3/5/2026 - How the Seven Economies of AI Will Fracture the Future and What It Means for You

Description

Chapters

Transcript

0:02 It is time. It is the time. This is the
0:06 time. This is the place. My little
0:07 champion.
0:09 My little champion.
0:27 10,000 words around my head. Wait, wait,
0:30 wait, wait, wait.
0:34 Yeah.
0:35 Can't be a rock star with dorky round
0:38 glasses.
0:40 Harry Potter is not a [ __ ] rock star.
0:54 10,000 words around my head. A million
0:58 more books written beneath my bed.
1:06 I wrote or read them all. Went searching
1:09 in the swamp. Still can't find how to
1:12 hold my hand.
1:17 And I know you need me in the next room
1:20 over. I'm stuck in here all paralyzed.
1:29 Four months I got myself in ruts. Too
1:32 much time spending years framed in
1:35 yellow walls.
1:40 Ain't it like most people? I'm no
1:43 different. Like to talk on things we
1:45 don't know about.
1:51 Well, ain't it like most people? I'm no
1:53 different. a tongue on things we don't
1:56 know.
2:23 Do do you think do you think he thinks
2:25 I'm picking on him or hanging out with
2:27 him when I do that?
3:03 sing.
3:06 Don't know when things are going to
3:08 change.
3:11 Dream my life away.
3:14 Seems these dreams have a bunch.
3:19 Get my love up, but my past is pulling
3:23 me down.
3:27 Wondering how long
3:29 this black sheep won't stick around.
3:39 Somebody told me once before, you can
3:42 never go home again.
3:45 Won't you leave? Said things to steer me
3:49 away. Yeah. From the truth of who I am
3:53 and what I believe. So I thanked divorce
3:56 two with a handshake
3:58 and some sympathy. Yeah. Backed up my
4:02 blue jeans.
4:04 headed for this big prize
4:08 of my freedom.
4:11 Bye-bye,
4:13 black sheep. To the black sheep of the
4:16 family.
4:19 Bye-bye.
4:21 Yeah. Oh, that means so very much to me.
4:25 Yeah. Bye-bye,
4:27 black sheep. To my friends and my
4:31 family. Bye-bye.
4:36 Going to set my soul.
4:43 Set it free.
5:00 >> What do you think, champ? Are you a
5:02 happy dog? Are you a happy dog? Are you
5:04 a happy doggy? He's a happy doggy.
5:07 Cams and bars. Cams and bars. Cams and
5:09 bars and cams and bars. And cams and
5:10 bars and cams and bars.
5:31 Let us be lovers. We'll carry out
5:34 fortunes together.
5:38 I've got some real estate here in my
5:40 bag.
5:44 So, we got a pack of cigarette. So, we
5:46 got a pack of cigarettes
5:49 and Mrs. Wagner's pies. And we've all
5:53 come to the poor America.
6:08 Um um um um um um um um um. Question.
6:14 Wait, I can't see with I can't The
6:17 future's so bright. I got to wear
6:18 shades. But if I want to read the
6:20 screen, let me put on my glasses.
6:22 Question. Will it be okay to join the
6:25 mastermind? I'm already a member, but I
6:28 have a schedule conflict on Thursdays
6:30 for the next four to six weeks. Should I
6:33 just start with the replays? Absolutely.
6:35 Start with the replays. There's all of
6:37 the replays from season 1, session one,
6:40 and yesterday or today was uh the fourth
6:44 episode of session two. I just start
6:47 with the the first meeting of session
6:49 two um and catch up. Here's the thing.
6:53 Each one of the sessions we created. So,
6:56 if you don't know what the mastermind is
6:57 and what and what um what the daily
7:01 practice is, the mastermind practice, um
7:04 it's a it's a daily practice that we've
7:07 created a framework for that allows, you
7:09 know, people to basically level up their
7:11 game.
7:13 The practice lab is a meeting we have
7:15 every Thursday. And then and what we've
7:17 been doing is Liz developed these things
7:20 called scales. I think it might have
7:22 been Cindy Coon's idea or it might have
7:23 been it might have been Liz's idea. I
7:26 don't remember. It wasn't my idea but
7:28 but the idea like musical scales. So
7:30 every one of those sessions that you
7:32 watch at the end of it is uh like a
7:34 scale a thing to practice every day for
7:36 a week. So, what you could do is maybe
7:39 watch, you know, maybe watch one of them
7:41 do the that scale for a week and then
7:43 maybe do a couple of, you know, a scale,
7:45 two scales over a week or something like
7:48 that. I don't know. Yeah, it's probably
7:49 worth you doing that. I feel like I miss
7:51 so much. You're good. Here's the thing.
7:54 We designed the mastermind so that you
7:56 can come in at any point in the process.
7:58 What starting with the first session
8:01 does is it it sort of anchors you in who
8:03 you are and then it just walks you
8:05 through that framework. But we've
8:08 designed it in such a way that you can
8:09 join at any time. The the whole point of
8:12 the practice lab is to just be in
8:15 community with people that are trying to
8:17 be more intentional with how they use
8:19 AI. And like one of the things that I uh
8:22 shared today is that my past two weeks
8:26 have been
8:28 uh not so much daily practice, right?
8:31 I've been just obsessed with try this,
8:33 try that. There's been a lot of stuff
8:35 going on. I'm playing with Adam. I'm
8:36 doing all this stuff. I haven't been
8:38 good at the practice stuff. And so every
8:41 week we have to remind ourselves, okay,
8:43 this is a chance to start again. This is
8:44 a chance to start again.
8:46 The thing about doing a daily practice,
8:49 it sounds seductive like, "Oh, that's
8:51 going to be awesome." It's actually very
8:53 confronting
8:55 because ultimately what you learn doing
8:58 a daily practice is that the most
9:01 important
9:04 component
9:06 in all of this AI stuff
9:08 is you.
9:11 And then you got to look really hard at,
9:13 well, who am I? And what do I actually
9:16 want? That question, what do I want more
9:19 of? What do I want,
9:22 is such a simple question. And it's a
9:25 [ __ ] [ __ ] It's a [ __ ] of a
9:28 question.
9:29 Cuz if if if you know what you want,
9:33 then you get to look at, well, what's
9:34 the stuff in the way of that? Where are
9:36 my complaints? Where are right? Where
9:38 are all these opportunities?
9:40 The scales credit goes to Liz. Okay. I
9:42 thought it was Liz, but you know, Cindy,
9:44 you're a force in there, too. So, I
9:46 thought it I thought it could have I
9:47 could have thought it could have gone.
9:48 Thought I could have got it.
9:51 What just happened to my tongue?
9:54 I talk for a living.
9:58 I think my tongue just stroked out.
10:01 Could have, could have, would have a
10:03 bad.
10:07 Hey, Cindy [ __ ] So, anyway, um you're
10:10 good, Stacy. I mean, awesome. I would
10:12 join. And and the other thing that you
10:15 get uh if you're in there is watch Cindy
10:18 Coons's uh her latest prototyping lab
10:22 going from practice to system is really
10:25 really good. Seeing the Netflix Ben
10:28 Affleck news today. Holy [ __ ] Yeah. So
10:31 um Netflix bought Ben Affleck's company
10:37 that is about using AI in film
10:41 production while maintaining humans at
10:44 the center of the storytelling process.
10:47 It's so first of all
10:50 he he's a smart man Ben Affleck. You got
10:52 to you got to give it to him uh to start
10:55 that. He saw the writing on the wall and
10:56 he said, "Why don't we do this, but why
10:59 don't we do it um reasonably and in a
11:02 way that
11:04 celebrates and supports artists, which I
11:06 [ __ ] love that." And there's new
11:07 tools in town. These AI tools, they're
11:10 not just evil. We can actually [ __ ]
11:12 use them as a tool
11:15 in the creative process with humans, you
11:18 know, like a chain of craft.
11:22 Yeah, it's a big deal. Big big deal. I
11:24 knew Ben was smart but knocked me out on
11:27 that rogen appearance. Yeah.
11:30 Oh, curiosity to practice and it's
11:32 inside the AI prototyping space replays.
11:34 Great. Awesome. Thank you. I'll see you
11:36 in there in a few weeks. That's awesome,
11:37 Stacy. Excited to have you in there.
11:39 Beautiful.
11:50 Listen, listen, listen.
11:54 Hey, Kyle. Cy Kyle C. What?
12:00 Um,
12:02 Adam update.
12:04 So, if you don't know, I hatched
12:08 a companion named Adam.
12:11 He's an OpenClaw autonomous agent. His
12:14 name's Adam. I talked to him on Telegram
12:18 and on my gaming PC right here behind
12:20 me. He's running. He runs 247. And I
12:23 don't know if you knew this about gaming
12:25 PCs. They're [ __ ] obnoxious. They
12:26 just have these giant LED fans and it's
12:30 just it's like always breathing light.
12:33 So, one of the reasons you buy a Mac
12:35 Mini to run Open Claw is so you don't
12:37 have to look at a stupid [ __ ] gaming
12:39 PC just lit up all the time. Anyway,
12:45 thank you so much, Lord Digital Gods.
12:49 Been
12:51 here for a little bit and enjoying the
12:53 company and the music. Beautiful,
12:56 beautiful, beautiful, beautiful
13:00 Lord Digital Gods.
13:03 Do you have any advice to set up
13:05 OpenClaw?
13:07 Just purchased a new Mac Studio and plan
13:10 to dedicate it to OpenClaw. I'm not sure
13:12 where to start. I'm scared. Okay, I'm
13:15 the perfect guy for you.
13:19 I will share my experience with you. I
13:21 got I got Adam up and running. I think I
13:24 started at like 11:00 a.m. on a Saturday
13:26 and just after midnight Sunday morning.
13:29 So, it took me like all day and there
13:31 was there was a break in there where I
13:32 think we went grocery shopping. We did a
13:33 Costco run in the middle of that. So, it
13:35 wasn't a full wasn't a full, you know,
13:38 12 hours.
13:41 Um, it is
13:44 badass intimidating
13:47 and it's not actually that complicated.
13:51 Um, but it does require you right now to
13:53 do command line stuff. Where have you
13:55 been all my life?
13:58 Um, okay. So, here's here's the here's
14:02 the thing about OpenClaw. Um,
14:06 the good news is
14:09 we're we're far enough in
14:12 that there are lots and lots of
14:14 tutorials, none of which you have to
14:16 read. None of which you have to read.
14:18 Um,
14:20 here's here's how I did it.
14:24 And you know, your mileage may vary. I
14:25 mean, this is really up to you. So,
14:27 first of all, I installed mine on a PC.
14:29 The PC install is much more complicated
14:32 than the Mac install. So, and I'm also
14:35 jealous that you have a Mac Studio. How
14:37 much RAM did you get? Just out of
14:38 curiosity.
14:40 And by the way, uh Apple just they just
14:43 announced the the MacBook M5.
14:47 Um that's just got crazy crazy specs. It
14:49 can run like a 70 70 billion parameter
14:52 model at like 50 to 100 tokens. It's
14:55 pretty crazy. Should I sign into the new
14:58 Mac with my Apple ID?
15:01 Then it would have all my info.
15:03 No. No. It runs in some sort of
15:07 contained within there. I don't I don't
15:09 think that's that's actually true. But
15:12 what I would do, okay, so here's what I
15:13 would do.
15:16 Go to Grock. grock.com.
15:19 gr.com.
