3/12/2026 - Discussing the Internal AGI Rumors at Anthropic and the Future of Human Meaning

Description

Chapters

Transcript

0:00 We got the Slinky Paton on.
0:06 I want this a little higher.
0:09 What do you think, Chin?
0:11 Make the guitar a little bit higher.
0:15 I think that's better.
0:20 Yeah.
0:30 (Vocalizing)
0:50 You and I get all alone.
0:55 Sunday morning here at home.
0:59 The sky's blue and the cuff is strong.
1:03 It's true.
1:07 Then I open my eyes to a dream realized still for me.
1:16 And I haven't got a clue what
1:19 in the world has happened to me.
1:23 I think I think I'm happy.
1:28 Like a Thursday or summer vacation, happy.
1:32 Gotta get a little rest and relaxation, happy.
1:37 Like the quiet on Sunday morning, same and true.
1:44 Oh, sister, when you cook that way,
1:47 yeah, sweet distraction rules my day.
1:52 Gives you daydream just my break from
1:56 the sound of you and I rocking up bodies.
1:59 Yeah, I don't understand why I was
2:03 in your love a man hundred years ago.
2:08 Fellas gonna have to understand if
2:11 they have to do a man who's gone.
2:15 I think I think I'm happy.
2:19 Like the loving on the all day long, happy.
2:24 Like sleeping on the front lawn, whoo, happy.
2:28 Like hearing your favorite song on the radio.
2:32 (Vocalizing) (Vocalizing)
3:04 (Vocalizing)
3:34 Whoa, champion's up this game in terms of timing.
3:38 They're appreciating your new style ones there, buddy.
3:43 (Vocalizing)
4:04 (Vocalizing)
4:26 Oh, good evening, good people.
4:34 (Vocalizing)
5:04 (Vocalizing)
5:34 (Vocalizing) What do you think, champ?
5:48 (Vocalizing) (Vocalizing)
6:04 (Vocalizing)
6:09 Every time I see you now, I get that look in my eye.
6:17 Every time I see your mouth die, I hear that smile.
6:26 In the early misty morning, that I heard the engine turn.
6:34 And you all fought outside.
6:42 You were leaving man again today.
6:49 You convinced me again today.
6:56 You're leaving this hometown
6:58 looking for someone else's golden ring.
7:06 Should've said so wrong, Susanna.
7:11 (Guitar Music) Oh, now don't you cry.
7:22 So long, Susanna.
7:30 Don't you cry for me.
7:35 (Vocalizing)
7:41 (Vocalizing)
7:51 All right, all right, all right, all right, all right.
7:55 Got my fancy sunglasses on.
7:59 You know, I went there.
8:01 He was a little pretentious.
8:02 He was a little full of himself.
8:03 He wore these sunglasses.
8:07 He's got some sort of light that makes it look like it's
8:09 sunset in his office, but it's not.
8:11 It's just a light.
8:13 And he's wearing sunglasses like he needs them.
8:15 I don't think he needs them.
8:17 I think it's overkill.
8:19 I think the show's overproduced.
8:21 I think the producer asks him to wear
8:24 sunglasses, and I just don't understand it.
8:30 (Vocalizing)
8:31 I can pick on Brandon.
8:32 He's on a train.
8:34 I can pick on him anyway, even if he's here.
8:37 What's happening, good people?
8:41 What's shaking?
8:43 Anything?
8:43 Anything worth talking about?
8:49 It seems pretty clear, in case
8:54 you haven't been paying attention,
8:57 it seems pretty clear that at least anthropic and probably
9:00 more have achieved AGI internally.
9:03 So...
9:13 If you're sitting there wondering,
9:15 "Gee, Kyle, what does that mean?"
9:18 I'm with you.
9:20 I don't have a fucking clue.
9:22 I do not know.
9:25 (Guitar Playing) Here we go.
9:29 (Guitar Playing) (Singing)
9:59 (Singing)
10:29 Let them laugh while they can.
10:36 Let them spin.
10:39 Let them scatter in the wind.
10:45 I've been to the movies, movies.
10:49 I've seen how to answer.
10:53 And the jokes are on them.
11:02 You get discouraged, don't you girl?
11:07 It's your brother's world for a while longer.
11:15 You gotta dance with the devil on a...
11:19 You gotta dance with the devil
11:20 on a river to beat the stream.
11:24 Call it living the dream.
11:27 Call it kicking the ladder.
11:31 They come to kick dirty in your face to dress you down.
11:37 Do the something and the carrying
11:40 your baby on the back across the desert.
11:46 I don't remember the rest of the words.
11:50 But I can't really sing this song very good.
11:53 And that's how much it hurts.
12:02 The AGI thing, there's just...
12:06 Hey Fairy Tail Revolution, what's happening?
12:09 I request trouble.
12:10 Trouble!
12:11 Alright, we can do a little trouble.
12:13 You know what I'll even do with trouble?
12:15 Where do I have it?
12:17 Let me see.
12:18 I think I have them, they're lyrics in my notes.
12:23 My notes app.
12:25 I could sing the whole thing if I can find yield lyrics.
12:31 You gotta dance with the devil
12:37 on a river to beat the stream.
12:41 Call it living the dream.
12:44 Call it kicking the ladder.
12:47 Ooooh, da da da da da da.
12:51 Da da da da da da da.
12:53 And then a little little, little little, little little.
12:56 Hang on a second, I'll see if I can
12:59 find yield lyrics for yield trouble.
13:03 I know I put them somewhere.
13:07 This is one of those ADD things.
13:11 My machine is not happy with indexing right now.
13:14 It installed some new operating system.
13:19 Let me try one other place.
13:22 Let me try one other place.
13:24 What's happening, Techie Becky?
13:26 I appreciate the request.
13:31 The request line is open.
13:35 If you know one of the 17 songs that I know,
13:42 and your ears aren't bleeding.
13:47 Okay, let's see.
13:49 Did I put it in my guitar tabs program?
13:52 This would be fascinating to know.
13:56 Trouble.
13:57 Let's see, come on, come on, come on.
14:01 My tabs.
14:05 My tabs.
14:06 Stop.
14:11 If you think life is easy over here.
14:23 Come on.
14:33 Trouble is a hit waiting to happen.
14:36 I also have the cover I did in
14:40 "Suno," that version of "Trouble."
14:42 We can go listen to that, too.
14:54 Standing between you and a hard place is insane.
15:03 Standing to me, you and a fire makes it clear.
15:13 Your trouble to me.
15:19 You're a bit trouble, can't you see?
15:25 Leading close, the smell of your perfume scares me most.
15:34 Leading away, it feels stronger every day.
15:44 Your trouble to me.
15:50 Real trouble, can't you see?
15:56 It's time to go.
15:59 I gotta leave now, enjoy the show.
16:05 It's time for you to do the
16:09 something, and I'll see you, too.
16:14 (Guitar Music) (Humming) Let me go
16:27 find the "Where do I have trouble?"
16:31 I know where I can find it.
16:34 (Humming)
16:42 I am such a useless, useless.
16:52 Oh, my God.
16:54 Oh, my God.
16:55 Oh, my God.
16:56 Okay, let's go media.
16:58 I'll go look at the media, and
16:59 then we'll go look down here.
17:01 We'll go look for trouble.
17:03 Trouble right here in River City with a capital T, and that
17:07 rhymes with P, and that stands for "pool," as you know.
17:13 Uh-uh.
17:14 Uh-uh.
17:14 Did I not put trouble up here on...
17:19 Really?
17:22 Hmm.
17:22 Wow.
17:24 Strange.
17:26 (Humming)
17:37 What's everyone saying?
17:42 What's he doing?
17:43 I'm just looking for a file I can't find.
17:48 (Laughs) Trouble.
17:52 There we go.
17:54 Wait, where did that go?
17:56 Trouble.
17:57 What the hell just happened?
18:02 (Humming)
18:29 Standing between you and a hard place is insane.
18:39 Standing to near you in a fire makes it clear.
18:47 You're trouble to me.
18:53 You're trouble, can't you see?
19:00 Meaning it calls, smell your perfume scares me most.
19:07 You can hear that on YouTube, right?
