
1:10:29 Um,
1:10:33 that's a trippy article. In fact, I got
1:10:35 to send that to some people.
1:10:42 Tik Tok question.
1:10:45 What What economy does academia and
1:10:47 science fit into? So, Cam, the the the
1:10:50 way I think about it is this. the the
1:10:54 layers like education, business, um what
1:10:58 did you say? Uh science,
1:11:01 um politics,
1:11:03 those are um horizontal layers and the
1:11:08 seven economies are vertical slices
1:11:10 through them. So think of the each
1:11:11 economy as level of AI adoption.
1:11:16 Um,
1:11:21 so you're going to have educational
1:11:22 institutions
1:11:24 that some are probably still in economy
1:11:27 one. They're just using books and
1:11:29 whiteboards or, you know, blackboards
1:11:31 and chalk and [ __ ] like that. And
1:11:34 they're not really using technology.
1:11:35 You've got some in economy 2 that are
1:11:37 using, you know, some technology, but
1:11:39 they're just, you know, they're still
1:11:40 using fax machines.
1:11:43 Then you're going to have economy three
1:11:45 educational institutions
1:11:47 that are using AI to like improve email
1:11:50 communications. Then you're going to
1:11:51 have economy 4, which is the transition
1:11:53 economy. They'll be doing a little bit
1:11:55 of innovation. And then you'll have
1:11:57 educational institutions doing five,
1:11:59 six, and seven. And same with science. I
1:12:01 have a feeling science is just going to
1:12:02 sort of lean quickly into five, six, and
1:12:06 seven. But a lot of scientific
1:12:08 institutions are slow and traditional.
1:12:11 So, so again, what's going to happen is
1:12:14 what the the reason I'm calling these
1:12:15 seven economies, you've got AI
1:12:19 accelerating up and to the up and to the
1:12:21 right
1:12:23 quickly.
1:12:25 And then you've got 99% of people on
1:12:27 Earth that don't even really know what
1:12:29 AI is. They maybe use Chat GPT, the free
1:12:32 version of it, like like you know, 16%
1:12:35 of them use chat GPT like a Google
1:12:37 search.
1:12:39 83% of them have never used AI. And AI
1:12:42 is accelerating away from us. So as it
1:12:44 accelerates away from us, it's going to
1:12:46 split. It's going to just fracture.
1:12:49 All these different behaviors, all these
1:12:51 different levels of AI adoption are
1:12:53 going to start functioning like separate
1:12:54 economies. And so the the stack of
1:12:57 education,
1:12:59 science, government, business, private
1:13:02 sector, you know, uh you know,
1:13:06 citizenship, citizenry
1:13:09 are all going to they're going to be all
1:13:11 in this layer in an economy
1:13:14 is I how I think it goes.
1:13:21 Kim Kek, my college has fully embraced
1:13:23 AI. That's amazing. That's great.
1:13:28 It's a seven economy cake.
1:13:31 That's really cool.
1:13:36 Um, let me copy this.
1:14:31 All right, I sent that off to some
1:14:33 people.
1:14:35 Um,
1:14:40 post on Oh, post on LinkedIn. I'll share
1:14:42 with my railroad account. Okay, I'll do
1:14:44 that. Let me go post this on LinkedIn.
1:14:46 Actually, it's a good idea.
1:14:48 Um, I'll I'll share my screen.
1:14:52 Yes, AI readiness project podcast is
1:14:54 tomorrow
1:14:56 um at 400 PM
1:14:59 Mountain time.
1:15:02 Oh, I'm sharing. Okay.
1:15:30 Okay. So, we're going to go next. We're
1:15:33 going to go um
1:15:37 this
1:15:40 tweet from
1:15:42 Jack.
1:15:46 Who's that Jack? Let's see.
1:15:51 Um X
1:15:53 is
1:15:55 very important.
1:16:07 He's declaring what it means to be an
1:16:15 economy five company.
1:16:19 referring to most companies who
1:16:26 are now economy
1:16:29 2 or three.