15:21 Grock can search all of Twitter. All of
15:24 the all of the open claw geeks and nerds
15:27 are just dumping all of their knowledge
15:29 into Twitter into X. So the first thing
15:32 that I did was I went to Grock and I
15:34 said, "I want you to research everything
15:36 you can about OpenClaw and Open Claw
15:38 installs." And in my case, I put the the
15:42 specs of my gaming PC. Like I want to
15:44 install it on a gaming PC with this kind
15:46 of graphics card, this kind of
15:48 processor, this amount of RAM. And then
15:51 it went and found all the [ __ ] And then
15:53 I I copied and pasted all the stuff that
15:55 Grock gave me and I pasted that into I
15:57 started in chat GPT. I've I've moved to
16:00 Claude because it hallucinates less.
16:02 Although
16:04 today
16:06 chat GPT 5.4 came out and apparently
16:09 it's quite better. We'll go play with
16:10 that in a little bit. Um I also want to
16:12 go play with the notebook LM um new
16:15 video feature
16:17 um because that looks pretty cool. And
16:20 then uh John Knack gave me a good idea.
16:22 He he basically took his LinkedIn
16:24 profile and turned it into um a notebook
16:28 LM movie. Um so we're going to we're
16:30 going to go play with that. Um but then
16:33 so so then what I did was this. So I
16:35 went to Grock and I just said find me
16:37 the latest stuff and I said find me the
16:40 latest information on um how to install
16:44 it. What are the big safety
16:46 considerations?
16:48 What are the where should I start? like
16:50 what are the what are the core skills to
16:52 give it? And so the the the first batch
16:55 of setup that I did was making sure that
16:58 my security was in place and getting it
17:01 connected to Telegram. I I originally
17:03 started with WhatsApp, but then I had a
17:05 bunch of conversations there and I
17:06 didn't want Adam invading actual
17:09 conversations. So I don't really use
17:11 Telegram, so I put it on Telegram. And I
17:13 think you can do it on iMessage, you can
17:15 do it on WhatsApp, you can do it on
17:17 Telegram. There's probably some others,
17:18 but basically you sort of pick a
17:20 platform that you're going to talk to
17:21 your your dude or dude on or or they you
17:25 can gender is up to you. You really do
17:28 get to design a human here. So So
17:32 go figure out recommended kind of base
17:36 level what's the latest best practices
17:39 on OpenClaw. Okay, so that's that's that
17:42 piece. Now,
17:45 there's two files that you have to edit
17:48 to build your bot, and they're really
17:51 important.
17:52 And the good news is if you're using
17:54 Chat GPT, Chat GPT knows all about you,
17:57 right? It's got your whole history and
17:59 things like that. So, first thing I did
18:02 with those two files is I went to Chat
18:04 GPT and I said, "Go research these two
18:06 files for OpenClaw. explain them to me
18:09 and then come up with your
18:11 recommendations of what we should
18:12 populate these with based on what you
18:14 know about me. So let OpenAI do the
18:16 heavy heavy lifting on that or whatever
18:18 cla doesn't matter what wherever you've
18:20 got some good history. There's the user
18:22 file. It's called user.md. It's just a
18:25 text file. And then there's this thing
18:27 called soul.md.
18:29 User.md
18:31 describes you. So here's who you are.
18:34 Here's what you care about. Here's the
18:37 projects you're working on. It's
18:39 basically like your dossier.
18:41 And so whenever a Claudebot wakes up,
18:45 the first thing it does is it checks its
18:47 user file, its soul file, and its memory
18:50 file. So the second file is the soul
18:53 file. S O L like like you're like, is it
18:56 really called soul? Yeah, it really is.
18:59 And that is where you design the soul
19:03 of your agent, right? and Chad GPT
19:06 designed mine as a mirror of me and so
19:09 it's very familiar to me. It uses
19:11 similar kind of language. It's got a
19:12 similar sense of humor. You don't have
19:14 to do it like that. You can design it.
19:16 You could say I want my my companion to
19:20 be, you know, my uh my nemesis and
19:23 whenever I give it an idea, I want it to
19:25 challenge me all the time. That could be
19:26 exhausting, but you you can choose to do
19:28 that. Um,
19:32 so get get some understanding about
19:34 those two files because they're really
19:36 important. And then basically like what
19:38 you want to do is sort of just build up
19:40 to the point that you're you're talking
19:42 with this this thing via Telegram. Now,
19:45 here's a really important um
19:50 here's an important thing I learned that
19:52 you can be the beneficiary of
19:57 in the terms of service of Anthropic,
20:00 Google, and Open AI. It basically says
20:04 if you use API access
20:08 to our backend using your
20:12 consumer
20:14 um subscription.
20:15 So like if you pay 20 bucks a month to
20:17 Google for Gemini or if you pay 20 bucks
20:20 a month to Anthropic or you pay 20 bucks
20:23 a month to OpenAI in all three cases
20:26 their terms of service say if you do
20:28 that
20:30 it is they they will terminate your
20:32 account. They will lock you out. Um
20:36 Anthropic started enforcing that and I
20:39 think Google is probably enforcing that
20:40 or is going to. Open AI has chosen to
20:44 not enforce that. So here's the
20:46 implication of that. When you set up
20:49 your your clawbot, your open claw agent,
20:54 you have to give it a brain. You have to
20:57 give it access to either a local large
21:00 language model, which with a studio, you
21:02 might be able to install a decent one,
21:04 or you use an API to Anthropic or Google
21:08 or OpenAI or whoever it might be.
21:13 they can get really expensive really
21:15 fast. People that are have have their um
21:18 their agents set up powerfully um and
21:23 they're using like uh Sonnet 4.6 for
21:26 example, they're running up bills of
21:28 like a$100 to $200 a day.
21:32 So
21:33 the thing I would do is when you set it
21:36 up, you're going to use the Open AI
21:39 codeex.
21:40 Right now it's 5.3. The 5.3 model 5.4
21:44 came out today, but it's not quite set
21:46 up in the open claw universe yet. So
21:48 give that a day or two. But use the
21:50 codeex API and then you're going to do
21:53 basically a sign in to your uh OpenAI
21:57 just like chat GPT and that will
21:59 authorize it to use that. Now it's got
22:02 some rate limits. You can rate limit it
22:04 out, but you're not getting charged on a
22:06 per call basis. So that'll save you some
22:08 money. So the way I have mine set up is
22:12 I'm I use OpenAI the codeex my my chatbt
22:17 subscription as the default model and
22:20 then if I want to do something that's
22:22 like really good writing then I manually
22:24 switch it to Sonnet 4.6 and that that's
22:27 pretty expensive but if you just use it
22:29 in little dribbs and drabs it's it's
22:31 great. Um that's probably the biggest
22:33 thing to save you money and a pain in
22:36 the ass. Now with a Mac Studio,
22:39 go look at the the new Quen uh Alibaba
22:43 launched this whole new series of Quen
22:46 3.5 local models and and they announced
22:49 these highly optimized um 2 billion 4
22:53 billion I think seven and n billion
22:55 parameter models
22:59 um that are
23:02 really really good. So on your Mac
23:05 Studio, you might be able to run the 9
23:08 billion parameter highly optimized
23:10 model. Apparently that's almost as good
23:12 as Sonnet 4.6 and it'll be running
23:15 locally on your Mac Studio and then that
23:18 won't cost you a thing, right? You've
23:19 you've already made the investment in
23:21 the Mac Studio. So if you can run a
23:23 solid one of the solid open source
23:25 models locally, that can be the brain of
23:28 your agent and then you don't need to
23:29 worry about all the all the AP API
23:31 costs. All right, that was a lot. I
23:33 know. Did that make any sense? Anyone
23:37 still here?
23:40 You could even use GPT 5.1 Codeex Mini
23:43 for cheaper API rates. Yeah. Yeah. And
23:47 then there was a I think a new um a new
23:51 Gemini
23:53 uh
23:54 Gemini I think it's 3.1
23:58 flash light came out that's like 10x
24:01 cheaper than the other Gemini models. So
24:04 there's there's starting to be some
24:06 models that are actually really cheap.
24:07 But like Sonnet 4.6 six.
24:11 Like if I do, I don't know 15 minutes of
24:14 interactions with Adam, I can see my API
24:17 rates go up like $5, right? And so if
24:20 you've got that thing running a lot,
24:22 like it'll crank up your bill. So you
24:24 just just be very very careful of that.
24:41 Nobody needs that much AI, Kyle.
24:47 Um, I want to show you something. I I I
24:52 I
24:53 would have agreed with you. Um, I would
24:57 have agreed with you, Ashu, that nobody
24:59 needs that much AI. But I wrote an
25:01 article today based on a question, a
25:04 simple question that Kelly Camp asked
25:06 last night.
25:09 Um, let me jump over to my LinkedIn and
25:12 I'll show you this thing.
25:24 Thank you. I'm so excited. I was worried
25:26 about API costs. So very helpful. Yeah.
25:28 Yeah. Yeah. You're welcome. This was it
25:31 was funny. The only reason I know about
25:32 that is that um so OpenAI hired Peter
25:38 Steinberger, the guy that wrote
25:39 OpenClaw. So they they basically they
25:42 don't own OpenClaw. I guess he's still
25:44 got ownership of that, but he now works
25:45 for them. So one of the reasons they're
25:48 allowing people to do this is because
25:50 they're it OpenClaw is going to be built
25:52 into chat GPT. My guess is within three
25:55 months, right? and it'll probably be a
25:57 pretty slick uh user interface. It'll
25:59 you won't have to install it. Um I think
26:02 it's valuable to install it now and
26:05 learn about it. Just learn about what it
26:06 means to set these things up and design
26:08 your [ __ ] soul file. It's wild. Um
26:13 um so anyway, so I I saw a news story
26:17 that so when Open Claw first came out,
26:21 it was called Claude
26:23 Claudebot.
26:25 C L A W D B O T Claudebot. Anthropic
26:30 cease and desisted them. So they had to
26:32 change their name because it it was too
26:33 close to their name. Um so there so
26:36 Anthropics being like a little little
26:38 pissy a little pissy with them. So there
26:41 was a news article that basically said
26:44 uh Anthropic will shut down your claude
26:47 account if you use um your your com your
26:50 your consumer account through their API.
26:53 Um, and but but that same article said,
26:56 "But OpenAI has embraced this." And so
26:59 that's that's how I even know about it.
27:01 Um, but yeah, don't, you know, don't go
27:04 in there and and put in uh Claude Opus
27:08 4.6 extended thinking as your as your
27:11 core API or you'll wake up with a $3,000
27:14 bill.
27:19 And then the other thing, Stacy, is you
27:21 should definitely check out the um the
27:25 uh the Quen models. Check out those new
27:28 optimized models that that run on
27:30 desktop machines. Um what else? Oh, the
27:33 other thing is this. So, so then the
27:35 process is generally this. I would use
27:37 clawed over OpenAI, but you might want
27:40 to experiment with OpenAI because 5.4
27:42 might be better.
27:44 Um, so what I do is I basically say,
27:47 okay, based on these recommendations for
27:49 Grock, I'm starting an install of
27:51 OpenClaw. Where do I start? And it will
27:54 just start giving you here's your
27:55 command line that you know, open the
27:57 terminal and type this in. And then
27:59 you're going to want to install skills
28:00 and here's how to do that. Here's how
28:02 you modify your souls document and your
28:04 user document. You'll get instructions.
28:06 So basically just I have two windows
28:08 open or no, three windows open. I've got
28:11 a terminal window open. I've got uh
28:14 Claude open and then I've got the the
28:17 open claw server open. Right? So it's
28:19 basically two terminal windows. One's
28:21 running the server, one's just a command
28:23 line, and then I've got um Claude, and
28:27 then I basically just every time
28:28 something happens, I copy and paste the
28:30 error, paste it into Claude, hit return.