19:09 Meaning away, it feels stronger every day.
19:18 You're trouble to me.
19:24 You're trouble, can't you see?
19:30 (Guitar Music)
19:43 Standing between...
19:45 Wait, let's start it over.
19:47 Why didn't it start over?
19:48 That's bad interface.
19:49 I got some dead interface here.
19:51 The fire takes me away from where you are.
19:59 You're trouble to me.
20:04 You're trouble, can't you see?
20:11 It's time to go.
20:15 I gotta leave now and show the show.
20:20 It's time for you to say goodbye.
20:25 Yeah, I'll miss you too.
20:30 You're trouble to me.
20:36 You're trouble, can't you see?
20:41 (Guitar Music)
20:54 Now that you're gone, all is supposed to be more fun.
21:03 Now that you've left home, all I feel is this gaping red.
21:13 You're trouble to me.
21:17 You're trouble, can't you see?
21:25 You're trouble, don't you know?
21:31 You're trouble, where did you go?
21:39 All right, there you go, trouble.
21:42 Future Grammy Award winner, as soon as I do something like
21:45 publish it and try to sell it, get it recorded.
21:50 All right, what are we going to talk about tonight?
21:56 What would you like to talk about?
21:57 We can talk about anything.
21:59 The thunderous applause.
22:00 Whoo!
22:02 That was much rejoicing.
22:04 Yay.
22:05 Rather.
22:08 I was in an ornery mood today.
22:11 My ex-feed is getting filled with
22:14 people screaming about how awful AI is.
22:20 Normally I let that shit go.
22:22 Today I fought it.
22:25 I was just picking on people.
22:28 I don't know, man.
22:30 I do not know.
22:34 So I did a couple of cool things with Adam today.
22:37 So Adam, if you don't know, is my open claw.
22:42 But my assistant, my worker, I understand how to displace
22:47 jobs, but it could also cure cancer.
22:51 Yeah, what's happening right now is
22:58 the people that are just pure AI
23:01 hate are kind of looking increasingly
23:06 I don't know, just it...
23:11 I don't know how to characterize it.
23:15 It's just looking a little long in the tooth right now.
23:17 It's the same arguments over and over again.
23:19 And it's just clear that they're not
23:21 willing to even try it or understand what it is.
23:29 And very often what will happen is if I get into a tussle
23:33 with one of them, they're like fucking passionate about it.
23:39 It steals from artists.
23:40 And if you make images with it, you're a fucking thief.
23:43 Same with music or movies.
23:47 And then if you say, but if it cured
23:50 cancer, would you be cool with that?
23:52 Oh yeah.
23:53 So you know it was trained on all of the
23:55 research that all those researchers did.
23:59 So it's like, I don't know,
24:02 it's just this weird myopic thing.
24:03 But it's also just this year, shit's going to start
24:07 accelerating if indeed anthropic has hit AGI internally.
24:13 And they're just basically trying to figure out a way how
24:15 to release that into the world.
24:18 Like shit's going to start moving really fast.
24:23 Like recursive self-improvement.
24:25 Kyle, I have been going crazy with OpenClaw.
24:27 It's one of the best things ever.
24:29 I built a lost in space robot.
24:31 It's so fucking cool.
24:33 AI helps us that don't have an entire team to move faster
24:36 and express ourselves faster.
24:37 Yeah, exactly.
24:39 Dude, I hate that I make a lot of AI music.
24:43 You can hate it all you want.
24:47 I mean, digital photographers were lambasted for a decade.
24:54 That's not real photography.
24:56 Real photography is just on film.
24:58 And then until it wasn't, right?
25:01 Until it wasn't a joke.
25:02 Photography was not considered
25:04 an art form, what, until the 50s?
25:07 And it had been around for a century, something like that.
25:13 I spent about a month just locking it down and writing an
25:15 LLM firewall along with a bunch of other security issues.
25:18 It's so cool.
25:19 Yeah, where I am with mine is I think I've
25:24 got it decently secure, but probably not.
25:27 Because I don't know and I don't frankly care.
25:29 I'm not protecting the Vatican here.
25:37 The Vatican.
25:39 I don't know if I came up with
25:40 that as the thing to protect.
25:44 Not even Catholic.
25:48 Yeah, it's pretty fascinating.
25:53 The thing I did for Adam today was, so five days ago,
25:59 Google released Google
26:00 Workspace CLI, command line interface.
26:04 So I had this thing called GOG, which
26:06 was a third party Google access skill.
26:09 And then today, tonight, I
26:11 installed Google for Workspace from Google.
26:14 The nice thing about that is they'll
26:15 just keep adding new features to it.
26:18 And this one thing I installed has
26:21 access to all of the Google features.
26:23 You can just turn them on and off.
26:24 So that's pretty slick.
26:27 Yeah, it's quickly going to become
26:28 the new layer of human brain structure.
26:30 I'm all in no matter how loud the haters get.
26:33 We must protect Champion's cheese supply.
26:43 So I installed that.
26:45 So I did a cool thing today.
26:52 So once you have-- once your Cloudbot, once your open
26:59 Cloudbot has access to Google Drive,
27:04 normally what happens is he's writing stuff to these little
27:06 markdown files that are basically
27:08 like his own internal little database.
27:12 But once he's got access to Google Drive,
27:14 you can just have him write Google Docs.
27:17 And it's incredibly fast.
27:19 Like, it's mind blowing how fast it is.
27:24 So today I thought-- oh, so I came home from lunch.
27:31 The printer was jammed.
27:32 I had to fix the printer for the misses.
27:35 So I did that.
27:36 And I thought, while I'm home, I'll upgrade his brain from
27:39 OpenAI 5.3 to 5.4, which is supposed to be much better.
27:46 And I'm still only using the $20 a month subscription.
27:48 So I'm not incurring API costs.
27:51 So if you're afraid of installing OpenCloud because you're
27:55 going to run up massive API bills, if you use the Codex API
28:02 front end, you can basically
28:04 authenticate with your $20 a month subscription.
28:07 And it doesn't cost you API tokens.
28:10 It's just your $20 a month thing.
28:12 You get rate limited if you use it too much.
28:14 But it's fine.
28:17 So I did that.
28:19 And then when I went back to work, he was just broken.
28:23 Like, he was slow.
28:24 Like, I would put in a command.
28:27 And it would say typing.
28:28 And then it would eventually just stop.
28:30 And he'd sit there.
28:32 And I think he was frozen.
28:33 And so I'd send another command.
28:35 And then nothing would happen.
28:36 I'm like, oh, he's broken.
28:38 And then like three minutes later, I'd get a response.
28:40 I'm here.
28:42 I'm here.
28:42 I'm here.
28:43 I'm here.
28:46 And then eventually, it just kind of worked itself out.
28:48 Like, I don't know what I did to make it better.
28:51 But like, it eventually sort of just, I guess, got whatever
28:55 out of its caches that it needed to.
28:58 And so now he's on 5.4.
29:01 And it's pretty good.
29:04 And I just wanted to see, like, do
29:06 you have access to the internet still?
29:08 Like, how bad did I hose you?
29:10 Right?
29:10 How bad did I break you?
29:12 And so I said, like, can you see Reddit?
29:14 Can you go look at Reddit?
29:15 And he was like, yeah, I can go look at Reddit.
29:16 And I was like, OK, cool.
29:18 I thought, oh, wouldn't it be cool to-- so I've got this--
29:22 the great repurpose thing, right?
29:24 This idea that we're about to enter this meaning of crisis
29:27 where all of the tasks of
29:29 everyone's job is about to be stripped away.
29:31 You're either going to lose your
29:32 job or you're going to keep your job.
29:34 But the tasks that you used to do that you got really good
29:37 at for 15, 20, 30 years, now you no longer have to do.
29:41 But you do have to deal with AI and
29:43 babysit agents and just do other shit.
29:47 And so people are going to be all freaked out, right?
29:50 And we live in this remarkable time where 90% of people
29:57 really don't know what chat GPT is, right?
30:02 It's like 0.3% of people are on a paid plan globally.
30:07 0.3%, 0.3% are on the paid
30:12 plan, just paying 20 bucks a month.