1:16:36 The TLDDR
1:16:40 is that
1:16:43 for 2,000 years
1:16:50 Large organizations
1:16:54 required
1:16:57 slow
1:17:00 hierarchies
1:17:04 to
1:17:08 manage the information flow
1:17:13 between
1:17:16 um working units.
1:17:32 He is actively replacing
1:17:37 the hierarchy
1:17:42 at block. I wonder if block is block
1:17:49 a company block labs block company
1:17:52 financial services he's actively
1:17:55 replacing
1:17:56 the hierarchy of block with
1:18:00 AI with with intelligence
1:18:10 if you are in business.
1:18:14 This should be
1:18:17 required reading
1:18:20 and a flipping wake up call.
1:18:27 If you are an employee,
1:18:34 figure out
1:18:39 which of the three roles
1:18:44 individual
1:18:47 individual
1:18:49 contributor
1:18:52 what did he say?
1:19:04 Directly responsible individuals.
1:19:11 Directly
1:19:13 responsible
1:19:18 individuals
1:19:22 or player,
1:19:27 coaches.
1:19:30 You
1:19:33 think you are qualified
1:19:37 for
1:19:45 Let's see.
1:19:48 Paul Rer.
1:19:52 Let's see. Michael,
1:20:00 Stellner,
1:20:06 Retesh, Patel,
1:20:10 Ben Jones,
1:20:14 let's see, Andy Scarantino,
1:20:19 Beth Commtock.
1:20:21 Let's see. Chris Hoyer,
1:20:28 what the Thomas Eric
1:20:32 Makier
1:20:35 IC's Dr.
1:20:37 These are tech company terms.
1:20:41 Yeah, they're they're tech company
1:20:43 terms, but like abstract them up. A
1:20:47 player coach is someone who can do but
1:20:48 can develop people. A DRRi is basically
1:20:51 think of it like an account manager with
1:20:54 with deploying and budget capability,
1:20:58 right? So, think about someone who can
1:21:01 say, "Hey, this customer needs this. I'm
1:21:03 going to I'm going to tell the
1:21:04 intelligence, here's what I need, and
1:21:07 the intelligence is going to figure out
1:21:08 how to get me that." And then individual
1:21:10 contributors will make sure all the
1:21:12 shit's working, right?
1:21:17 Uh
1:21:19 uh E A E Wait, ABCDE E F.
1:21:23 Does anyone else do that when you're
1:21:24 trying to figure out where you are in
1:21:25 the alphabet? You have to start at A. A
1:21:27 B C D F G I say
1:21:30 T U V V V V V V V V V V V V V V V V V V
1:21:31 V V V V V V V V V V V V V V V V V V V V
1:21:31 V V V.
1:21:33 Am I the only one?
1:21:38 I'm meeting Ben on Thursday.
1:21:45 Waiting for Kazio. Let's see at
1:21:50 G.
1:21:55 Did my sharing just go away? It did,
1:21:58 didn't it? God damn it.
1:22:14 H I
1:22:21 J
1:22:24 John Patel
1:22:26 J Jason
1:22:30 K
1:22:32 Kelly Anderson. We'll put Kelly. Chef
1:22:34 Kelly's on the list. Um, H I J K L
1:22:41 Log Logan Kilpatrick. We'll give him
1:22:43 one.
1:22:44 M.
1:22:46 We'll do Marique.
1:22:49 N. Oops.
1:22:53 All right. I'm bored now.
1:23:00 That's good. [ __ ] it. That's enough.
1:23:04 We got our thing.
1:23:09 All right. There you go. So, if you go
1:23:10 to my LinkedIn account right now, it
1:23:13 says, "This tweet from Jack on X is
1:23:15 really important."
1:23:23 Okay, heading out folks. See y'all later
1:23:24 in the AI salon. Mary Carol, pace out.
1:23:28 Yes. ABC D, etc.
1:23:32 Beautiful. Beautiful, beautiful people.
1:23:34 Fantastic. All right, it's getting to be
1:23:37 that time. It's getting a little late
1:23:39 there. Little late there in the evening.
