05/04/2026 -Why Developers Are Migrating Between OpenAI and Anthropic as Models Evolve

Transcript

1. 0:18[music] Heat. [music] [music] Hey, Heat.
2. 0:39[music] >> [music]
3. 1:25[music] [music] [music]
4. 1:42[music] [music] [sighs] >> AI Learning Lab. May the fourth be with
5. 1:58you. Hey everybody, what's happening? I was muted. Now I'm not muted. I mean, I'm unmuted.
6. 2:12Happy AI learning lab in the uh in the AI salon. Welcome everybody. Welcome, welcome, welcome. Uh May the fourth be
7. 2:21with you. Any Star Wars nerds out there? I would think within the AI salon it's probably we're pretty pretty heavy on
8. 2:28the on the nerd to non- nerd quotient. [laughter] So I I would think um so so welcome inside the AI salon. So you're
9. 2:38in here, you're watching this um you know please comment, figure out the best way to watch this and stream it and you
10. 2:45can pop the window out. You can do um you can do chats within the uh within uh the the AI salon itself. Um you can do
11. 2:54chats right here in the Mighty Network. There's Kelly Camp is in here. Oh, good. You were just muted. I thought it was
12. 2:59me. No, it was me. It was full on me. Uh let's see. All right, that's good. Beautiful. Fantastic. Um so, welcome,
13. 3:11welcome, welcome. If you're new here, my name is Kyle Shannon. This is the AI learning lab. It's Monday night. So, the the way the
14. 3:19format's going to work is we're going to we're going to be going live at 7:30 Mountain time Monday, Wednesday, and
15. 3:24Friday. Monday's going to be like big idea night or big news item night. Tonight, we're going to talk about
16. 3:31OpenAI versus anthropic and what's going on. What is the dynamic with with people jumping ship from one of those to the
17. 3:38other and it seems to be ebbing and flowing a bit. I want to talk about what's going on with that. We'll talk
18. 3:43about some other news items. So, so Monday's going to be the the nickname uh over the past year or so of of Mondays
19. 3:51has been meltdown Monday. So, I tend to kind of lose my crap. Someone will post a comment somewhere and uh I'll kind of
20. 4:00lose my crap about something or other. Um so, so I don't think tonight's going to be a meltdown night, but you never
21. 4:06know. If you have questions about what's going on, pop it in the comments below. Um, if you want to start conversations
22. 4:13in the AI salon, there's an area called the community feed or the community chat. Uh, that's on that lefthand
23. 4:19navigation. Um, this live session, the AI learning lab live has its own chat. So, you can chat within that. You can
24. 4:27chat within the StreamYard where we can see it and pin it pin it to the screen. Um, so there's all sort sorts of
25. 4:32different ways to communicate. The purpose of bringing the AI learning lab from outside to inside the salon is is a couple of things. Um,
26. 4:45the AI salon is is a community. We've got about 4,000 members and yet in a typical week we only have about 200 members that are active,
27. 4:56right? So, so it's a it's it's essentially a 5% activity rate. I would love to get that number to 10%. I would love to get that
28. 5:06number to 20%. But one of the things that we really looked at and and I want to thank Andy Scarantino who's who's the
29. 5:14operations person that's making sure that all of the stuff within the community is running as smoothly as
30. 5:19possible. One of the things that she pointed out is a lot of our activity was outside the community and so I was
31. 5:26putting all this energy outside the community and it wasn't doing anything to feed the community here. When we
32. 5:32first started the salon, it was kind of the opposite of that, right? Everything was being driven to Discord or when we
33. 5:38first started Mighty Networks, there was just a single Tik Tok channel. I was driving everyone to a single channel uh
34. 5:44within the AI salon and there was lots of activity there as a result. Because of a lot of reasons, the the
35. 5:53focus of the AI learning lab went from one channel, Tik Tok, to four channels. there was Tik Tok and YouTube [laughter]
36. 6:01because because Tik Tok was going to go away and it was going to be illegal and then it wasn't and then so we've had
37. 6:07this sort of bizarre year year and change of of living on both and then plus because of Streamyard we can also
38. 6:15stream to LinkedIn and we can stream stream to X. So, we've got all this audience, all this activity happening
39. 6:21outside the salon and for good or bad, I wasn't actively pushing people in inside the salon. And so, so that's part of it.
40. 6:32So, big community. Um, we want to dramatically increase the active users in the community. So, that's part of it.
41. 6:41We also have member le spaces inside the salon and you know, they've noticed a drop in participation as well. So that's
42. 6:49things like um the the AI life hacks group that Brandon and Claire run, AI for business, AI for all minds, the neurodeiversity
43. 6:59um sub community that Gareth runs, um advocacy and policy groups, all that sort of stuff is are really interesting
44. 7:06subcommunities within the AI salon that just weren't getting uh weren't getting attention. Um and so we're looking to do
45. 7:13that. So um so I'm really excited about that. Um, and yeah, and and you know, I I want to like spin the thing back up.