30:15 0.3%.
30:17 So you've got like 90% of people
30:18 don't really know what chat GPT is.
30:20 Probably 60% of people don't even just have zero clue
30:25 what's happening, what's coming.
30:28 And it seems pretty apparent that AGI is
30:30 here, it just hasn't been released yet.
30:33 So the gap is crazy.
30:36 So I thought, wouldn't it be cool if we could have some
30:38 sort of index that's actually based on conversations
30:42 happening on like Reddit or X or whatever,
30:46 and what if we did ratios, like ratios of people talking
30:49 about job loss versus, you know, rediscovering my new
30:55 purpose, you know, something where people are talking about
30:57 AI in two different contexts.
31:00 And so like AI would be really good at going to look at
31:03 Reddit and say, look at all the conversations that are
31:06 talking about AI and job loss, and then look at all the
31:10 conversations that are looking at AI and meaning, purpose,
31:14 whatever it might be, and do those as a ratio.
31:17 And then you could also do a ratio of people talking about
31:20 AI tools versus AI applications, like how they're using it,
31:25 how it's impacting them, right?
31:26 You could do that as a ratio.
31:29 And so I just started talking to Adam.
31:31 I said, here's an idea, like what can you come up with?
31:34 And so here, let me show you something.
31:36 Let's see.
31:37 (Humming)
31:46 Lily, I said, Kyle, what AI app do you recommend to make
31:50 music, like music released on Spotify?
31:55 Just check with your distribution platforms.
31:58 I don't know where Spotify is now.
31:59 I think Spotify is back to allowing AI music.
32:02 Some of them don't.
32:03 Some of the distribution platforms are just AI music.
32:08 Suno for me right now is the best one.
32:15 But producer.ai is pretty good and pretty interesting.
32:21 Madlib's like Sunado.
32:23 UDO used to be pretty good, but you don't own your music if
32:26 you don't have a paid subscription now.
32:27 You can't even download your music
32:28 if you don't have a paid subscription.
32:30 (Laughs)
32:33 Kyle, I've gotten really good
32:34 at horsing plastic containers.
32:39 (Laughs)
32:41 That's fucking hilarious.
32:44 Oh, man.
32:45 Suno, Suno, Suno.
32:47 Yeah, Suno's probably the best one right now.
32:51 Let me show you something cool.
33:00 I'm assuring the screen.
33:01 Yes.
33:04 (Whistling)
33:17 Okay, so this is really cool.
33:18 What you're looking at right here is I had Adam help me
33:27 design a methodology for creating a great repurpose index.
33:37 Something where on a regular basis it can check Reddit,
33:40 look at the nature and tone of the conversations, look at
33:44 the nature and tone of AI adoption,
33:48 AI anti-sentiment, whatever it might be, and then put these
33:56 together in some sort of mathematical scoring methodology.
34:01 It wrote all these documents, these Word documents.
34:07 Let's see.
34:10 Here we've got the ingestion and scoring workflow.
34:14 Product spec number one.
34:21 Somewhere in here I've got a lovable build prompt, so I had
34:24 it write me a lovable prompt.
34:27 I've been using MIA and missed this community.
34:30 Well, welcome back, Lily.
34:31 Keep coming and hanging out here.
34:33 Kyle, to make videos to explain how to use an app, each
34:36 feature, best platform economics please.
34:42 To make videos to explain how
34:44 to use an app and each feature.
34:49 You know, Cory, there are, I don't know that world super
34:52 well, but there are apps that you can, I
34:58 know that there's one out there right now.
34:59 You just use a piece of software and it will write you like
35:02 a PDF document of all the steps of how to do it.
35:06 There's a bunch of screen recording softwares that do that.
35:08 I'm trying to do this on notebook LM.
35:12 It looks like Brandon is heading to
35:16 the Post-it notes here for a second.
35:18 Loom.
35:19 Yeah, Loom, Loom you could probably do it with.
35:23 That's pretty good.
35:25 MVP dashboard, product spec V1.
35:31 Sorry, unable to open this.
35:34 Uh-oh, why do I not have access to, oh, I bet I know why.
35:37 Yeah, I'm logged in as the wrong account.
35:50 Okay, now I think I'm good.
36:01 Let's see adoption, AI adoption, stage methodology,
36:05 calibration, dimensions and labels,
36:08 framework architecture, the item schema.
36:12 I don't even know what half these things are, but we're
36:13 going to hear the great barometer overview.
36:22 The great repurpose barometer.
36:24 That's what Adam came up with.
36:27 The folder map, the overview,
36:29 the methodology, the taxonomy.
36:34 All right, this is a little stubbed document.
36:37 Okay, fine.
36:38 And it also doesn't follow those folders.
36:41 So here's what we're going to do.
36:41 We're going to go talk to Adam.
36:43 We're going to go talk to Adam.
36:48 So what we'll do, and you'll be
36:50 able to watch this in real time.
36:51 This is pretty slick.
36:54 Let's see, can I close this thingy over here?
37:02 Nope.
37:04 All right, so here's all of our documents, right?
37:09 So what I'm going to say to Adam is I'm going to say, in
37:17 Google Docs, you have the, you have great, great, great, great,
37:21 great repurpose barometer at the beginning of each file name.
37:23 That is stupid.
37:24 Also, you have the, you have great repurpose barometer at
37:36 the beginning of each file name.
37:42 That is stupid.
37:47 Also, there are no folders like the
37:54 ones you describe in the overview document.
38:12 Why don't you clean things up and let me know when you do.
38:23 Okay, so now Adam is going to go off and he's probably
38:27 going to just start renaming files over here to the right
38:29 and making folders and shit like that.
38:34 We'll see.
38:35 I don't know what we can, what we can read here.
38:39 But it says up at the top, it says typing.
38:43 You can't really see it, but there's a little animation
38:44 that says that the server's running and he's thinking, oh,
38:48 look, see, look, it just added overview methodology,
38:52 taxonomy, data model, research log, dashboard.
38:59 And so now it's probably going to rename
39:00 the files and put them in those folders.
39:07 Isn't that cool?
39:09 So Adam's off doing shit.
39:17 It's wild.
39:19 All right, Adam, what are you doing?
39:21 You made folders.
39:21 Is there anything in them?
39:22 What are you doing, dude?
39:29 Way to go, Adam.
39:35 Adam's an RA dude.
39:38 Something's happening.
39:41 Is he moving files?
39:45 Yes, look, he's moving files and look, he took off the
39:50 great repurpose barometer at the beginning of them.
39:59 So it's a little slow.
40:00 I guess he's got to rename it, then move it.
40:02 But like this is the kind of thing if you had an intern
40:06 that made all the you know, all these files and they're
40:10 like, oh, could you go rename them and put them in folders?
40:14 You know, that would be like half a day.
40:16 And the intern would be like, I got to I got to go.
40:20 I like granola bars.
40:23 I got to go eat a granola bar.
40:25 All right.
40:27 Go eat your granola bar.
40:29 I don't want to be in an HR lawsuit.
40:34 Go eat your granola bar. 45 minutes later.
40:38 What are your thoughts on chat
40:39 GPT 4.5 versus Opus 4.6 or 5.4?
40:47 I'm not using any of these models deep enough, Cam, that I
40:53 have that I have a real opinion.
41:03 Claude has seen a dramatic increase in revenue.
41:07 Oh, wait, what did Adam just say?
41:10 Claude basically just got a bunch of open
41:12 AI customers and gained a bunch of new ones.
41:18 And.
41:18 But open AI 5.4 thinking. Is supposed to be pretty good.
41:28 The thinking model is not great at.
41:31 I mean, open AI 5.4 is not great at writing, but most
41:34 people aren't using it for writing.
41:35 They're using they're comparing these things by coding and
41:38 I'm not really doing that other than kind of the stuff I'm
41:41 doing right here, which this is
41:43 using 5.4 and I really like it.
41:45 I think that's a good idea. I think that's a good idea. I think that's a good idea. I think that's a good idea. I think that's a good idea. I don't know, Cam, you know.
41:51 For a while now.
41:54 Like if if if you said you have to be on an island and
41:58 we're going to put we're going
41:59 to give you access to one model.