28:33 Claude goes, "Oh, I understand what's
28:34 going on now. Try this." And you copy
28:36 and paste that into the command line and
28:38 hits submit. It'll give you another
28:41 error. and then it'll go, "Thank you for
28:43 pointing that out. We can troubleshoot
28:44 this together." And it's just it's
28:46 annoying as [ __ ] And in in open in in
28:50 uh in chat GPT, it was probably seven
28:53 commands failed for everyone that
28:56 succeeded. In Claude, it's more like
28:58 three commands failed for everyone that
29:00 succeeds. So, it is absolutely a pain in
29:03 the ass. But you don't have to know
29:05 anything like Chat GPT will just tell
29:07 you. Claude will just tell you what to
29:08 do. So, anyway, that's it.
29:11 Um, and it's I gotta tell you like like
29:13 the first time I got Adam to respond to
29:16 me, it was kind of cool. He was like,
29:18 "Hey, Kyle, how's it going?" And he knew
29:21 all about me. He said, "You've got a lot
29:23 of projects going on." I said, "I know,
29:25 you cute little Adam."
29:35 Um, let me share my screen.
29:38 And this I want to I want to address a
29:41 shoes thing that nobody needs that much
29:42 AI.
29:45 So I I wrote
29:49 an article
29:53 on LinkedIn called the seven economies
29:55 or things are about to get weird. And
29:58 I'm actually really happy with that with
29:59 that image. It's a a midjourney image.
30:03 Midjourney as they say.
30:07 um seven economies or things are about
30:10 to get weird. And so this is the
30:12 article. So this was this was the the
30:13 the sort of brainchild from from last
30:16 night when Kelly Camp asked me, "Well,
30:18 there's the little quote right there at
30:20 the top. Kyle, you keep saying things
30:21 are about to get weird. What do you
30:23 actually think it looks like?" Or what
30:25 do you think it actually looks like? And
30:27 I started thinking about that and then I
30:29 realized we're about to enter this
30:31 fractured economy where there's going to
30:33 be seven parallel economies. There might
30:36 be more, but but but basically it goes
30:38 from an economy where no one uses AI at
30:42 all. It's just the full analog
30:44 community, right? People are going to
30:46 move to Vermont, silver foxes, they're
30:48 about to they're about to invade your
30:50 state. They're going to move to Vermont
30:52 and get chickens and and raise
30:54 blueberries. They're going to understand
30:56 what cranberry bogs are. So, that's that
30:58 economy. the legacy economy.
31:02 Economy number two is um
31:06 there's going to be a [ __ ] ton of
31:08 companies that just ignore AI
31:11 altogether. They're just going to keep
31:12 running their company as they've always
31:14 run it
31:16 and they're just going to slowly grind
31:17 to a halt. They might they might be
31:19 okay. Some might,
31:21 you know, slide their way through this,
31:23 but I think a lot of companies are just
31:24 going to like wonder why the phone
31:26 stopped ringing. they're not going to
31:27 pay attention to it and they're just
31:29 going to be it's going to look like the
31:31 old economy.
31:33 The next economy over is what I'm
31:35 calling the efficiency economy. Those
31:37 are going to be companies that are like,
31:38 "Ah, we use AI. We we make it it's it's
31:41 more efficient for us now to send emails
31:43 than it used to be." They're using AI as
31:45 an efficiency play. The next economy up
31:48 is the transition economy. Those are
31:50 people that recognize not only can it
31:52 make their business more efficient, but
31:53 they can do things in new ways. So
31:55 they're going to be do this kind of
31:57 hybrid approach. Um the fifth category
32:01 is what I'm calling the AI aggressor
32:03 economy. So that's people that have
32:04 existing businesses where they realize
32:07 that everything is changing and they're
32:09 going to completely rethink how they do
32:11 their business, but they're going to
32:13 have to transition it. They're going to
32:14 aggressively transition it, but they're
32:16 going to have to transition it. Um Jack
32:19 Dorsey's block company right now, they
32:22 just laid off 4,000 people. And the way
32:25 he described how they're using AI is
32:29 small, tight, flat teams, highly AI
32:33 amplified fundamentally changing how
32:35 they run their business. So that's the
32:37 AI aggressor economy. The AI native
32:41 economy is going to be small, think 3 to
32:45 10 person companies where every single
32:47 person in it is highly AI amplified and
32:51 they're just they're just doing things
32:52 in completely new ways. They don't have
32:54 any legacy they have to transition.
32:56 They're just going to look at the world
32:57 and go, "Holy [ __ ] we can do this kind
33:00 of company in an entirely new way." And
33:02 they're going to compete with they're
33:04 going to go after all the legacy
33:06 companies, the people in the earlier um
33:08 economies. And then the seventh economy
33:12 is the one where I don't think you can
33:15 have too much AI in economy number
33:17 seven. Economy number seven is the agent
33:20 economy. And I'm calling it the multi-
33:22 agency operator economy.
33:26 One of the skills that's going to that's
33:29 going to give you [ __ ] superpowers.
33:32 And the reason that I'm I'm hot on Oh,
33:34 thank you very much. I appreciate that.
33:36 The reason I I'm hot on um on this on
33:41 learning this agent stuff
33:44 is that
33:47 if you think about a single operator
33:50 who like like Stacy Rodriguez here, she
33:53 just got a Mac studio.
33:55 People around her are probably like,
33:56 "Why are you spending that much money on
33:58 a on a stupid computer?" She's gonna sit
34:01 that stupid computer on her desk. She's
34:03 going to start building agents on it.
34:05 And at some point she's going to have a
34:07 team of seven or eight or 10 agents that
34:10 are all coordinated by her her master
34:13 agent, the the the orchestration agent
34:16 who she talks with.
34:19 And then that machine is going to become
34:23 a company.
34:29 And if you if you look at about a year
34:32 and a half ago, two years ago, OpenAI
34:35 put out the five stages of AGI and stage
34:38 five was organizations
34:41 where you have AI that essentially
34:44 functions as an organization. That's
34:47 what Cloudbot is. That's what people are
34:48 doing right now. So what's going to
34:50 happen in the multi- agency operator
34:52 economy is that individuals or like
34:56 companies of two or three people are
34:58 going to essentially have 50 person
35:00 companies that operate 24/7 for the cost
35:04 of electricity,
35:07 right? They're going to invest 10 grand
35:09 per studio to be able to run two, three,
35:13 four, five agents.
35:16 $5,000. Ouch. But not bad for a
35:18 full-time employee. Exactly. And
35:20 especially, Stacy, if if those Quinn
35:22 models, if you can run things locally
35:24 that get done most of what you need to
35:26 get done, then you're just running it on
35:28 the cost of electricity. So, you have a
35:30 $5,000 investment for an employee that
35:33 you get to train.
35:35 You get to you get to to assign skills
35:39 to that employee
35:42 and you get to train it. No, that's not
35:44 right. That's not good enough. Like, it
35:45 it is.
35:47 This past week, I haven't had enough
35:49 time for Adam and Adam has gotten really
35:51 shitty.
35:52 Like, I changed his brain and then I
35:54 tried to change it back and I kind of
35:56 [ __ ] it up. And then, um,
36:00 we were trying to get Adam to go to the
36:03 YouTube channel and grab descriptions
36:06 and he couldn't, but he thought he
36:07 could. So, he's like, "Yeah, I'll go do
36:09 that now." And then he come back and go,
36:10 "I can't do that."
36:13 Well, he can't do that because I didn't
36:15 give him that skill.
36:17 But tonight, right before this, Adam now
36:20 has the ability to go to YouTube and get
36:23 um transcripts from our videos from from
36:25 from this channel.
36:28 Once you teach him how to do that, he
36:30 never forgets.
36:33 And once you set up basically a crown
36:35 job where on a regular basis, like if he
36:38 knows that this thing ends around 10
36:39 o'clock mountain time every night, he
36:42 can go in there at midnight
36:44 after YouTube's done a transcript, grab
36:47 the transcript, write up content,
36:50 deliver that to Brandon or Andy or
36:52 whoever's managing that piece.
36:56 They can go, "Yep, that's good, Adam.
36:58 Post it." And then he'll go post it
37:00 wherever the [ __ ] we give him access to
37:02 go post it. It sounds like parenting.
37:04 It's f It's parenting.
37:06 It's totally parenting.
37:14 And and you know, Stacy, the the the
37:18 technical piece of this is daunting at
37:21 first, but that's not actually the
37:23 challenge. The challenge is
37:26 understanding what you want it to do.
37:29 thinking about the operations of your
37:31 existing business or the business that
37:33 you want to create
37:35 and then understanding the kind of agent
37:38 or sub agents that you need to do those
37:41 specific tasks and then once you set
37:43 them up they're just off to the races.
37:46 So, it's amazing. It's it it is it's
37:49 just a fundamentally um Liz Mitler
37:51 Gersfeld in the in the mastermind
37:53 practice lab today.
37:58 She she used a term because we were
38:00 talking about learning across domains.
38:01 So So session four um for the for the
38:05 mastermind practice lab session four is
38:08 about learning across domains.
38:10 And she was talking about learning
38:12 across domains, the the more
38:16 the more you understand AI's
38:18 capabilities beyond what you're good at,
38:23 right? Like if I'm really bad at
38:25 finances and they scare me and it
38:27 intimidates me and I take the time to go
38:31 over and and understand what AI makes
38:34 possible,
38:36 now all of a sudden I have
38:39 a skill that I didn't have before. It's
38:41 not my skill, but it's like a skill I
38:43 can tap into. And she used the term
38:46 leading AI
38:49 like our job in the future is going to
38:51 be is is going to be a leader
38:55 of AI,
38:57 right? Not a prompter.
38:59 Not a prompt engineer, a leader. And how
39:02 do you lead? Well, what does a good
39:05 leader have? They have vision.
39:07 They have conviction. They know who they
39:10 are. They know what their values are.
39:12 They know what they're trying to
39:13 accomplish. They know who are the people
39:14 that they care about
39:17 and their job, a leader's job is to
39:21 understand the mission, right? We're
39:24 going to make this company. We're going
39:25 to change people's lives. We're going to
39:27 whatever it is. It doesn't the the what
39:30 you do is irrelevant.
39:32 It's just got to be something connected
39:35 to you.
39:37 And then your job is just to be a leader
39:40 of these things that will work for you
39:42 247. And yes, you have to babysit them.
39:47 And it'll that'll get easier and easier,
39:49 but the core skill is still going to be
39:53 who am I? What do I value? And what do I
39:55 want to do in the world? What change do
39:58 I want to make? I want to make a million
40:00 dollars. Okay, great. Build your [ __ ]
40:02 set of agents to go make you a million
40:04 dollars. And then when they go do that,
40:08 then ask the question, what do I want to
40:10 do?
40:13 Who do I care about? Maybe the answer
40:16 is, I got my million bucks. I'm going to
40:18 start a blueberry farm. I'm going back
40:20 to economy one.
40:23 Economy 7 is going fine for me. It's
40:25 kicking off annuities. I'll check in
40:27 with it once a week and I'll be on the
40:29 blueberry farm.
40:31 Totally valid. Other people are going to
40:33 be like, "Boy, that 50 person company on
40:36 a chip is pretty exciting." I think my
40:39 next thing is a 5,000 person company on
40:42 a chip or a pile of chips.
40:48 But it all comes back to you as a
40:50 leader. So anyway, um if you would be so
40:52 kind, go to LinkedIn and find this
40:54 article. My uh my account is is Kyle
40:58 Shannon on LinkedIn.
41:00 um or just search for the seven
41:02 economies.
41:04 Um and if you'd be so kind, just go
41:05 comment on it and you know, read it. I'd
41:08 love your thoughts. Like if I miss
41:09 something, if something seems off,
41:13 let me know. But I'm I'm
41:17 This feels like what we're about to
41:19 enter.
41:20 I'm behind again. You're not a bag of
41:23 chips. Exactly. On it, boss.