1:23:42 Don't even mess with LinkedIn anymore.
1:23:44 It's messy. Yeah, I know. I'm I'm
1:23:46 actually switching over to Substack. So,
1:23:47 do me a favor.
1:23:49 Um, go to Substack and my Substack is I
1:23:54 am Kyle Shannon.
1:23:56 Let me make sure that's right.
1:24:06 Yep. I am Kyle Shannon.
1:24:09 All right.
1:24:11 Go to go to Substack and I am Kyle
1:24:14 Shannon. All right. All right.
1:24:17 Right. Kyle, check out the irregulars
1:24:19 channel. I will do that right now.
1:24:28 Swing dig gank swing ginking.
1:24:44 Oh, that's so good. Seven economies
1:24:46 shuderie board. Oh, that's so good.
1:24:55 Oh my god,
1:24:58 that's so good.
1:25:01 We got to extend the top of it. Or is
1:25:03 the top of it there? Oh, it is.
1:25:07 It's got a little jankiness in the uh in
1:25:11 the in the words. But damn, that's
1:25:12 really cool, actually.
1:25:15 So good.
1:25:17 It's so good. Stability over change.
1:25:20 Young cheddar improved margins.
1:25:23 Garganola blue chip blueberry compost.
1:25:26 That should be economy one.
1:25:32 That's really cool. That was really
1:25:34 good. Really, really good. That graphic
1:25:37 makes me hungry. I know, right? It's
1:25:39 really good.
1:25:41 All right.
1:25:48 Oh, Lord.
1:25:52 Okay, people. I'm going to get out of
1:25:54 here. Um, so tomorrow, AI readiness
1:25:58 project, that's at 4:00. Um, go to the
1:26:01 AI salon. If you haven't been there in a
1:26:02 while, go there. I want to start getting
1:26:05 us back into conversation in between
1:26:09 these meetings. I want to start getting
1:26:11 things hopping again. Shit's about to
1:26:14 get weird. Shit's about to go down.
1:26:19 We don't want to be doing this alone. I
1:26:21 know it's [ __ ] exhausting.
1:26:23 I get it. If anyone gets it, I get it.
1:26:30 Will you be on this Friday holiday? Um,
1:26:37 I don't know what the holiday is.
1:26:40 What? I don't even know what month we're
1:26:42 in.
1:26:44 It's Good Friday. I will be doing
1:26:46 something good Friday.
1:26:48 It's also first Friday, so I might be
1:26:50 late. So, I'll do something a little in
1:26:52 informal on Friday. We might even do an
1:26:55 experiment. So, one of the things
1:26:58 Brandon and Andy and I worked together
1:27:00 today in our L10 meeting to to figure
1:27:03 out like if we're going to do AI
1:27:06 learning lab inside the community,
1:27:07 what's that look like? What's the what's
1:27:09 the most elegant way to do that? And so,
1:27:12 over the next month, we're probably
1:27:13 going to be experimenting. So maybe
1:27:15 we'll maybe we'll take Friday and
1:27:17 experiment. Um so Friday might be might
1:27:20 be fun
1:27:22 depending on your definition of fun.
1:27:24 Office hours this Friday. Yes. This
1:27:27 Friday is a really important one. That
1:27:29 Jack Dorsey thing that I just read, I'm
1:27:32 not going to read the whole thing on
1:27:33 Friday, but we're going to talk about
1:27:34 that on Friday.
1:27:36 That's a That's a really important It's
1:27:39 a It's a really important um
1:27:42 thing that J Jack just put out there.
1:27:44 It's the first thing that I've seen
1:27:46 where someone has actually declared what
1:27:49 economy five is going to look like.
1:27:52 You've got a going concern. You're about
1:27:54 to reinvent it. There's your road map
1:27:58 right there. Incredibly well thought
1:28:00 out. Like
1:28:03 really important. really important.
1:28:08 I'm exhausted with trying to figure out
1:28:10 clott GPT Pro like beer commercials.