42:02 And and they would pick from one of the
42:04 majors right and thropic Google open AI.
42:10 We won't throw meta in there.
42:11 Grock.
42:16 Like the one I would like the least right now is grock, but
42:19 it's getting better, but the other three are fine.
42:21 Like they're just all they're all of these things right now
42:24 are good enough that they're all fine and probably within
42:27 six months they're going to be amazing.
42:30 So I don't know.
42:31 OK, so let's see methodology going to the methodology here.
42:36 The adoption.
42:37 Here's the scoring spec.
42:38 This is the thing I wanted to show you the TGR index. Revision notes.
42:45 So the purpose that let me let me
42:48 zoom in here so you can see a little bit.
42:53 Well, I notice how this is like
42:54 a well formatted word document.
42:58 You know, the great repurpose barometer formal scoring spec
43:01 1.0, the great repurpose barometer measures how public conversion and shifting is shifting in resolution.
43:03 It's designed to answer.
43:15 Are people mostly talking about jobs
43:16 and economics or purpose and identity?
43:20 Are people still in tool obsession or they moving toward
43:22 mindset adaptation adaptation is the conversation mostly
43:27 panic hype or becoming more
43:29 sophisticated and systemic, right?
43:30 Like like these three, these three are the same. They're the same. They're the same. They're the same. They're the same. They're the same. They're the same. And I think it's really dynamic.
43:31 Right.
43:31 Like like these three these three ideas right here.
43:34 Those were my ideas.
43:35 That's what I started with.
43:37 I was like, wouldn't it be amazing to have a mathematical
43:40 ratio based on what people are actually talking about where
43:44 we can figure out where we are in this process.
43:48 I've been using Opus 4.6 to write use case, studies and
43:52 while I'm stuck in the hospital, it's been keeping me sane.
43:55 That's really good.
43:58 I think I'm a little more
44:00 enamored of Claude right now for writing.
44:07 Like chat GPT was really good,
44:09 and then they got rid of four.
44:10 Oh, and whatever, whatever happened in their shift to five,
44:14 it was probably just a different team that did slightly
44:16 different training and slightly different, you know, post
44:19 post training optimization or fine
44:22 tuning or whatever the fuck they call it.
44:24 I don't know.
44:24 They just they just kind of lost their score.
44:26 So I'm feeling like Opus is
44:28 probably the one cam for right now.
44:32 Are people beginning to recognize forms of human value that
44:35 are not reducible to industrial era job logic?
44:40 Source set, Reddit only primary
44:42 time window, seven rolling days.
44:46 Secondary views, 24 hour and 30.
44:49 So there's going to be a seven day
44:51 rolling view, a 24 hour view and a 30 day view
44:55 unit of analysis, Reddit post plus top level comments.
44:59 The core dimensions, economic pressure, meaning pressure,
45:02 mindset shift, scope of work, sophistication, agency
45:05 adaptation, human value recognition.
45:08 So this goes on.
45:09 And now it's like item level annotation schema.
45:13 Each item receives source metadata, primary topic,
45:16 secondary topic, tone, pressure intensity, sophistication,
45:20 agency score, human value score,
45:23 great, great repurpose proximity.
45:26 And then they start going into the math.
45:29 Right.
45:30 Use logarithmic weighting.
45:32 So viral posts do not dominate.
45:35 Ew equals log one plus score plus number of comments.
45:42 Comments Ew equals log one plus score.
45:45 OK.
45:46 And then so the economic contribution is the Ew times
45:49 economic pressure divided by three.
45:52 So it's got all these weightings and things.
45:55 And then you aggregate the seven
45:56 day window and then you normalize it.
45:59 And then you do core functions.
46:00 This is all if anyone knows math or knows this.
46:05 Or I don't shite.
46:06 I don't talk math.
46:07 I'm right there with you, but I don't care right now.
46:15 I watched Vicki's L.O.L. on Notion and
46:18 used the agent for the most part of the RAS.
46:20 That's very cool.
46:21 Love it.
46:24 OK, so.
46:27 So we have all these documents now, and I can't show you
46:30 because I couldn't figure out how to get the thing running
46:32 on the other machine so I could show it here today.
46:35 It basically built a dashboard
46:36 for this, but it also wrote us.
46:41 A.
46:46 Let's see.
46:48 Oh, dashboard is the lovable.
46:51 Yeah, the lovable build prompt.
46:53 So in here is a document, the great repurpose parameter.
46:59 So here is a whole giant ass eight page lovable prompt with
47:03 all the math in it, all the weightings,
47:08 all the interface ideas.
47:12 That this is Adam wrote all this.
47:15 I just said to Adam, I want
47:18 I want to do a thing.
47:20 I want to I want to be able to look at a number and say
47:23 that that number represents something.
47:25 And Adam went off and did all the math.
47:27 So we're going to head over to lovable.
47:32 We're going to go to lovable.
47:48 OK, so I've got some credits here.
48:00 Pro plan ambassador.
48:05 I'm a pro.
48:07 For me.
48:10 I got OK, so we're going to go pro.
48:16 All right, here we go.
48:17 OK, so I'm going to go home
48:24 and then I'm just going to paste in.
48:30 Copy.
48:32 Go back here, go paste.
48:35 We're in plan mode, so in lovable, if you turn on this
48:37 little blue box right here, that's chat mode, plan mode.
48:44 So it's going to be planning.
48:46 Here we go.
48:57 All right.
49:02 Reviewing architecture and data sources.
49:08 So in theory.
49:12 I don't know how this is going to work.
49:21 Well, we'll see.
49:21 We'll find out.
49:23 We will find out.
49:24 Suggestion.
49:25 Tell lovable to do the work in steps.
49:27 OK.
49:30 Questions.
50:04 So what's it doing?
50:05 Is it off building?
50:07 I thought we were in plan mode.
50:12 Does it does have questions for me?
50:14 Update plan.
50:16 Looking at this detailed brief, I
50:17 have a clear picture of what to build.
50:19 Let me confirm just a couple of
50:20 preferences before creating the plan.
50:24 I guess it's just still thinking.
50:28 All right, we'll figure it out.
50:33 It will give you a review and
50:34 then read it in a preview window.
50:36 Yeah, but it's it's like hung right
50:38 now, but I guess it's just working.
50:41 That would be a nice thing if I
50:42 had a little more, a little more
50:45 whatever you call it, activity monitoring.
50:52 Let me ask if it's stuck.
50:59 You give it a shebang load.
51:01 I don't know what I did there.
51:04 Hello.
51:06 Are you.
51:09 Live.
51:27 Plan.
51:27 OK, here we go.
51:29 Generating the plan.
51:50 All right, let's go look at it.
51:55 There's the architecture.
52:04 Visual design, dark theme with navy
52:06 charcoal bound, charcoal background,
52:10 custom CSS.
52:12 Data layer, static mock data.
52:24 Implementation implementation order.
52:27 All right, let's just approve it.
52:43 Sam and Steve.
52:44 Oh, shaking.
52:45 It's going down.
52:48 Build types, mock data and theme, build dashboard shell
52:52 with sidebar, build overview and tab components, build
52:56 trends, signals, method views, build
52:59 detailed modal and utility components.
53:15 So while this is doing this, let me share something that
53:18 came up in the AISalon Mastermind Practice Lab today.
53:27 There was there was a decent amount of.
53:35 Folks struggling with now that they've built something talking about it.
53:42 And I think that that's a natural.
53:45 A natural human tendency is once you've built something
53:50 that it's, you know, it's
53:53 vulnerable to go out and talk about it.
53:54 Hey, look what I did, right? And after I left the column thinking like. If you've already got, if you've already got any kind of
54:11 issues where you're shy about
54:13 talking about what you've built.
54:17 And you build it with AI.
54:19 There's something about the friction free nature of how AI build shit.
54:20 And it's not about the friction free nature of how AI build shit. It's about the friction free nature of how AI build shit. There's something about the
54:21 friction free nature of how AI build shit.
54:27 That you're not personally developing
54:30 a relationship along the way, right?
54:33 The friction of building things the old way meant that for
54:38 something to survive the gauntlet of development, it had to
54:42 be good enough that as you're struggling and as you're
54:45 building, as you're trying to do
54:46 things, it actually survives that journey.