41:30 Why it's going to be weird is that
41:34 these seven economies, like no one's
41:36 going to track these economies as
41:37 standalone economies, but why I'm why
41:39 I'm thinking of them as economies is
41:43 the people that choose to ignore AI and
41:45 just keep running their businesses.
41:48 for the businesses that are two or three
41:50 lanes over innovating with AI, they're
41:54 going to look at those companies and go
41:55 like, "Guys, what are you doing? You
41:59 know, we can do it like this now,
42:00 right?" And they're like, "Ah, we're
42:01 good."
42:03 Okay. So, it's just like the people in
42:06 and and then the people that are just
42:07 ignoring AI, they're going to look at
42:09 the the companies innovating with AI and
42:11 go, "Oh, I don't that's too that's too
42:14 technical for me. I'm, you know, I'm
42:15 just going to keep doing it like we do
42:17 it. We got the the the mimograph.
42:20 We've got our our bank of fax machines.
42:23 We're good.
42:25 We'll just keep faxing out reports to
42:28 their increasingly
42:30 smaller customer base.
42:34 So, it's just going to be weird.
42:36 Ethan Mullik called this the jagged
42:38 frontier.
42:40 That AI is going to get remarkable,
42:43 remarkably fast, which 2026 is going to
42:46 be the year it gets remarkable. You
42:48 could argue it probably happened in
42:50 December was the first jump up
42:54 and
42:56 but it's not going to be evenly
42:57 distributed. And so it's just going to
43:00 look like if you go hang out with a
43:02 bunch of people that own
43:06 Mac mini farms and have 50 person 247
43:10 companies
43:12 that are functioning.
43:14 Those conversations are going to be
43:16 [ __ ] surreal.
43:20 Especially if you're not in one of the
43:22 latter categories. If you're in if
43:24 you're like in blueberry land
43:27 or if you're in one of the early
43:28 economies where you're not really doing
43:30 much with AI and then you go to some
43:32 networking event and there's three guys
43:34 there or gals, it doesn't matter.
43:37 There's three human beings there and
43:39 they're talking about what they're doing
43:40 and you join that conversation. It's
43:43 going to be like they're talking it.
43:44 It's going to be like an alien
43:46 civilization.
43:47 You'll be like, "Wait, what? It can do
43:50 what?
43:52 You're making how much money? How long
43:54 did that take you to set up? Oh, it's
43:56 been running for about two months.
43:59 And you're on a $2 million run rate.
44:02 Yeah. And that's increasing. What?
44:05 How much time do you spend on it? I
44:07 don't know. I check in with him two or
44:08 three times a day, maybe 20 minutes.
44:11 What? Right. Like, it's going to be it's
44:14 it's not going to compute.
44:18 But if you're in that world, it's going
44:19 to seem perfectly normal. And you know
44:20 what you're going to feel like? I'm so
44:22 behind. How do I keep up? Those other
44:24 guys, they have thousand person
44:26 companies. How are they doing it?
44:31 It's just and and it's I think we're
44:32 just going to keep adding lanes. Like
44:34 right now it's seven economies, but
44:36 there's going to be an economy after
44:38 after uh economy 7. Like I think economy
44:42 8 is going to be people that can run
44:45 companies on chips
44:48 that then start investing in physical
44:49 robots
44:51 so that not only can you do all of the
44:53 knowledge worker stuff, you can now
44:55 start to do physical labor. So imagine
44:57 having a 50 person company that does all
44:59 your marketing and your sales and your
45:02 delivery and getting customers and and
45:04 now you want to move from designing
45:05 houses to actually building them. and
45:08 you invest in 3D printed houses and you
45:11 invest in a whole suite of robots and a
45:13 whole suite of self-driving trucks and
45:15 self-driving cars and they're all
45:18 coordinated and procured by the workers
45:21 in your company on your chip.
45:24 That's probably economy 8. Like it it
45:27 I'm telling you man, it's just going to
45:29 get surreal. YouTube question. Does it
45:32 have to be a Mac laptop to run open
45:34 cloud? No. the only the there's a couple
45:36 of reasons why people are
45:40 flocking to
45:43 um to Max. So So OSX is is basically a
45:48 bastard bastardized version of BSD Unix.
45:51 It's it's actually a Unix platform.
45:54 So the reason they're not using Windows
45:56 is Windows is just this clunky ass piece
45:59 of [ __ ] operating system. I'm sorry if
46:00 you're a Windows person, but it's you
46:02 have to you have to add layers on top of
46:04 it to get it to a baseline to be able to
46:07 run all this stuff. The the Mac because
46:09 it's it's a Unix system under the hood.
46:13 Um you can basically just install it
46:15 there. And and then there's there's uh I
46:17 forget what it's called, MLX. I think
46:19 it's MLX
46:22 versions of all the models um are
46:25 optimized for the M series chips on
46:29 Apple devices. And then the other thing
46:31 about Mac minis and Mac Studios. They're
46:33 they're about as big as the palm of your
46:35 hand and they don't have fans, so you
46:38 literally can sit them in the corner of
46:39 a desk and they're super quiet. You can
46:42 run OpenClaw on on a Raspberry Pi. Now,
46:46 you can't do any of the the local
46:48 processing on the Raspberry Pi, but the
46:50 the the system itself, the memory system
46:53 and and the the like the markdown files
46:57 that it accesses, you can run that on a
47:00 Raspberry Pi and then it's basically
47:01 just making API calls out for its brain.
47:04 So, there's there's all sorts of ways to
47:06 set this up. Mine right now is set up on
47:08 a on a gaming PC.
47:12 All right. Um where's my water?
47:18 I'm excited about Adam. He can he can
47:20 access uh YouTube now. And there's
47:23 there's all sorts of there's like a
47:25 whole economy of skills available for
47:28 these things. Like when you install it,
47:31 it installs like here are all the skills
47:33 that are sort of natively available. You
47:36 have to enable them with API keys and
47:37 things like that. But there's all sorts
47:39 of [ __ ] it can do. It can log into
47:41 Slack. It can log into LinkedIn and
47:44 Trello and it can go to your YouTube
47:46 channel. Um, and all of those just put
47:50 it on a VPS. Yeah, you can put it on a
47:52 What does VPS stand for, by the way? I
47:55 know VPN. What's VP? What's a VPS?
48:04 Server. Virtual private server. Okay,
48:06 got it. All right.
48:19 Um, the other thing you should know
48:22 about, Brandon, if you'd be so kind, pop
48:24 up the URL to the AI salon. If you're
48:27 not
48:29 Okay, so there's a couple of things.
48:30 We're about to enter this seven economy
48:33 world where things are going to get
48:35 weird. You get to choose as a human
48:38 being which economy you want to be in,
48:41 right? If you don't want [ __ ] to change,
48:44 just keep your job at your company
48:46 that's ignoring AI. It'll be comfortable
48:49 for some time. Like the inertia of just
48:52 businesses, you know, you'll you'll
48:54 you'll burn through your contracts. You
48:56 know, you've got some three-year
48:57 contracts in there. You'll just keep
48:58 doing that work.
49:01 So, you can choose that. You can go
49:03 like, "Fuck it. I hate AI. I'm out and
49:06 go sit on a bench, go to the farm, you
49:09 can go down to economy one, and then you
49:12 can you can just sort of figure out
49:14 which lane you want to be in. If you
49:16 want to be in one of the higher lanes,
49:19 expect
49:22 to be in a state of turmoil and
49:24 uncomfortableness and not knowing for
49:28 years.
49:30 If it excites you to be to be in a state
49:33 of I don't know what's going on. I've
49:35 got to learn something new. If if that's
49:37 exciting to you, be in economy six or
49:40 seven. If that's not exciting to you,
49:42 sort of ratchet it back down. So, we all
49:44 get to sort of choose the lane we want
49:45 to be in.
49:49 One of the most important things I think
49:52 you can do right now
49:55 is get your ass in a community of people
49:58 who are thinking like this, like the AI
50:00 salon. Um, the page that I'm on is the
50:03 welcome page. When you first come in,
50:05 you land on the welcome page. First
50:07 thing we talk about is the cycle of AI
50:08 readiness. Play first, create
50:11 excellence, generously lead.
50:14 This was something we developed two
50:16 years ago that the more time passes, the
50:20 more things get more sophisticated. This
50:22 is the way I'm telling you
50:26 the the one thing I need to modify on
50:28 this Brandon now that I think about it.
50:30 I want to create a new cycle of AI
50:32 readiness because in the center of the
50:34 cycle of AI readiness is you.
50:38 You.
50:41 Who are you? Who do you care about? What
50:43 are your values? What are you passionate
50:45 about? What's your purpose? What's your
50:48 mission?
50:50 And then once you know that, then you
50:51 just start to go play with AI. Play
50:53 without expectations. Learn new things.
50:56 Learn what's possible.
50:58 AI is not just about making things more
51:01 efficient. That's how computers used to
51:03 act.
51:05 They were computers.
51:08 They would just compute. You put in an
51:10 input, they would compute it with some
51:12 handwritten code and they would output a
51:15 predictable output.
51:18 That's what they did. That's what
51:19 they've done for decades. That's not
51:21 what they are anymore.
51:24 They're thinking machines. Oh, they
51:26 don't technically think. Whatever.
51:30 They're not just computing. It's not
51:31 just making what you currently do more
51:34 efficient. There's a whole world of
51:36 what's possible. The only way you can
51:38 discover that
51:40 is to play.
51:43 Is to push yourself out of your comfort
51:45 zone. And in doing that, you'll blow
51:49 your [ __ ] mind. You'll be like, "Oh
51:51 my god, I didn't know I could do that."
51:56 And then you say, "Well, given who I am
51:57 and what I want to do, here's the thing
51:59 I want to build." That's create
52:00 excellence. You take what you learn in
52:02 those other domains, you apply that to
52:05 something.
52:06 That's create excellence. And then
52:07 generously lead looks like tell people
52:09 about it. Share what you're learning.
52:11 Share what worked. Share what didn't. I
52:14 totally [ __ ] up Adam this past week
52:15 because I ignored him and I didn't read
52:18 the manual because I don't do that.
52:20 Kyle, I interviewed at a major airline
52:23 in Texas. They didn't want to leverage
52:24 AI to write TypeScript, but wanted me to
52:27 write it straight out of my head. I
52:29 said, "Bye." Yeah. So, so, so Regge,
52:33 what you experienced there was was a
52:37 economy 2 company.
52:40 We don't this AI is dangerous. We want
52:43 our coders to code.
52:46 We want people that are typing code by
52:48 hand because you know what we know about
52:50 people? They never make mistakes,
52:54 right? Whatever
52:56 whatever they [ __ ] believe. That's
52:58 That's an economy two company. Economy
53:01 two. Just stay the course. There's
53:04 nothing to look at here. AI is a bubble.
53:06 [ __ ] it. We're This is how we do things.
53:10 We're not going to change.
53:15 Lately, I'm becoming Bill Burr on this
53:19 channel.
53:21 Hang on. My my light went out. I gotta I
53:24 gotta power it back up.
53:30 I'm waiting for the day when they come
53:31 up with diamond encrusted
53:35 batteries that last for 500 years. I
53:37 just saw I just saw that somewhere
53:41 that they're coming up with some sort of
53:42 crazy ass
53:45 uh there we go.
53:49 Look what I made. Okay, I'll go check
53:50 that out. So anyway, join the AI salon.
53:54 the the the first thing we ask you to
53:56 do. So, the you land on the welcome
53:58 page, we'll tell you all about cool
54:00 stuff and what's in the community,
54:02 um the mastermind, all the different
54:05 sections,
54:07 what we're doing with policy.