1:28:12 Yeah, tastes great, less filling. Um,
1:28:18 I don't know what to say. Like the the
1:28:20 AI right now, it all still sucks. It
1:28:23 it's not going to within
1:28:26 six months.
1:28:29 We have a different problem when the AI
1:28:31 doesn't suck,
1:28:33 right? when it tastes great and it's
1:28:35 less filling and it's cheap
1:28:38 and it'll get you drunk like Bllandons
1:28:43 and like absin all at the same time.
1:28:49 It's going to be a different problem.
1:28:52 We got to get We got to get reconnected
1:28:53 in community. We got to get reconnected
1:28:56 in community.
1:28:58 Okay. Do you think Opus 4.6 sucks too? I
1:29:00 don't think so. I think Opus 4.6 is
1:29:02 good. It's funny today. Oh, this is this
1:29:05 is a wild [ __ ] insane [ __ ] thing.
1:29:10 So,
1:29:12 so there's this audio
1:29:14 synthesis thing called Cartisia and it's
1:29:17 really good and they've got this
1:29:21 data policy that if we want to use it
1:29:24 for our pharma clients, we have to turn
1:29:25 on data protection. And if we turn on
1:29:28 data protection, we can't log into their
1:29:31 playground. We we can log into their
1:29:33 playground and we can make an audio file
1:29:37 from text we put in,
1:29:40 but you can't download it.
1:29:42 There's they they removed the download
1:29:44 button and they're like, "Well, that
1:29:45 goes against you. You you turned off
1:29:47 security privacy, so you can't download
1:29:50 it on the the account you're logged
1:29:52 into."
1:29:53 So, I actually have to build an
1:29:55 application that uses their API to be
1:29:57 able to pull my my [ __ ] file down.
1:30:00 So, I went into Claude Code today and I
1:30:01 said, you know, here's the [ __ ] they
1:30:04 gave me. Go build me a front end where I
1:30:07 can input text and select from one of
1:30:10 the stupid [ __ ] voices in Cartisia
1:30:13 and get my [ __ ] file. So, it it spent
1:30:16 like half an hour coding. I don't know
1:30:17 if it finished it or not. I I left the
1:30:19 office before I got to it.
1:30:22 Um, I disagree that AI isn't great yet.
1:30:25 It's mathematical. Did you mean
1:30:26 mathematical? Um, it is mathematical.
1:30:30 You tell him, Kyle.
1:30:35 Yeah. You know. All right, I gotta go.
1:30:37 My eyes are blurry. I can't read like
1:30:40 that, you know. You know, man. Um, come
1:30:44 check out the AI readiness project
1:30:45 tomorrow. Go to the AI salon. Go post in
1:30:48 the community feed. Go post in the
1:30:50 community chat. Go post in look what I
1:30:52 made. Go post in the regulars. Join the
1:30:54 [ __ ] mastermind. We just rolled out
1:30:56 today. We had um a workshop that I led
1:31:00 on um figuring out what is your oneofone
1:31:05 thing that you're best in the world at
1:31:08 and like really owning that, discovering
1:31:10 it and owning it. It was a really cool
1:31:13 workshop today. We brought in people
1:31:14 from Andy's Humans 2.0 um group of
1:31:18 coaches and we mixed them with the with
1:31:20 the AI salon people. We crossed the
1:31:22 streams.
1:31:24 It was really cool. It was a good event.
1:31:27 It It's only happening in the
1:31:29 mastermind. All the the um the great
1:31:32 repurpose stuff is is going to be really
1:31:34 [ __ ] cool. I would get your ass in
1:31:36 the in the mastermind. We're
1:31:37 experimenting with it over the next six
1:31:39 months. Um it's going to be good stuff.
1:31:41 It's gonna be really good stuff. All
1:31:43 right, cool. Peace out. I think y'all
1:31:46 are amazing. Appreciate you hanging out.
1:31:48 You have a good night. I'm late. Gareth,
1:31:50 that's it. I'm leaving. Bye.