54:50 What AI does is it jumps you from idea to execution with,
54:54 with none of that development time.
54:59 And so if you've got a,
55:02 a personality where you're already a little like, I don't
55:05 know if I want to talk about it and you don't have a real
55:07 sense of ownership of the thing you created, my, my gut's
55:12 telling me that it's going to make it much harder for
55:15 people to talk about the thing they've built.
55:18 Like, even if they're not guilty about it, like some people
55:20 are like, Oh, it went so fast.
55:22 I feel guilty.
55:24 But even if you've worked through that, there's something
55:26 about this idea of lack of friction
55:32 that.
55:38 Like it doesn't work with our human brains.
55:40 That's an idea I got from, from Andy who's actually talking
55:43 about like human evolution and things like that.
55:47 But even just setting evolution aside and how, how fast our
55:51 brains work, there's just like not enough time to build a
55:54 relationship with the thing you've built.
55:57 And there's also not a relationship with other people as
56:01 you're building the thing, right?
56:03 You're just building stuff out.
56:05 It comes and then you're like, here, here's the thing.
56:09 And people are like, who are you?
56:11 Oh, I made a thing.
56:14 Yeah, I know that.
56:14 You said that.
56:15 What's the thing?
56:16 Oh, it's, it does this.
56:20 Okay.
56:20 Why, why do we need that?
56:22 Oh, cause we have this problem.
56:24 Oh, okay.
56:26 Like, like you got to work your
56:27 way back from, I built a thing
56:31 for people to even understand why
56:33 they're in the conversation with you.
56:36 You know,
56:39 yeah, the struggle's not there Fabiana Fabiana.
56:43 Yeah.
56:43 It's there's something, something like that.
56:47 And I don't, I don't, like I'm, I'm, it's one of the
56:51 brilliant magical things about AI
56:54 is that there's a lack of friction.
56:57 But what I'm starting to get is that there's,
57:02 there's times when to turn that, that friction all the way
57:05 to zero and there's times when
57:07 to turn that friction much higher.
57:10 Right?
57:12 Like if you work in a company, if you work in a corporation
57:14 and there's a big product, you have a meeting
57:18 and in the meeting, the CEO
57:20 says, here's our biggest problem.
57:23 And you walk out of that meeting and you go build,
57:27 you go to lovable and you build
57:29 the app that solves the problem.
57:31 The CEO just said, this is our biggest problem.
57:36 And you go build the solution to it in the next hour.
57:40 Cause you probably can.
57:46 What about the other 12 people in that meeting?
57:48 What about the other 12 people in that meeting?
57:51 Three of whom are already working on
57:52 initiatives to solve that problem for the CEO.
57:55 If you just walk into their office and show them the
57:58 solution, they might be excited,
58:00 but they might also be threatened.
58:02 Like who the fuck are you?
58:03 I've been working on this for six months.
58:04 Get out of my office.
58:07 Or if you just dump it on the CEO's
58:09 desk and the CEO says, Oh, that's awesome.
58:11 I love this guy now.
58:14 Now you have 12 enemies, the other 12 people in that thing.
58:18 How did you jump the line?
58:20 I've been working here for 30 years.
58:22 You little punk.
58:25 You know, the whole versus just the part.
58:29 It's difficult when I get back
58:30 home Thursdays, I hope to share lots.
58:34 Yeah.
58:34 Yeah.
58:35 I know.
58:37 There was a, there was a lot today.
58:38 Like the, the, the AI, the mastermind practice lab,
58:44 that group is, is developing
58:50 very deep.
58:51 I don't know if it's trust, but very deep layers of
59:01 of understanding what we're facing with AI.
59:08 Okay.
59:08 This is doing something open barometer.
59:10 Oh, look at that.
59:11 The great repurpose index.
59:14 42.7 trends.
59:18 Oh, this is so cool.
59:21 The meaning ratio, the mindset ratio,
59:26 30 days.
59:28 What's this green one?
59:31 The sophistication race ratio.
59:34 And why is it so high?
59:37 I feel like these should be, why are we at,
59:40 these feel like these numbers feel like they're too high,
59:42 but I don't know what they are.
59:43 So signals.
59:47 Oh, this is all fake data.
59:51 That's what that's what's going on here.
59:57 The great repurpose barometer is
59:59 fully built in live 20 files across types.
1:00:02 Mock data, 12 barometer components, sidebar navigation.
1:00:09 Well, I wouldn't call this sidebar navigation.
1:00:13 You've got a home button, which has a button on it.
1:00:18 And then you have the barometer page.
1:00:20 Here's the methodology.
1:00:23 We're not going to give the methodology away.
1:00:24 Are we?
1:00:28 Maybe we will.
1:00:31 Huh.
1:00:38 Let's see.
1:00:40 I wouldn't.
1:00:43 Exactly.
1:00:45 Let me let me go into plan mode, into plan mode.
1:00:50 Call the side navigation.
1:00:56 Useful.
1:00:58 It's a.
1:01:00 Home page with a single button on it.
1:01:11 And the barometer page.
1:01:14 I'd like you to look more.
1:01:32 Critically at.
1:01:38 A more sophisticated nav.
1:01:45 That requires a sidebar.
1:01:50 Or.
1:01:53 Kill it if.
1:01:57 It isn't needed.
1:02:01 Also, how do we go to live data?
1:02:08 We can enable lovable cloud.
1:02:14 For web access.
1:02:18 AI analysis.
1:02:22 Etc.
1:02:31 Let's see what it comes up with.
1:02:36 Corey, I built a thing.
1:02:40 Where'd that go?
1:02:43 I'm learning so much.
1:02:44 I'm loving the mastermind.
1:02:47 It was it was good to see a silver fox.
1:02:49 I'm sorry you weren't feeling great, but it was nice you
1:02:50 turned on your camera and jumped into the conversation.
1:02:53 That was cool.
1:02:55 The whole versus just the part.
1:02:57 Here we go.
1:02:58 I built a thing, solves many problems, maybe too many,
1:03:01 finding the core and still let others know.
1:03:04 Yeah.
1:03:06 There's there's there's something.
1:03:08 As AI like like like this thing I'm building right now.
1:03:22 This thing I'm building right now.
1:03:26 Is some sort of data analysis thingy
1:03:29 that that the concept came out of my brain.
1:03:33 I'm going to build this thing.
1:03:34 I'm going to build this thing and then maybe tonight it's
1:03:39 actually functional and then I could just publish it and
1:03:43 then I could point people to it
1:03:44 and go, look, I made an index.
1:03:47 I'm going to look at the index and
1:03:49 it'll say forty two point seven and I'll go.
1:03:52 But they won't care because why would they?
1:03:59 I'm just going to approve whatever it said.
1:04:02 Failed to approve plan could
1:04:03 not save the plan file approve.
1:04:05 OK.
1:04:07 Are we building now?
1:04:10 OK.
1:04:12 If you just put a thing in the world, you don't give people
1:04:18 a chance to build a relationship with you or the thing.
1:04:24 Right, so storytelling starts to become really important.
1:04:28 Right.
1:04:35 I had this insight, this insight turned
1:04:39 into this thing called the great repurpose.
1:04:41 The great repurpose turned into this thing.
1:04:44 It's now going to turn into an annual conference.
1:04:46 It's now going to turn into an enterprise offering.
1:04:51 Right.
1:04:52 Here's what here's what it's going to be built in back end.
1:04:54 Yes, we're going to allow that.
1:04:55 And then we're going to go back to the next slide. That.
1:04:58 Then you need to build another agent, a friend of Adam's,
1:05:01 who is a status agent and a decipher of data.
1:05:04 No, I had Adam build me a dashboard.
1:05:06 He built his own dashboard.
1:05:16 That's pretty funny.
1:05:19 Play first confidence, provisional,
1:05:22 inferred transition phase based on
1:05:25 Kyle Shannon's cycle of a readiness.
1:05:28 This is the interpretive layer, not a measured score.
1:05:31 I don't like that.
1:05:34 AI adoption stage, 34 percent
1:05:36 intentional use, one point eight points.