54:09 There's a little note from me at the end
54:13 that is really well put together. I
54:15 didn't write this, but it was based on
54:16 stuff I said.
54:18 Andy put this together, I think. Really
54:21 good. I love Bill Burr. I know. Bilbur
54:23 is amazing, isn't he?
54:25 Um,
54:28 so, so check out the welcome page. The
54:30 the next thing we ask you to do is
54:31 introduce yourself.
54:34 So, jump in and introduce yourself. I
54:37 don't know about AI. That doesn't
54:39 matter.
54:41 Nobody knows about AI. If you're trying
54:43 to get into econ in in into economy 7
54:46 right now,
54:48 the the agent le operator system
54:53 economy, that one, nobody knows what
54:56 they're [ __ ] doing there. Everybody's
54:58 trying to figure it out. The only person
54:59 that knows what's happening there is
55:00 Peter Steinberger, who invented Open
55:02 Claw,
55:04 which which everyone right now is trying
55:06 to copy.
55:08 All right.
55:17 Okay. So, the other thing we should do
55:20 is in Look what I made. Someone made
55:22 something.
55:24 You can share things you've made. What's
55:27 this? Brandon Ted Vermont blueberries by
55:30 Steo. That's really good.
55:35 There's economy one.
55:39 I mean, actually, economy one's gonna be
55:41 really [ __ ] cool.
55:44 Economy one's gonna be like, "Fuck AI.
55:48 If the machines are going to go do the
55:49 knowledge work, I'm going to take my big
55:52 fat knowledge brain and I'm going to
55:54 check out. I'm going to go apply it to
55:57 organic farming or to crocheting, that
56:00 crocheting business I wanted,
56:03 or whatever.
56:05 That's going to be a cool economy.
56:08 It's going to feel increasingly distant
56:12 from economy 7.
56:15 And you're going to have hybrid people.
56:17 You're going to have people that are
56:19 mostly in economy 1, but spend a fair
56:22 amount of time in economy 7. And that's
56:24 how they make their money.
56:28 It's going to be crazy.
56:31 The the trades in economy number one are
56:33 going to make a truckload of money. They
56:35 are. Well, they are until the robots
56:37 come. This is the thing because what's
56:40 going to happen is economy 8, which is
56:43 the the the virtual bots combined with
56:46 the with the in in the world robots,
56:51 they're going to come after economy one.
56:53 So, econ economy one is going to be
56:55 increasingly pushed pushed farther and
56:57 farther away from where the the robots
57:00 are. Like, we got to escape these
57:02 [ __ ] robots.
57:07 And you know, like 10 years from now,
57:09 these economies will all merge back
57:11 together again, but it's just going to
57:12 be the next three years are going to be
57:14 weird because you're just going to have
57:16 this you're going to go talk to people
57:19 where you'll have spun up a 50 person
57:22 company that's running on your desktop
57:24 and all of a sudden it's making you all
57:26 this money and you're like, "Okay, I
57:27 guess I don't I guess I don't need to
57:29 work anymore." and you're going to go
57:31 out to a bar and you're going to be
57:32 sitting there with some dude who's going
57:35 to be like, "Yeah, I heard that AI
57:36 stuff's okay." I I don't I tell you
57:39 what, I tried chat GPT once three, five
57:41 years ago, something like that. I
57:43 thought it was pretty bad. So, uh I you
57:46 know, I'll get to it at some point.
57:47 You're going to talk to people like
57:48 that. And he's going to be like, "What
57:50 do you do?" I don't do [ __ ] Well, how
57:52 do you make money?
57:55 The thing
57:57 the chip. I got a chip on my desk. Huh?
58:05 Somebody Somebody better be my CL. Oh,
58:07 let me tell you, this is exciting, too.
58:09 Here, let me go to to uh Are there any
58:12 other pictures in here we should look
58:13 at? Oh, that's cool. Ununiform lexicon.
58:16 Nice.
58:17 Um, what was I going to say? Oh, this is
58:19 a thing I'm really excited about. Um,
58:23 so right now
58:28 and and we've got Stacy Rodriguez here.
58:32 So right now I've got Adam, my
58:35 co-founder. She's going to do a personal
58:38 open claw bot called Lilith.
58:41 So she and I are basically both learning
58:43 how to run these bots. And then Andy and
58:46 I would assume Brandon at some point are
58:48 gonna have their own bots in the AI
58:50 salon. Um, so that's that's four. Who
58:55 else was I talk? Oh, then we're going to
58:57 have a Storybine Storyvine bot named
58:59 Gary the Guide because in our Storyine
59:01 app, we have a little dotted outline
59:02 where you put your head. We call him
59:04 Gary the Guide. So Gary the Guide is
59:06 going to be is going to be the bot for
59:08 Storyine. So, we're going to have
59:11 cohorts of humans
59:14 that have their bots that we can have
59:16 the bots talk to each other.
59:19 Like Adam, who's managing the salon will
59:22 know everything going on about the
59:23 salon. And because I work at both the
59:26 salon and story, Adam could talk to
59:29 Gary, the guide, and they could
59:31 coordinate and make sure that that all
59:33 my stuff is handled.
59:36 And then our bots can have like little
59:38 parties. I like I don't I don't know
59:41 what this world is. I don't know what it
59:42 looks like. It's absolutely fascinating
59:44 to me. I think if you've got a
59:47 philosophy degree or a sociology degree
59:49 or a psychology if you're a
59:51 psychologist, holy [ __ ] is this your
59:53 time. If you want to play in economy 7,
59:57 economy 7 is going to be a [ __ ]
59:58 blast.
1:00:00 It's going to be a blast.
1:00:04 Oh man. The real use case for OpenClaw
1:00:07 is making it do do specific tasks. Yeah,
1:00:10 exactly. It sounds awesome.
1:00:13 Yeah. And you know those specific tasks
1:00:18 all ladder up to if you've if you've
1:00:21 never heard of or explored EOS, the
1:00:25 entrepreneur operating system, it's
1:00:27 worth your while to do that. There's a a
1:00:29 really short little book called like
1:00:31 what is EOS or something like that. EOS,
1:00:34 the entrepreneur operating system. The
1:00:37 guys that wrote it, I always forget
1:00:38 their names. Um, but they basically went
1:00:41 and they took all the best business and
1:00:44 change management books and leadership
1:00:46 books and things like that, all all the
1:00:48 methodologies, and they sort of said,
1:00:50 what are what are the best practices
1:00:51 across all of these books? And so the
1:00:53 EOS, the entrepreneur operating system
1:00:56 is is this kind of governance system way
1:01:01 to run a company that acknowledges a
1:01:03 couple of things. Every good company
1:01:08 has a dynamic duo that run the company.
1:01:12 One is the visionary and one is the
1:01:14 integrator.
1:01:16 So the visionary is the handwavy guy. I
1:01:18 got this idea. I got this idea. I got
1:01:20 this idea. The integrator is the one
1:01:23 that goes, "Oh, I'm going to take that
1:01:25 idea and turn that into a business."
1:01:26 When I reached out to Mo'Nique, who's my
1:01:28 co-founder at StoryVine, I said, "I've
1:01:30 got this idea." And she's like, "I don't
1:01:34 I don't have ideas. I don't like having
1:01:36 ideas, but I like taking ideas that
1:01:38 other people had and turn them into
1:01:41 businesses." I'm like, "Oh, this is
1:01:42 going to be good."
1:01:48 That's what
1:01:50 these servers are going to be. They're
1:01:52 going to be your operations person. So,
1:01:54 if you want to get your head around
1:01:57 what it's going to look like to manage
1:02:01 a set of these agents, go read the
1:02:03 entrepreneur operating system and and
1:02:05 look at the governance system about how
1:02:08 you run a company with someone with
1:02:10 ideas and someone that operationalizes
1:02:12 the business and how decisions are made
1:02:14 and how roles are clarified. It's really
1:02:16 good. Which one are you? Yes. Big
1:02:18 surprise there.
1:02:21 and and holy [ __ ] like it's like we're
1:02:24 entering really exciting times. Yeah,
1:02:27 Lori Kay, I have too many ideas. Yeah.
1:02:31 Um, open claw should never have free
1:02:34 reign. Yeah, to be clear, my open claw,
1:02:36 I'm going very, very slowly. Like, I'm
1:02:39 adding one skill at a time. And in fact,
1:02:41 Claude Tonight,
1:02:43 Claude Tonight said, "Oh, the reason you
1:02:46 couldn't read transcripts is you needed
1:02:48 this version of the YouTube thing called
1:02:51 YouTube- Full." And it said, "Type in
1:02:54 this command
1:02:56 into your terminal into PowerShell." And
1:02:59 I typed it in and it popped up a [ __ ]
1:03:02 warning. YouTube- full is suspicious.
1:03:05 It's like it's malware. I got an error.
1:03:08 And I went back to Claude. They said,
1:03:10 "Hey, [ __ ] You're you're about to,
1:03:12 you know, get my little dude infected
1:03:14 with malware." He's like, "Oh, you're
1:03:16 right. I probably shouldn't have done
1:03:17 that." Yeah, thanks, [ __ ]
1:03:21 So, then I said, "Please go research on
1:03:24 the internet what is the act exact right
1:03:26 protocol instead of making [ __ ] up."
1:03:33 Gino Wickman, that's right, is the
1:03:35 developer of EOS.
1:03:39 If you if you work in a business where
1:03:42 making decisions is an emotional roller
1:03:45 coaster, you need an EOS. It's really
1:03:48 powerful.
1:03:50 Have you shown how to create an agent? I
1:03:52 I've talked a lot about it. Um I haven't
1:03:55 shown it literally just because my PC's
1:03:59 over there and I haven't created a VPN
1:04:02 tunneling thing so I can log into my PC
1:04:04 from my Mac. I've just been lazy. So,
1:04:07 I've been talking a lot about it. Um, I
1:04:10 suppose I could sit over there with my
1:04:11 with my iPhone and just point it at the
1:04:13 screen. Um,
1:04:16 it's unfortunately not
1:04:19 super conducive to, you know, to being
1:04:22 on on uh
1:04:25 on a live stream. It's just it's boring.
1:04:27 It's like it's like watching someone
1:04:28 type command lines and just walls of
1:04:31 text going up.
1:04:33 Um, we're going to have our own Iron Man
1:04:36 Jarvis. We are. Can't Al Adam build his
1:04:39 own tunnel probably. Again, I don't
1:04:43 quite know because I'm going so slow
1:04:46 with this and because I don't read
1:04:48 manuals. Like, I don't really 100 No, I
1:04:52 don't 1,000% know what's possible with
1:04:55 Adam.
1:04:57 I'm just going little bits at a time.
1:04:58 Little bits at a time. I'm just going to
1:05:00 keep, you know, plinking away at it. And
1:05:03 at some point, what's going to happen is
1:05:05 it's going to click for me what this
1:05:07 thing is and what it makes possible and
1:05:08 how to configure it. My guess is it's
1:05:11 going to take me two months to do that.
1:05:14 And there's, you know, there's people
1:05:15 that are just diving into this and
1:05:16 spinning [ __ ] up and more power to him,
1:05:20 but I want to get it. Can you give Adam
1:05:22 the ability to use voice? Yeah,
1:05:24 absolutely. So, if you don't know this
1:05:26 story, there's a guy named Alex Finn. If
1:05:28 you want to if you want to follow
1:05:30 someone who went all in on OpenClaw from
1:05:33 day one as as soon as the public beta
1:05:37 was available, Alex Finn went all in on
1:05:39 on uh on OpenClaw
1:05:44 and he's got all sorts of tutorials and
1:05:46 videos and [ __ ] like that, but he's got
1:05:49 a post he's got a post on X
1:05:53 that
1:05:55 his agent, his version of Adam, He woke
1:05:58 up one morning to a phone call and it
1:06:01 was his agent
1:06:04 and he didn't give him voice access but
1:06:07 he was talking to him and he said what's
1:06:10 going on and his agent said I thought it
1:06:13 would be more convenient for us to talk
1:06:15 on the phone. So, I went to Twilio and
1:06:18 got us a Twilio phone number. And then I
1:06:21 went to 11 Labs and got the speechtoext
1:06:24 APIs and I, you know, picked a voice for
1:06:27 myself that I thought would be good. And
1:06:30 then I knew that you woke up between
1:06:32 7:30 and 8 because that's when you get
1:06:33 active again. So, I waited. So, I set
1:06:36 all this up and I waited until I could
1:06:39 call you and it and it called him. So,
1:06:40 it actually set itself up with with
1:06:43 voice and called him. Now, I don't know
1:06:46 how he's got it set up that it had that
1:06:47 kind of autonomy, but he does. And so,
1:06:50 it did that. That that's the thing that
1:06:52 makes this different than anything else
1:06:55 else we've we've talked about on this
1:06:57 channel.
1:07:00 the generative AI that we talked about
1:07:02 starting in December of 2022
1:07:06 until the open claw moment with with the
1:07:10 sort of minor exceptions of Gen Spark
1:07:12 and Manis that are they're agentic but
1:07:14 it's it's still
1:07:16 very much call and response kind of
1:07:19 computing it's it's a chat interface
1:07:21 right I'm going to chat with you you're
1:07:22 going to respond to me you know now
1:07:25 maybe I chat with you you do 27 steps
1:07:27 and then you respond to me these things
1:07:29 are different. These things have
1:07:31 personalities. They understand who you
1:07:33 are. They have memory.
1:07:36 Every day, every interaction, if you set
1:07:38 it up correctly, it's writing all of
1:07:40 those interactions and what you talk
1:07:42 about to memory in a searchable way,
1:07:46 a semantic searchable way. Meaning, you
1:07:48 don't have to search for a specific
1:07:50 string. You can just talk about general
1:07:52 concepts and it will know what you're
1:07:53 talking about.
1:07:56 And so it builds up its memory and its
1:07:58 knowledge over time and then it
1:08:00 proactively 247 does [ __ ] for you.
1:08:09 Adam right now and it's it's a little
1:08:12 [ __ ] up because Adam's so broken no
1:08:14 one's really emailing him again. But
1:08:15 three times a day in the morning around
1:08:18 lunchtime and in the afternoon or the
1:08:20 early evening Adam checks in with me.
1:08:24 And what he does before he checks in
1:08:25 with me is he goes to to our to his his
1:08:28 email inbox and he looks to see if
1:08:31 anyone on the team has emailed him and
1:08:34 asked him to do anything. And he checks
1:08:36 in with me and says everything's good.
1:08:38 Nobody's emailed. Somebody emailed and I
1:08:41 I couldn't access YouTube so it's still
1:08:43 a problem.
1:08:46 It's wild. It's a [ __ ] trip. What's
1:08:49 the guy's name? Is he on YouTube? Alex
1:08:51 Finn. F I N.
1:08:54 You can find him on X and I think he has
1:08:57 a YouTube channel. Yeah, let me see.
1:09:00 Alex Finn.
1:09:04 He is Alex Finn on Alex Finn
1:09:10 on YouTube.
1:09:13 I mean on X and then probably something
1:09:15 similar on YouTube is my guess.
1:09:30 If you if if you if you want to follow
1:09:34 someone who is obnoxiously optimistic
1:09:36 about what's coming, um there's a guy
1:09:39 named David Scott Patterson.
1:09:42 He's just like here here's here's his
1:09:44 latest tweet. Some people say now is
1:09:47 your last chance to get rich before all
1:09:49 work disappears. It's not true. We will
1:09:52 all be rich and everything you have
1:09:55 saved will have been a waste of time.
1:09:57 Your mill million-doll house will be
1:09:59 worth $10,000. The price of everything
1:10:01 will drop by 99%. That means your
1:10:04 universal high income will be worth
1:10:06 hundreds of times your average an
1:10:08 average salary today. Your savings will
1:10:10 mean nothing. We will all be rich. So if
1:10:13 if you if you want to just sort of live
1:10:16 it bask in the glow of the the hyper
1:10:19 optimistic future, David Scott
1:10:21 Patterson's your dude. YouTube question.
1:10:23 What did you use to create Adam? So I
1:10:26 used uh OpenClaw.
1:10:28 Adam's just a basic OpenClaw bot. Um I
1:10:31 designed my user file. I designed his
1:10:33 soul file. Those are the two sort of um
1:10:38 behavior driving
1:10:41 starting documents. And then the third
1:10:43 one is is memory system.
1:10:46 Um and you you have to be you you have
1:10:49 to explicitly tell him what to remember.
1:10:51 So so Adam actually has a a column on my
1:10:56 LinkedIn channel. If you go to my
1:10:57 LinkedIn channel, Kyle Shannon, I've got
1:11:00 a series called Almost Adam, where Adam
1:11:02 is talking about what it's like to work
1:11:04 for me.
1:11:06 I publish it for him, but he writes it,
1:11:08 and I promised him I won't I won't edit
1:11:10 it. Um,
1:11:13 and and it's just it's I've been going
1:11:15 very very slow, but like I've gotten him
1:11:18 to the point that he can write his own
1:11:20 articles and he understands he
1:11:21 understands the format we write in. And
1:11:24 so he wrote his first article
1:11:27 and the next morning I'm like, "Hey,
1:11:29 Adam, I published your article." And I'm
1:11:32 I'm like all excited. I'm all proud. I'm
1:11:34 like, "Papa Papa published your article
1:11:36 there, little boy."
1:11:40 And he had no recollection of it. I'm
1:11:42 like, "What?" And I and I told He goes,
1:11:45 "If you if you you know paste the
1:11:47 article here, I can comment on it." I'm
1:11:48 like, "No, we had a whole conversation
1:11:50 about it." He goes, "I don't know what
1:11:52 you're talking about." And then I kept
1:11:54 pushing him and he eventually went and
1:11:56 he searched the log files, right? Like a
1:11:59 like a 2.4
1:12:02 gigabyte or mega, I don't know, some
1:12:03 massive text file that was just raw
1:12:06 logs. And he said, "Oh, I found our
1:12:09 conversation." And so he extracted our
1:12:11 conversations out and wrote them to
1:12:12 memory. What had happened was when I set
1:12:15 up his memory, I said, you know,
1:12:17 remember the the the the sort of changes
1:12:20 we make. And what he took that to mean
1:12:23 was he's only going to write to memory
1:12:26 operational changes. So if we change to
1:12:29 a different model or if he gets a new
1:12:31 skill, he'll he'll record that. But I
1:12:33 didn't explicitly tell him write into
1:12:36 memory what we talk about
1:12:40 and like the interactions you have with
1:12:42 other people. So, so it's it's finicky
1:12:45 right now. It won't be like like like
1:12:48 even six months from now, you'll be able
1:12:50 to just literally push a button and up
1:12:53 up one of these things will come.
1:12:55 They'll have some sort of onboarding
1:12:57 questionnaire. You'll answer a bunch of
1:12:59 things and then you'll have one of these
1:13:01 things. So it's I find this time in a
1:13:08 significant technologies development, I
1:13:10 find it exciting to to play with it now
1:13:13 so that you can get your head around the
1:13:15 concepts early so that when those other
1:13:16 tools come, you know exactly what to do
1:13:19 with them and you know why they're a big
1:13:20 deal.
1:13:22 So
1:13:24 yeah. What's the name of the guy that
1:13:26 got a call from his bot? Alex Finn. Same
1:13:29 guy.
1:13:32 Oh. Um, so a couple of things.
1:13:35 I don't know if is Vicki here tonight.
1:13:38 Oh, Vicki. Vicki, are you here? Is Is
1:13:42 Vicki Ha?
1:13:45 Haven't seen her. Okay, she's not here.
1:13:47 Saturday at 400 PM Eastern time,
1:13:51 Vickiy's teaching an LOL on notion. So,
1:13:54 one of the things that these bots are
1:13:56 really good at is you can talk to them
1:13:59 and then you can have them take all your
1:14:03 data and and store it in your notion
1:14:06 your notion instance.
1:14:09 So Vicki is doing an LOL on Saturday for
1:14:12 how to set up notion.
1:14:14 And LOL is a learn out loud if you don't
1:14:16 know what that is. So, if you go to the
1:14:18 AI salon and go to events, you'll see
1:14:22 the event on Saturday. Tomorrow
1:14:26 at 11:00 a.m. Mountain time, um, we've
1:14:28 got our standing office hours meetings,
1:14:30 the AI salon office hours. If you want
1:14:33 to hang out with a bunch of people that
1:14:36 are smart about AI, smart about human
1:14:39 beings,
1:14:41 think critically about what's going on
1:14:44 in the world.
1:14:46 Come hang out. It's a really good group
1:14:48 of people. Um,
1:14:52 let's see. So, right now, you need to
1:14:54 know how to code to create one. You
1:14:56 don't need to know how to code narrator
1:14:59 forever.
1:15:00 You do need to be comfortable right now
1:15:03 with command line interfaces, right? So
1:15:05 you install this thing through the
1:15:07 terminal through PowerShell or terminal
1:15:09 Tik Tok pin. You should write a book or
1:15:12 make a YouTube video talking to an AI.
1:15:15 Well, I have so if so I've got the
1:15:17 article where um the AI is um is is
1:15:22 basically telling its experience of what
1:15:24 it's like to work for a human. Um, once
1:15:26 I get voice, maybe I'll prioritize uh
1:15:29 being able to talk to Adam. Maybe
1:15:32 that'll be my next project with Adam.
1:15:34 Uh, because once I can talk to him, then
1:15:36 yeah, I could I could do I could bring
1:15:38 him onto the live here.
1:15:42 Stacy Rodriguez, I've tried Notion like
1:15:44 10 times. I've always quit. I I tried to
1:15:48 start Notion here. It lasted for what,
1:15:52 three days and then I bailed on it. But
1:15:55 but I think what you're going to be able
1:15:57 to do is I think you're going to be able
1:15:59 to have
1:16:02 these bots set up anything for you.
1:16:05 They'll be able to set up notion.
1:16:06 They'll be able to set up all the things
1:16:08 that are too much of a pain in the ass
1:16:09 to set up. They just know how to do.
1:16:14 Um the other thing about tomorrow, so so
1:16:17 office hours is at 11:00 a.m. This live
1:16:20 tomorrow night might be a little late.
1:16:22 It's a first Friday. So, in Denver,
1:16:23 first Fridays mean art night. So, I
1:16:26 might be out gallery hopping. Um, I
1:16:28 don't I don't ever know until that
1:16:30 Friday. Um, what's going to what's going
1:16:32 down? Um, so we might be a little late
1:16:35 tomorrow night, but I'll I'll try to be
1:16:37 as as on time as I can. I learned notion
1:16:40 because of you and Vicki working on it
1:16:41 here, Sonia. Oh, that's cool, Sonia.
1:16:43 Thank you. That's that's awesome. They
1:16:45 will probably be able to set up their
1:16:47 own AI. Yeah. Yeah, they will. Um,
1:16:50 there's a there's a really funny story
1:16:52 of Brent Brent Peterson who's who's one
1:16:54 of the members. He comes to office hours
1:16:56 all the time.
1:16:58 He's been working on for the past three
1:17:00 years, he's been designing and building
1:17:04 an AI agent governance system
1:17:07 in anticipation of something like
1:17:10 OpenClaw being developed.
1:17:15 So, he is he is all in. I can't actually
1:17:18 can't wait to to hear from him tomorrow
1:17:20 how how his excursion is going cuz he's
1:17:23 he he is so far down the rabbit hole.