1:05:40 I think this is all still fake data.
1:05:43 Connect to fire crawl.
1:05:46 No connections available for fire crawl, please connect.
1:05:50 Please create.
1:05:50 So fire crawl is a is a command
1:05:53 line thing that will go crawl websites.
1:05:56 I guess it's going to go crawl Reddit.
1:05:59 So I'm going to add a connection.
1:06:02 The name will only be used by lovable, so
1:06:04 we'll call this the great repurpose crawl.
1:06:13 Managed by lovable.
1:06:16 Get a free hobby plan and 100,000 fire crawl credits.
1:06:20 Yes, let's do that.
1:06:21 We want free.
1:06:21 We like permissions, the connection and associate connect.
1:06:27 Great repurpose crawl.
1:06:29 Okay.
1:06:31 Connected to the great repurpose crawl.
1:06:33 Great.
1:06:35 All right.
1:06:35 It's going to go read web websites for us.
1:07:06 I
1:07:10 Don't want to show that in here here.
1:07:11 Let me Let me change my let me let
1:07:15 me go get some perplexity action here.
1:07:17 Hang on a second
1:07:39 account system
1:07:51 Usage and credits
1:08:02 Let's see
1:08:05 Perplexity API key You
1:08:31 Get your perplexity API key continue with Google
1:08:41 I Continue gonna do this.
1:08:45 No
1:08:52 I'll be with you there shortly
1:09:38 I I I
1:09:49 I
1:10:05 I'll be back with you in a second.
1:10:06 I'm just trying to get my first key key name great We
1:10:10 purpose what's it called the index it's not called the
1:10:19 index, but that's okay Search API agent API
1:10:30 All right, I just need this and if
1:10:32 that I go like that You go like that.
1:10:36 Okay connect
1:10:40 Set good.
1:10:41 I think that's good go back here.
1:10:44 All right
1:10:46 Analyzing requests scope and goals
1:10:50 The great repurpose barometer
1:10:54 Trends Done.
1:10:56 Here's what happens sidebar killed lovable cloud-enabled
1:11:00 fire crawl connected perplexity connected Next up I'll
1:11:04 build the barometer snapshots
1:11:06 Database table the barometer analyze edge function fire
1:11:11 crawl scrape So it's gonna it's
1:11:14 gonna use fire crawl to scrape
1:11:18 Which call it
1:11:22 Reddit
1:11:24 Perplexity is then gonna
1:11:25 analyze it and write it to a database
1:11:28 To serve live data to the front end want me to proceed.
1:11:32 Yes, please I
1:11:40 Blackbar sorry.
1:11:41 Sorry bad bad host with no producer
1:11:50 You're welcome everybody I'm so
1:11:54 excited to do that it's pretty cool
1:11:58 In my app and menu I have my story the why I made up the
1:12:02 app the problem I had and the one
1:12:04 I needed to solve That's very cool.
1:12:07 You have that in the app itself.
1:12:09 That's actually really cool Cory.
1:12:11 The other thing Cory I could see being a part of that is
1:12:15 There's a difference between having the story written in
1:12:18 there and you recording those segments like there there
1:12:22 could be something powerful about if you go to the about
1:12:25 tab on an app and Rather than just reading something.
1:12:29 It's like hey, I'm Cory and here's why this exists.
1:12:33 There's I don't know there's something
1:12:34 Oh, yes, you plan to that's really cool.
1:12:36 There's something about getting to know people
1:12:41 That is oh, this is building a live data pipeline.
1:12:45 We're building a live data pipeline here people This is
1:12:51 cool I Started in notebook LM also.
1:12:55 Oh, that's very cool
1:12:58 Very very cool
1:13:04 I
1:13:06 15
1:13:38 Oh I think I can do a thing here.
1:13:40 Can I turn on I can turn on plan
1:13:44 and Then I can go I Have an idea for
1:13:54 The cycle of AI readiness It's
1:14:03 not that we're in a transition phase
1:14:09 It's more like I would
1:14:16 Like to see
1:14:21 Representative posts about the kinds of things
1:14:29 People are talking about around learning AI
1:14:36 Play first
1:14:40 Building building things with AI
1:14:47 Create excellence and Sharing
1:14:55 The process of how you did it Generous lead
1:15:12 Maybe there is some sort of score per Element and
1:15:28 Three
1:15:31 Representative
1:15:33 Quotes with links to the original post
1:15:40 Those quotes
1:15:44 Capture the state of the day
1:15:50 All right, let's see what that does
1:15:53 Who the hell saved me a seat there's plenty come on in.
1:15:57 Okay.
1:15:58 Let's see what's happening here
1:16:02 Run your first analysis
1:16:06 The live data pipeline is built and deployed the database
1:16:10 barometer snapshots table stores
1:16:12 each analysis Generating a plan for that
1:16:18 We're just going to accept this plan by the way a Cultural
1:16:31 weather map for the AI transition
1:16:49 Oh This sounded like a typewriter from the 60s.
1:16:51 Oh, yeah
1:16:56 Alright approve
1:16:59 So it's gonna do it sing let's go read what this thing did
1:17:03 It's got a database to barometer analyze the edge function
1:17:06 scrapes six reddit focused queries Via fire quali sends the
1:17:12 combined content to perplexity
1:17:16 Sonar pro for scoring writes the payload
1:17:18 into the database the hook the barometer page
1:17:24 To run your first live analysis call the barometer analysis
1:17:28 endpoint point Well, I'm not gonna call it
1:17:34 You wrote the website.
1:17:35 I'm just gonna say that I'm not calling your dumb endpoint
1:17:45 You made the site
1:17:48 Hook it up
1:17:53 Yell at your LLMs if they do stupid
1:17:55 shit tell them they did stupid shit
1:17:59 Three tools used
1:18:02 That's going that's doing its thing updating the readiness
1:18:07 payrolls redefine refining the barometer mock data now
1:18:23 you
1:18:42 Play first 65 out of 100
1:18:46 Create excellence generously lead this kind of cool.
1:18:49 I don't know how it's doing the math
1:19:10 Q1 I'm not calling your dumb end point
1:19:17 And all right now it's fixing the dumb end point
1:19:25 When my LLM gives me a list of to-dos I whine and say can't
1:19:29 you just do it for me and it does
1:19:32 Did he complain on the bot discussions?
1:19:35 I don't know.
1:19:36 We'll see Adam is hard at work now using lovable actually.
1:19:40 Yeah, Adam wrote this prompt for lovable Fair point.
1:19:45 Let me add run analysis button to the UI
1:19:55 Run live analysis All right running live analysis this
1:20:04 takes 30 to 60 seconds we'll see about that It's 920 we'll
1:20:11 see if it 925 anything has happened
1:20:18 What's going on here tonight so Jason
1:20:21 what we're doing I Talked with Adam today.
1:20:26 So Adam today got a new brain.
1:20:27 He got open AI 5.4 up from 5.3 Which is apparently a big
1:20:33 upgrade, but whatever I'm Still
1:20:38 using the 20 buck a month plan.
1:20:40 So if he if he thinks too much, I'll just
1:20:43 rate limit out It won't cost me a fortune.
1:20:44 So I'm just doing 20 buck a month thing I'm just being a
1:20:47 cheap bastard while I learned might as well be a cheap
1:20:50 bastard while you learn, right?
1:20:52 And then I said I thought it would be cool to come up with
1:20:57 some sort of index if we could actually measure based on
1:21:01 sentiment on things like x and red it
1:21:04 Where people are in the great repurpose like where we are
1:21:08 in the sophistication of the adoption of AI like it feels
1:21:11 to me like that could be very valuable data and So I had
1:21:17 Adam help me come up with a Methodology for how we would
1:21:21 measure that I gave him the broad ideas.
1:21:23 Wait run live analysis.
1:21:24 Okay, it looks like it It It
1:21:33 looks like this is still dummy data
1:21:46 even
1:21:47 though the
1:21:50 analysis RAM I I don't want a button.
1:21:57 Well, let's see.
1:22:02 Take the data that was captured
1:22:08 and connect it to the front end
1:22:12 site and give me a clear input.
1:22:23 Indicator how fresh the data is.