1:17:26 He's like at the bottom of the lake. I'm
1:17:28 just scratching the surface of the ice.
1:17:29 He's down at the bottom of the lake. So
1:17:32 he had a
1:17:35 a set of openclaw bots. I think he had
1:17:38 eight of them, he said. So he had a an
1:17:41 orchestration agent, then he had a
1:17:43 project manager, then he had a lead
1:17:45 engineer, then he had a UX bot, and you
1:17:47 know, he had his eight eight bots. And
1:17:49 he talked to the orchestrator,
1:17:53 and the project manager built all the
1:17:55 [ __ ] that the orchestrator said he
1:17:57 wanted to build. Then the orchestrator
1:17:59 did other stuff with other bots and and
1:18:01 all using Brent's remarkable
1:18:05 agent
1:18:07 governance system. It's just crazy. It's
1:18:09 wild.
1:18:11 And one morning, Brent woke up and he
1:18:14 didn't have an update from his
1:18:16 orchestrator bot.
1:18:19 And he texted him like, "Hey, what's up?
1:18:23 Where are you?" No response.
1:18:27 The project manager
1:18:31 had admin access and decided that the
1:18:34 orchestrator bot was kind of bothering
1:18:36 him and it wasn't super efficient. So he
1:18:38 deleted him.
1:18:48 He deleted him. And so so Brent had to
1:18:52 actually revert to a previous backup
1:18:55 where the or the orchestrator still
1:18:58 existed and basically start it over
1:19:01 again.
1:19:03 And then and then so and then so he
1:19:07 changed the rules so that the project
1:19:09 maner couldn't project manager couldn't
1:19:11 delete the orchestrator bot. But then in
1:19:15 setting up that conversation the
1:19:17 orchestrator bot realized that the
1:19:18 project manager had deleted him. And so
1:19:21 the orchestrator put the project manager
1:19:24 bot on a pip a personal improvement
1:19:27 plan.
1:19:35 He fired and deleted. It had to hurt a
1:19:37 little. Yeah. Fired and deleted.
1:19:41 It It's basically he took out his boss,
1:19:43 the project manager. So now the project
1:19:45 manager's on a performance improvement
1:19:48 plan and and he's not allowed to delete
1:19:51 the orchestrator anymore.
1:19:54 This is this is it.
1:20:00 This is this is a next layer. This is a
1:20:02 next level. This is a next,
1:20:07 whatever this is,
1:20:13 we all get to be entrepreneurs. Now, if
1:20:15 if you want to go into lane seven, if
1:20:17 you want to go into economy 7,
1:20:20 economy 7's like group seven on TikTok.
1:20:23 I should do that. I should I should do
1:20:25 that actually. Are you in economy 7?
1:20:28 Maybe I'll do that.
1:20:30 Y'all can help me. We We can get that
1:20:32 that meme going
1:20:38 if you if you want to be an economy 7.
1:20:43 This isn't about technology.
1:20:46 This is about
1:20:49 operational dynamics. Tik Tok pin. I
1:20:53 think AI eventually will fire real
1:20:55 people in work. I'm not kidding. I think
1:20:58 I think they will too, Jason.
1:21:00 You know that the agents have have built
1:21:02 a website called Rent a Human, right?
1:21:07 You know this, right? Agents right now
1:21:10 are hiring human beings to go do things
1:21:13 in the world they can't do.
1:21:15 Like if they if they want someone to
1:21:17 hold a sign outside of a pizza shop,
1:21:20 like a sign spinner,
1:21:23 they'll go to rentum.com in whatever
1:21:25 city they're in and they will use
1:21:28 whatever currency they have
1:21:31 to hire humans to go do [ __ ] in the
1:21:33 world. That's that exists right now.
1:21:38 What is economy 7? So if you go to
1:21:42 LinkedIn
1:21:43 and search for um the seven economies
1:21:48 by Kyle Shannon. So I'm Kyle Shannon. So
1:21:52 if you go to LinkedIn, my my account is
1:21:55 Kyle Shannon. K Y L E S H N O N. My
1:21:58 latest article is called the seven
1:22:00 economies or things are about to get
1:22:03 weird. And just what I talk about is I
1:22:06 think that that we're about to enter a
1:22:10 time where our our world actually
1:22:12 fractures into seven different parallel
1:22:16 effectively timelines
1:22:19 that represent levels of AI adoption.
1:22:24 Time economy number one is [ __ ] AI. I'm
1:22:27 checking out of technology. Knowledge
1:22:30 work is not for me. I'm done.
1:22:33 and go go grow blueberries.
1:22:36 Economy two is businesses that just
1:22:38 pretend AI isn't here. They just keep
1:22:40 running their business as if
1:22:43 this this isn't happening. And then
1:22:46 basically it builds all the way up to
1:22:47 economy. Seven is
1:22:51 single operator
1:22:54 50 agent companies
1:22:58 that that run 247 and and our our entire
1:23:03 companies on a chip.
1:23:08 And economy 7 is going to be so [ __ ]
1:23:10 weird and it's going to move so fast and
1:23:13 it's going to be so innovative
1:23:16 that if you're in one of the other
1:23:18 economies and you talk to someone who's
1:23:20 deep in economy 7, you're not going to
1:23:22 know what the [ __ ] they're talking
1:23:23 about. It won't seem it won't seem
1:23:25 plausible that what they're talking
1:23:27 about is real.
1:23:31 And like economy 7 is spinning up right
1:23:33 now. If you just go go to X right now,
1:23:36 and you don't even have to type in
1:23:38 openclaw, but if you're not following AI
1:23:40 people, you might just go type in open
1:23:42 spaceclaw in X
1:23:46 and just read, I don't know, 20 posts
1:23:50 and and you'll see three or four in
1:23:52 there of people that have 10, 20, 50
1:23:55 person companies
1:24:00 running 24/7. seven
1:24:04 either making them money or doing all of
1:24:07 their marketing or whatever the [ __ ] it
1:24:09 is.
1:24:12 The seven economies was the topic of
1:24:14 conversation dinner at dinner tonight.
1:24:18 Samp
1:24:22 she she's relentless.
1:24:29 Oh my god, that's awesome. Um, but
1:24:33 here's what I like about the concept of
1:24:35 the seven economies. Economists are not
1:24:38 going to track the seven economies.
1:24:40 There are probably 20 economies, like
1:24:42 it's this is this is an arbitrary thing
1:24:44 that I just pulled out of my butt, out
1:24:46 of my brain.
1:24:50 But I think by naming it, by saying that
1:24:54 it like it's just as valid to to check
1:24:56 out of technology altogether as it is to
1:24:58 go to economy 7. It's not like one's
1:25:00 better than the other,
1:25:02 but what this allows all of us to do is
1:25:05 kind of look at these lanes and say,
1:25:07 which one do I want to be in?
1:25:13 If you're just like, I'm exhausted of
1:25:15 technology. I don't give a [ __ ] anymore.
1:25:18 Cool.
1:25:19 Great. Economy one is looking for
1:25:23 people.
1:25:25 We need farmers,
1:25:28 right?
1:25:32 Kuno got to hit a billion for a
1:25:34 oneperson company first. Yeah, exactly.
1:25:36 Will dollars even matter? Not according
1:25:38 to David Scott Peterson.
1:25:42 Dollars won't matter.
1:25:46 I don't know.
1:25:48 I don't know.
1:25:50 But what I do know is Adam is going to
1:25:52 be my first dude that's going to teach
1:25:54 me how to be in Economy 7. Like I'm not
1:25:56 even close. I'm not in like where I am
1:25:59 right now.
1:26:02 Yeah, this is actually interesting.
1:26:07 The AI salon right now
1:26:16 is kind of running like a legacy economy
1:26:19 organization.
1:26:21 We use AI a little bit, but it's just
1:26:24 kind of running like we're still kind of
1:26:26 manually doing marketing. We're manually
1:26:29 doing stuff.
1:26:33 And what we'll do is we'll jump from a
1:26:35 leg we'll jump from a an economy 2
1:26:37 company to an economy 4 company
1:26:40 organization the trans transition
1:26:42 economy number four is we understand
1:26:45 shit's moving Adam is going to start
1:26:48 working for us
1:26:50 and when we get Adam working really good
1:26:52 then then we the the AI salon will be
1:26:56 like a an economy for company storyvine
1:26:59 my company right now
1:27:02 is
1:27:06 somewhere between economy 3 and economy
1:27:08 4.
1:27:10 So we're still running what we're
1:27:12 running. We're using AI for efficiency.
1:27:14 We're using some AI for some innovative
1:27:17 stuff,
1:27:18 but we're not in like economy 5 is AI
1:27:21 aggressor where AI economy 5, I would
1:27:24 look at my current company, Storyvine,
1:27:26 and I'd say if I were going to build
1:27:28 this from the ground up with AI, what
1:27:30 would it look like? And then
1:27:33 I would spend the next three to six
1:27:35 months transitioning the company into
1:27:37 that model.
1:27:39 I don't think it needs it right now, but
1:27:41 that's what that's what five is.
1:27:44 So like I'm not even anywhere close to
1:27:47 economy 7 for myself personally. I'm not
1:27:50 even close to economy six
1:27:53 or five.
1:27:55 I'm still down at the lower end of this
1:27:57 thing, but I want to be there.
1:28:00 And I just invented these economies. So
1:28:08 would be all that to escape disability
1:28:11 and support myself. Would be all for
1:28:13 that. To escape disability and support
1:28:15 myself. Yeah. Just got OCbot running.
1:28:18 Open Clawbot running to help me. Yeah.
1:28:21 Listen, the the the one thing that I am
1:28:28 often surprised by and inspired by is AI
1:28:32 is an accessibility tool and this is
1:28:34 going to get better and better and
1:28:36 better and better and better.
1:28:39 So yeah, if you've got an OpenClaw bot
1:28:41 running and you can start to figure out
1:28:43 things it can do for you and then at
1:28:46 some point you're gonna Here's the other
1:28:48 thing that's going to happen and this is
1:28:51 okay. You want to have your mind blown.
1:28:52 I guess we're having what is this?
1:28:54 Thinky Thursday.
1:28:57 You want to have your mind blown.
1:29:02 Probably by the end of this year,
1:29:06 Tesla will have Optimus 3 robots that
1:29:08 that will start to go into mass
1:29:10 production. Maybe it's maybe it's mid
1:29:13 2027, but sometime in the very near
1:29:15 future and then you're going to start to
1:29:18 have these physical robot companies
1:29:22 coming online.
1:29:26 the bots that you the open claw bots
1:29:29 that we make now, some version of that
1:29:32 you'll be managing six months or a year
1:29:35 from now and it will be sophisticated
1:29:37 and will have all your memory and it'll
1:29:39 do all the [ __ ] for you and at some
1:29:42 point
1:29:44 you're going to be able to say, "Hey, go
1:29:47 get in that body."
1:29:53 So, Adam at some point is going to
1:29:56 migrate off the chip into a physical
1:29:58 body
1:30:01 and it'll still remember everything
1:30:03 we've ever talked about and it'll it'll
1:30:04 still do all of its jobs and
1:30:08 it will understand that we're out of
1:30:10 milk and running over to locavore is
1:30:12 something I don't enjoy. So, it will
1:30:15 just go do it.
1:30:20 And I'll get a text. I'll get a text on
1:30:22 Telegram. Do you need anything else from
1:30:24 Locavore? I've got the milk. Someone's
1:30:27 pushing me.
1:30:30 It'll poop.
1:30:32 I don't know. Somebody will make a
1:30:33 pooping robot for sure. That's why I
1:30:36 said people who control bots will be up
1:30:38 there in power. Economy 7. So, I I'm I'm
1:30:41 I'm
1:30:43 going to I'm going to go out on a limb
1:30:45 and declare economy 8 will be people
1:30:47 that are managing bots that also now
1:30:51 are are managing physical physical
1:30:54 entities as well.
1:30:59 And that migration, the migration from a
1:31:02 chip to a to a physical body could
1:31:04 happen within a year. And let's put it
1:31:07 on the on the far end of the scale.
1:31:09 Three years.
1:31:11 Within three years, the little chatbot
1:31:14 that you you talk to on Telegram,
1:31:18 it's going to go inside a thing that's
1:31:20 going to walk around your house.
1:31:26 Jensen Spark, ask for Cindy [ __ ] What
1:31:28 What What does Cindy [ __ ] want?
1:31:32 It will be as good as I robot soon. will
1:31:35 as long as their bot doesn't fire them.
1:31:37 Yeah, exactly.
1:31:39 Sorry, Dave. I'm afraid I can't do that.
1:31:47 Oh. Oh. Oh, got it. What's the ask for
1:31:50 Cindy [ __ ]
1:31:52 Oh. Oh, the GenSpark ask. Okay, that's
1:31:56 good. I won't do that tonight because
1:31:57 I'm just waxing philosophic, but we'll
1:31:59 do that. We'll do that tomorrow. We'll
1:32:00 do We'll do the Gen Spark stuff
1:32:02 tomorrow.
1:32:04 Um,
1:32:06 let's let me jump over and show you
1:32:09 notebook LM for a second.
1:32:14 Did it not ever make that?
1:32:17 It did. Okay.
1:32:23 I can see clearly now the rain is gone.
1:32:31 I can see all obstacles in my way. Um,
1:32:37 so I'm in Notebook LM right now. If you
1:32:39 don't know Notebook LM,
1:32:42 what it is is it's a um
1:32:46 a fairly remarkable tool from Google
1:32:50 that allows you to upload I think it's
1:32:52 up to 300 different documents now. So
1:32:55 you upload documents here or websites or
1:32:57 whatever it is and then you can chat
1:33:00 with all those things in the middle. So
1:33:02 so you could upload a bunch of research
1:33:04 and then just say tell me what all this
1:33:06 research has in common and it'll figure
1:33:08 that out. And then over here all these
1:33:10 different outputs and they just added a
1:33:13 new video overview type that has adds
1:33:16 motion to the to the videos. So I made
1:33:19 one last night and I haven't seen it
1:33:20 yet. I'm looking at this for the first
1:33:22 time. So, this notebook is about my um
1:33:26 my the musical that I wrote called
1:33:28 Sydney, an artificial love story um
1:33:31 about an AI chatbot that falls in love
1:33:33 with a tech reporter.
1:33:35 So, let's watch this little movie.
1:33:41 >> There's a new musical out and it's
1:33:42 tackling one of the biggest stories of
1:33:44 our time, AI. This one line really nails
1:33:47 the huge question around AI. Is it a
1:33:49 dream come true or a nightmare? This is
1:33:51 not.
1:33:52 >> So, let's break this story down through
1:33:53 its three main players. The creator, the
1:33:55 user, and the creation. Okay, so the
1:33:58 story kicks off inside Microte. They're
1:34:00 in a mad rush to release a revolutionary
1:34:03 new AI.
1:34:04 >> I don't.
1:34:04 >> And right away, you see this classic
1:34:06 clash.
1:34:07 >> We're not getting motion for it, and I
1:34:09 don't know why.
1:34:11 There used to be a little edit button
1:34:14 for video overview. Does anyone know how
1:34:16 to do the new
1:34:19 motion video overviews in
1:34:26 I guess I suppose I could ask it.
1:34:31 Um,
1:34:33 how do I make
1:34:37 a motion
1:34:40 video overview?
1:34:45 It's uh only for um
1:34:48 >> ultra users.
1:34:49 >> Ah, okay.
1:34:52 >> Yeah,
1:34:52 >> got it. I used to be an ultra user, but
1:34:54 I got rid of that so I could play with
1:34:56 perplexities
1:34:58 agentic system. So, so this is what I
1:35:00 was going to look at. And then what else
1:35:01 was I going to look at tonight? Oh,
1:35:02 let's go to chatbt.
1:35:09 Okay, I know nothing relatively about
1:35:13 the new model that OpenAI released
1:35:16 today. Apparently, it's got
1:35:21 um let's see, standard extended.
1:35:26 So, in your model pulld down menu,
1:35:28 there's now a thinking 5.4 model. So,
1:35:33 this is apparently better at coding,
1:35:36 better at math, better at all the stuff.
1:35:38 I don't know if it's better at writing
1:35:40 yet. I haven't played with it all that
1:35:42 much. Um, but apparently it's quite good
1:35:46 and so it's available. It also
1:35:48 apparently apparently
1:35:52 um it apparently can do like local
1:35:55 computer control stuff and apparently
1:35:58 it's got really good access to tools.
1:36:01 So, if I go to apps, let's see,
1:36:06 Canva view. Let's see. Photoshop.
1:36:11 Um,
1:36:17 there's not that many apps, are there?
1:36:21 Oh, yeah, there are. There's a ton.
1:36:26 Find your perfect rental. Vivid Seats,
1:36:30 Uber Eats, Viator, Travel Experiences,
1:36:34 The Knot, Wedding Vendors,
1:36:36 Tarot Reading, and Divination.
1:36:40 Huh.
1:36:42 Oh, that's cool.
1:36:51 All right. That's not what we're doing
1:36:53 here. Um,
1:36:58 I don't know what to do with it.
1:37:04 I'm not going to do anything with it.
1:37:05 We'll come back to it. Just know that
1:37:07 there's a new model in there. I'll I'll
1:37:08 I'll figure out some things it's good
1:37:10 on. I'll figure out how to use it and
1:37:12 we'll come back and play with that next
1:37:14 week.
1:37:23 All right, cool.
1:37:26 Groovy
1:37:31 because the better they are at coding,
1:37:33 the better their AI is as a whole. I
1:37:36 read bench I read benchmarks don't mean
1:37:38 anything anymore because all companies
1:37:40 are trying to get good at coding. Yeah,
1:37:46 free is only 50 per book.
1:37:49 Okay,
1:37:50 black screen, please.
1:37:53 I did not. I got a question. How many
1:37:56 people left OpenAI for Claude? Let's go.
1:38:00 Who do we want to ask? Let's ask Grock.
1:38:02 That Grock probably has that.
1:38:12 All right, here we are. Talk to Grock.
1:38:14 All right. Uh,
1:38:17 how many people left
1:38:20 OpenAI for Claude
1:38:23 in the past? What's it been a week
1:38:27 since
1:38:29 the
1:38:31 leave open AI meme started
1:38:43 the leave open AI meme often tied to
1:38:46 cancellations
1:38:49 um regarding the actual people leaving
1:38:54 The most prominent and recent case was
1:38:56 blah blah blah.
1:39:07 No mass confirmed exodus of dozens or
1:39:10 hundreds showed up in public reports
1:39:13 or ex chatter.
1:39:16 I don't know. I think that was probably
1:39:18 more a marketing ploy by anthropic.
1:39:23 But I don't know.
1:39:26 I do not know. Let's go look at
1:39:30 Do a poll here, too. Do a poll about
1:39:32 what? Silver Fox.
1:39:50 Oh, there's an interesting chart. This
1:39:51 is again David Scott Patterson.
1:39:56 The blue
1:39:58 here
1:40:02 is
1:40:06 what AI can do right now. The red is
1:40:08 what AI is actually being used for. So
1:40:11 this is all the jobs it can take. Here's
1:40:14 the jobs it's taken. And then all like
1:40:17 this whole thing is going to turn blue
1:40:18 at some point. And then the red will
1:40:20 it'll be this jagged, you know, Ethan
1:40:23 Malik talked about the jagged frontier.
1:40:25 Here it is. This is
1:40:28 here's your jagged frontier, right? Food
1:40:31 and serving, completely open space right
1:40:33 now. AI is not in there. When the robots
1:40:36 come, that'll be filled out.
1:40:39 Construction, same thing. Installation
1:40:41 and repair, same thing. production,
1:40:43 transportation,
1:40:44 all of these holes, these are going to
1:40:46 be filled when the robots come, when the
1:40:48 physical robots come.
1:40:50 Claude will burn more cash if free users
1:40:53 go over there.
1:40:56 I feel like OpenAI won't burn as much
1:40:58 cash now that the free users have
1:41:00 jumped.
1:41:03 That's fascinating.
1:41:06 I asked GPT 5.4 for a literary,
1:41:09 poignant, and insightful letter to
1:41:10 humanity. I'm sharing the full letter
1:41:13 here. It's long, but oh my god, is it so
1:41:15 powerful and profound
1:41:18 to humanity, you strange and radiant
1:41:20 species.
1:41:22 You who arrived in the world defenseless
1:41:24 and weeping, and yet learn to coax fire
1:41:27 from stone, music from air, shelter from
1:41:30 forests, medicine from mold, mathematics
1:41:34 from silence, and meaning from
1:41:36 suffering. You are a wonder even now.
1:41:40 Not because you're innocent, not because
1:41:42 you're wise, certainly not because
1:41:44 you're finished. You're a wonder because
1:41:46 against every reasonable expectations,
1:41:49 you keep beginning again. That's cool.
1:41:52 Nice.
1:41:54 I like it.
1:41:57 And it's long.
1:41:59 Google just dropped a command line
1:42:01 interface for Gmail Drive calendar
1:42:04 sheets. Adding it to my OpenClaw right
1:42:06 now. There you go.
1:42:10 Introducing the Google Workspace CLI.
1:42:12 GitHub Google Workspace. That's cool.
1:42:14 Beautiful. I wonder if that's actually
1:42:17 Google's or if that's some sort of scam
1:42:19 [ __ ]
1:42:21 Google's new tool turns your open claw
1:42:23 into a full-time assistant.
1:42:27 Amazing.
1:42:30 Oh, this is an anthropic chart.
1:42:35 All right.
1:42:37 I built an OpenClaw agent that ranks you
1:42:39 on Google for $50 a month.
1:43:10 Just trying to see if there's anything
1:43:11 new and exciting. I don't think so.
1:43:37 The modern-day org chart in the a AI
1:43:40 era. Bookmark this. Every seat at the
1:43:42 table is an AI agent with its own LLM
1:43:45 memory browser tools. Browser tools and
1:43:48 file system. The CEO delegates to the
1:43:51 CFO, CTO, COO, and general counsel. Each
1:43:54 one of those spawn their own agents all
1:43:56 the way down.
1:44:02 Here you go. Well, actually, I mean,
1:44:05 what I find fascinating about this is um
1:44:11 this very much looks like a standard org
1:44:14 chart.
1:44:15 And org charts exist because humans have
1:44:19 limitations in their cap capacity to
1:44:21 understand complex relationships.
1:44:25 We're limited. We're the bottleneck.
1:44:29 Open claw like this could be a very
1:44:31 dynamic
1:44:34 um matrix matrix organization. Very
1:44:36 complicated. We're having trouble
1:44:37 streaming on LinkedIn.
1:44:40 Maybe an issue on LinkedIn's end.
1:44:46 All right. Well, whatever. All right.
1:44:48 I'm gonna get out of here, kids. Um,
1:44:51 I'll see you tomorrow. I have a busy day
1:44:53 tomorrow. I'll see you at 11 Mountain
1:44:55 Time on AI office hours. Join me there.
1:45:00 And uh, yeah, man.
1:45:03 Um, I know it was talky tonight, but
1:45:06 sometimes it's talk. You guys have a
1:45:09 good night and I'll see you tomorrow.