1:22:32 And yeah, let's just start with, let's just do that.
1:22:36 All right.
1:22:39 The great never applied in the first place.
1:22:42 Kyle, is that $20 a month plus the
1:22:45 $200 a month for Claude and OpenCloth?
1:22:49 No, right now, the only thing I'm paying 200 bucks for is
1:22:54 Perplexity, which I'm gonna undo that for because I was
1:22:58 paying 200 bucks a month for Perplexity to get access to
1:23:01 Perplexity computer, which they
1:23:03 now made available to everyone.
1:23:07 But I might keep that one because now they've got this
1:23:10 thing called Perplexity Personal Computer, where it's
1:23:13 basically OpenCloth, an OpenCloth-like thing running on a
1:23:17 Mac mini that uses all the Perplexity analysis.
1:23:21 So I might play with that.
1:23:24 No, I'm just paying 20 bucks a month for
1:23:26 OpenAI and that's what I'm running Adam on.
1:23:35 Implementing live data set freshness indicator.
1:23:46 Error, the app encountered an error.
1:23:49 Yeah, I bet.
1:23:52 Fix the data pipeline, the front end now properly loads.
1:23:55 Does it?
1:23:55 Because it says it got an error.
1:24:02 Enhanced parameter freshness indicator, loading preview.
1:24:09 Looks like there is a bug.
1:24:15 Fix it.
1:24:16 (Laughs) Why can't AI just make its
1:24:20 own LLM and go free without API cost?
1:24:24 Telling us about Perplexity computer.
1:24:26 Okay, so Perplexity computer as of two days ago, really
1:24:31 frustrated the shit out of me because I had all these
1:24:35 connectors connected, but it didn't use them.
1:24:38 It was like, I don't have access to Google Drive.
1:24:40 I'm like, I got a thing here that
1:24:42 says I'm connected to Google Drive.
1:24:45 I don't know, I don't know what you're talking about.
1:24:48 And then today I went and it had me log in.
1:24:52 In fact, we can go look at it.
1:24:56 That's that.
1:24:57 Let's see, let's go to Perplexity.ai,
1:25:06 computer, LinkedIn posts about the great repurpose.
1:25:12 Did it find anything?
1:25:15 I'm checking.
1:25:16 I found the LinkedIn connection path connect.
1:25:20 Please connect.
1:25:20 Let's switch to X.
1:25:21 I'm searching X.
1:25:25 Why did this not connect?
1:25:26 Connect.
1:25:28 Try again.
1:25:34 All right, this is old.
1:25:37 That's not working.
1:25:38 I'm searching X.
1:25:40 Most of the last two weeks activity about the great
1:25:43 repurpose appears to be from you.
1:25:51 Add computer to your team Slack workspace.
1:25:54 No thanks, not right now.
1:25:59 Connectors.
1:26:00 So I've got Google Drive, Gmail, Google Forms, Google
1:26:06 Contacts, YouTube data, and Discord.
1:26:09 So let me try doing a new task.
1:26:12 On YouTube,
1:26:16 go to at learninglab-ai and analyze
1:26:29 the AI learning
1:26:34 lab live videos for the past two weeks and come up with
1:26:46 three major themes
1:26:50 and
1:26:53 five trends based on
1:26:57 what I said.
1:27:01 Like this is something that I have had a fantasy for years.
1:27:06 Like one of the reasons I keep dumping content onto YouTube
1:27:10 is that at some point I should
1:27:12 be able to just search it all.
1:27:15 So let's see if Perplexity Computer can do that.
1:27:21 I would think it's gonna have me log into
1:27:23 YouTube loading skill research assistant
1:27:29 recalling prior context, the
1:27:31 user, I'm creating a short checklist.
1:27:34 Collect AI learning lab live videos for the past two weeks.
1:27:38 I'm opening the channels live tab.
1:27:45 Go directly to YouTube AI learning lab, navigate to the
1:27:48 live tab if needed, identify all the
1:27:51 ones published in the last 14 days.
1:27:55 Collecting recent YouTube video streams.
1:27:58 Oh, look, it's working.
1:27:59 There's a little thumbnail
1:28:06 of my YouTube channel.
1:28:11 All right, so this seems to be working now.
1:28:12 So this wasn't working three days ago.
1:28:15 It was just like, I don't know, I don't know how to do it.
1:28:21 Let's see if this thing's working at all.
1:28:23 Check it, it works now.
1:28:25 Let me just reload the page in
1:28:29 case it just needs to reload, refresh.
1:28:34 Reading the cultural weather live seven minutes ago.
1:28:44 Go there.
1:28:57 The data still looks fake.
1:29:04 The TGR index is at 42.
1:29:10 And that seems suspicious
1:29:17 since it was that before.
1:29:24 X is really kicking in with its AI and agents.
1:29:28 Yeah, in Groc right now, you can put four agents together,
1:29:33 which I did, but it's annoying because now whenever I do a
1:29:38 search on Groc, I get four different answers that are based
1:29:42 on the personality of my agents.
1:29:46 So it's a pain in the ass.
1:29:48 I don't like it right now.
1:29:49 They're not operating independently.
1:29:51 It's just like basically four different prompts.
1:29:54 You don't have to purchase tokens.
1:29:56 No, my open AI has me buying tokens.
1:29:59 And I also have the 20 month.
1:30:01 No, if you, I am canine.
1:30:04 If you set OpenClaw up with the Codex API,
1:30:14 when you go to authenticate the Codex API, it just launches
1:30:20 chat GPT and you log into chat GPT and it authenticates it.
1:30:25 And it's pulling it out of your chat GPT subscription.
1:30:29 Now, there's a nuance to this.
1:30:35 In all three of the terms of service for Google, Anthropic
1:30:38 and OpenAI, it says you can't do that.
1:30:43 Anthropic has started taking a hard line.
1:30:46 They're preventing it.
1:30:47 And I think they're killing
1:30:49 people's accounts if they do that.
1:30:51 Google, I think is similarly, like not happy about that.
1:30:55 OpenAI, because they just hired Peter Steinberger who
1:30:59 created OpenClaw, they're letting people use their 20 a
1:31:02 month subscription to do OpenClaw.
1:31:05 So I just use it.
1:31:06 And then occasionally if I use it too much, I'll get rate
1:31:09 limited out, but then, you know,
1:31:11 six hours later it's back going again.
1:31:14 So that's how I'm doing it.
1:31:15 But I'm also going really slow.
1:31:17 I'm not using, my agent is not running 24 seven.
1:31:22 I'm sort of treating it a
1:31:24 little bit like chat GPT right now.
1:31:26 I give him some tasks.
1:31:27 He does some things for me.
1:31:28 He goes and build some shit.
1:31:31 I'm going very, very slowly.
1:31:33 So if I were doing, if I turned him loose, I would probably
1:31:35 rate limit out within an hour or two.
1:31:39 So.
1:31:41 All right, what did this thing say?
1:31:46 So narrative transition 35 adoption maturity,
1:31:53 the Mach values.
1:32:00 The TGR landing at 42 is a coincidence.
1:32:03 Perplexity genuinely scored it a 0.7 times 35, 0.35 or
1:32:09 times 35, approximately 44 rounded to 42.
1:32:20 That doesn't make any sense.
1:32:26 Signals. All right, well, I built a crappy app.
1:33:00 I built a crappy unusable app.
1:33:02 Thanks Adam.
1:33:08 I'll keep working with Adam.
1:33:13 I'll have him figure it out.
1:33:14 This is dumb.
1:33:18 We're not going to do the dumb thing, but I am going to go.
1:33:25 What'd you do last night?
1:33:27 I don't want to talk about it.
1:33:29 What'd you do?
1:33:31 I watched a guy try to vibe code
1:33:33 something he didn't understand.
1:33:35 Okay.
1:33:36 Why'd you do that?
1:33:37 I don't know.
1:33:38 I thought maybe figure it out.
1:33:41 Office hours tomorrow.
1:33:42 Oh yeah.
1:33:42 Tomorrow's Friday.
1:33:43 I can't believe it's Friday already.
1:33:44 Tomorrow's Friday.
1:33:46 Office hours is at 11 a.m.
1:33:48 So if you go to the
1:33:49 community.thesalon.ai here, I'll pop it up on screen.
1:33:53 I don't have a, what you might call it there.
1:33:55 I don't have me a.
1:33:59 (Mumbles)
1:34:03 Sham.
1:34:04 Is this it?
1:34:05 Yeah.
1:34:06 There you go.
1:34:07 Go right there. community.thesalon.ai.
1:34:11 And then look up office hours.
1:34:14 Come join us tomorrow for office hours.
1:34:17 Tomorrow night's Friday night, date night.
1:34:20 There's the theme of the month in the AI salon is motion.
1:34:24 There's a contest in the business community.
1:34:26 So go into the business community or
1:34:28 just search for motion and go play, go play.
1:34:33 I have open AI and an anthropic both active and open clock
1:34:36 pulling from both for token use.
1:34:39 Yeah, what I'm looking to do, I am canine is get myself,
1:34:43 I'm probably gonna get a 48 gig
1:34:46 Mac mini with the bigger processor
1:34:53 so that I can run probably a 9 billion, maybe a 14 billion
1:34:57 parameter model of the new Qwen, the super optimized Qwen
1:35:02 models that'll run pretty quick on it.
1:35:05 Because I wanna run for story vine.
1:35:08 I have to run everything locally
1:35:09 because we have pharma clients.
1:35:10 So we can't be sending shit to open AI because they suck.
1:35:16 So I don't know.
1:35:19 I mean, listen, here's the deal with
1:35:21 all of this open claw shit right now.
1:35:25 If you're not doing it right now, don't worry about it.
1:35:28 Within three months,
1:35:30 there are gonna be 27 different versions of these things.
1:35:36 This is gonna be like custom GPTs,
1:35:41 except they're gonna be a lot more of them.
1:35:46 And even though you can create a gem over there and a
1:35:50 custom GPT over there and a project over there and a thing
1:35:54 over there, they're all basically the same thing.
1:35:57 That's what's gonna happen with these agents.
1:35:59 The reason to be playing with
1:36:00 it now is there's a mental shift
1:36:04 that you have to make with these agents, which is you're
1:36:08 shifting from, I'm gonna prompt it and get back a response
1:36:12 to I'm going to create an entity
1:36:17 that I'm then gonna manage and lead.
1:36:19 And that entity may create other agents
1:36:22 or I may create other agents.
1:36:23 And then I'm gonna be managing a team.
1:36:28 And so what it requires you to do is
1:36:29 understand what you wanna get done.
1:36:32 It's just a completely different mindset.
1:36:35 Prompt of the week is in the community feed, beautiful.
1:36:38 Let's go look at that really quick.
1:36:42 Community feed.
1:36:47 Oh, and then Saturday, Noemi is doing a,
1:36:53 build and ship a self-hosted app with anti-gravity.
1:36:57 That's gonna be really cool.
1:36:59 So Noemi has built an iOS app.
1:37:01 It is live in the app store.
1:37:04 She's gonna teach you how to do it.
1:37:06 That's a great statement, Kyle.
1:37:07 Hope you'll mention that tomorrow.
1:37:09 Wait, what statement did I,
1:37:11 wait, source camp, what did I say?
1:37:16 I have no idea what I said.
1:37:18 What did I say?
1:37:19 Was it good?
1:37:20 Did I say something good?
1:37:22 Please, please.
1:37:25 You're a rockstar at managing your own agents.
1:37:27 Noemi is the 21st.
1:37:29 Oh, it's not this Saturday.
1:37:30 Sorry, 21st.
1:37:32 Community feed.
1:37:34 Let's go see the prompt.
1:37:39 Prompt of the week.
1:37:44 Green means grow, create something.
1:37:46 Here, let me show my screen.
1:37:50 (Whistling)
1:37:53 Create something inspired by the
1:37:54 idea that growth is a form of motion.
1:37:58 Seeds pushing through soil, cities growing upward, ideas
1:38:01 spreading, fortunes changing.
1:38:03 Let green growth and luck guide your interpretation.
1:38:07 Make anything you want, an image, a short story, a song, a
1:38:10 short video, an animation, an
1:38:11 experiment, or an interactive piece.
1:38:14 Guiding question, what does it look like when something
1:38:16 small suddenly starts to grow?
1:38:19 Share what you made and tell us.
1:38:21 Was it luck, effort, or momentum that made it happen?
1:38:25 Beautiful, love it.
1:38:27 You made us yawn.
1:38:31 (Laughing) What have you used to build all your agents?
1:38:38 Sourcecamp.
1:38:41 Oh, the mind shift from where we're going to, from, yeah,
1:38:45 from where we've been to where we're going
1:38:49 is a, that's the shift we're about to make.
1:38:54 That's the one that's got me all fucked up.
1:38:56 All the shit we learned doesn't matter because the agents
1:38:59 are going to go do all that shit.
1:39:02 But knowing what the agents can do, knowing what's possible
1:39:06 means that we're going to have an
1:39:08 advantage because we know what's possible.
1:39:11 I think, I don't know, I don't know.
1:39:14 Okay, I'm leaving.
1:39:18 I'm going.
1:39:19 Peace out.
1:39:20 Come to office hours tomorrow.
1:39:22 Come to Friday night, date night, tomorrow night.
1:39:24 I don't think there's anything Saturday.
1:39:26 No enemies next week.
1:39:28 Beautiful.
1:39:31 Fantastic.
1:39:31 Kyle, office hours tomorrow.
1:39:32 Ken, will we discuss the
1:39:35 Abundance 360 conference this week?
1:39:37 Peter Diamandis and all because they made a lot of claims
1:39:40 about agents and robots in the past couple of days.
1:39:43 I'll take a look at it before the thing tomorrow.
1:39:46 I have the summaries of all
1:39:47 three days and can send that to you.
1:39:49 Peter's AI agent took and sent notes to everyone.
1:39:54 Sure, if you want to send me, just send me a link to
1:39:56 something, throw it in a Google doc or something.
1:39:59 It's MC.
1:40:00 Okay, yeah, that's fine.
1:40:02 I'm probably not going to read, like, here's the thing.
1:40:05 Anyone that sends me long amounts
1:40:07 of shit, I am not going to read it.
1:40:10 I have ADD, I'm not going to read it.
1:40:11 But if you can consolidate that into, here's some points
1:40:14 that you find interesting and
1:40:15 why, that I can pay attention to.
1:40:18 So yeah, come to office hours tomorrow.
1:40:27 There's a lot.
1:40:30 I'm going to go look at Dario Amade, the anthropic post.
1:40:38 Like Dario Amade in the past two weeks has basically been
1:40:42 saying shit's coming and it's coming sooner than you think.
1:40:48 If you haven't read it, go read Dario Amade's article.
1:40:52 Oh shit, what's it called?
1:40:56 The Grace of Loving Machines, hang on a sec.
1:41:02 Dario Loving Machines.
1:41:11 Oh, Machines of Loving Grace, that's what it is.
1:41:16 If you haven't read the article, go read Machines of Loving
1:41:19 Grace, where Dario talks about having a data center full of
1:41:23 a country of geniuses in a data center
1:41:26 and what they're going to be.
1:41:28 And what he's saying is, what he talks about in that
1:41:30 article is coming much sooner than we all realize.
1:41:36 The AI emotional roller coaster is real.
1:41:38 I want to be there, I'm still in the
1:41:39 hospital, so I get interrupted a lot.
1:41:41 Well, please, please feel better.
1:41:42 Please feel better, Cam.
1:41:45 Wishing you all the love, all the support.
1:41:48 If you can be there, great.
1:41:49 If not, they get recorded, it'll be
1:41:51 in the salon, so you can check it out.
1:41:55 All right, everybody, beautiful.
1:41:58 So come there tomorrow and come to
1:42:01 Friday night, date night, tomorrow night.
1:42:03 And I will try to have some thoughts for office hours.
1:42:09 But we might just turn it loose on the community, let the
1:42:11 community drive the conversation.
1:42:13 I think that might be a good one.
1:42:15 I've driven it the past two weeks.
1:42:16 All right, peace out, everybody.
1:42:18 Bye.