46. 7:21Spin it spin the the conversation back up. Here's the thing that's going on. There is so much happening in AI right now. AI is
47. 7:33getting more and more and more powerful. It's getting more and more and more complex. for someone like me who, you know, I've
48. 7:41sort of prided myself on being able to um to keep up with it and talk about what's going on with it. Uh I haven't been able
49. 7:52to do that and everyone in the space that I know is in the same place that it's almost impossible to be able to
50. 7:58keep up with the tools and the tools are getting more and more complicated. And so I think that that that overwhelm is
51. 8:06overwhelming a lot of people. And so one of the things that we're doing inside the AI salon is things like the AI salon
52. 8:11mastermind practice where people are creating a daily practice around AI. Um that I'm I'm you know I'm a part of and
53. 8:19and help you know uh facilitate that community. Um but if there's an instinct to check out because things are getting too
54. 8:28overwhelming, I think this is the wrong time to check out. I think this is the time we need to check back in and
55. 8:34connect as humans and connect as people struggling to figure out what this AI stuff means and how we keep ourselves at
56. 8:41the center of the conversation. So, let's see. I think the earlier time will also bring more people in. Yeah,
57. 8:47hopefully that um AI life hacks rocks. I agree with that. It'll it'll be cool to see you all. I never get to see the Tik
58. 8:53Tok comments. Oh, that's cool. Yeah, exactly. Um yeah, if you have any if you have any comments, please please put
59. 8:59them in the comments. If you're if you're commenting in the StreamYard chat, I'll be able to see them in here.
60. 9:06Um, if you're commenting in the AI salon chat, I'll be able to see that as well, uh, at different times. Um, let me see
61. 9:13if there's other stuff that I wanted to talk about here. You know, one of the other things about
62. 9:29about community especially a community like the AI salon where we've got people from all walks of life,
63. 9:43right? We've got people who are retired. We've got people who are starting businesses. We've got people who are
64. 9:48working in big businesses, small businesses. Um, everyone's exploring AI, but everyone's got these very different lenses that
65. 9:57they're looking at the world through. And so therefore, they've got very different lenses about how they're
66. 10:03applying AI. And one of the things about being in a community like this is that because it's so overwhelming, because
67. 10:11it's really hard to understand what which way to look, like what do I look at? What do I pay attention to? Is to be
68. 10:24engaged with people that are trying to figure it out too, right? everyone everyone is trying to figure out what in
69. 10:31the hell do we do with with this AI stuff. And so what ends up happening is, you know, Kelly Camp will talk about,
70. 10:39you know, something that she did. She she presented in a group of CEOs and realized that, you know, the CEOs don't
71. 10:47have a clue, right? Or Jim Ross will say, you know, here's what I'm doing uh with with um with my storage community.
72. 10:56And the storage community is really embracing this AI stuff because they're able to 10x, you know, they're single
73. 11:02operator companies able to 10x what they're what they're possibly doing. Um, let's see. Um, oh, so in terms of chat,
74. 11:11so I just got this from producer Brandon. Um, producer Brandon, if you're chatting in other areas in the AI salon,
75. 11:18producer Brandon is monitoring those chats as well. So, anything interesting, he'll bring over here so I'll be able to
76. 11:23see it. Also, you can I think it's in the upper right hand corner of the chat. There's a pop out button. You can
77. 11:31actually pop out the the the video screen and then you can open another tab where you can be doing a chat inside the
78. 11:37AI salon. So, you can do both at the same time. So, um experiment with what works best for you and where you want to
79. 11:44chat and where we want to keep the conversation going. Um I need this community to keep me grounded and not so overwhelmed. Um
80. 11:53yeah. Yeah. So, so that's that's some of what's going on. That's some of what's going on. Um, we're going to do this for
81. 12:00a while. There's a couple of things we're going to do. One of the things we're going to look at is stats. I [laughter] personally
82. 12:08have never looked at stats. It's not a good thing. It's not a good thing at all. I just, you know, would
83. 12:15historically come online and just talk and whatever happened happened. Um, this is a really important time and this
84. 12:23is an important time to have our community growing and vibrant and engaged. Um, I don't want the energy to
85. 12:30dissipate. I want it to increase. And so what we're going to do is we're going to look at what we're doing here and look
86. 12:36at is this driving, you know, new members? Is this driving um, engagement within the community? My hope is yes. My
87. 12:45instinct is yes. So anyway, that's what's happening. Okay, Charles Adams, I found a chat window. Good. Char Charles
88. 12:52is here. Welcome, welcome, welcome. For those of you that are regulars that are starting to figure out how this works,
89. 12:57if you can find others, we've got almost 30 people in here right now. Um, if people are lost, go find [laughter] them
90. 13:03and and, you know, share your best practices, which which I'm excited about. Um, it looks like a rectangle. Oh, okay. In the
91. 13:11upper right hand corner, the pop out thing. Look for the rectangle. It'll it'll pop out the uh the screen.
92. 13:17Um, okay. So, let me show you. I want to show something here.
93. 13:32Share screen. So, okay. So, this site right here
94. 13:47the AI learning lab explorer. I think you can see this. Yes. Yes. Um this is a site that I made with um
95. 14:03with Codeex, OpenAI's Codeex. And if you don't know what Codeex is, there's a thing from Anthropic called Claude Code
96. 14:12um and Claude Co-work. They kind of live in the they're they're a downloadable app that that live on your computer and
97. 14:21you can develop software in them. You can use them to manage files on your computer. Codeex is the same thing from OpenAI. Um
98. 14:36I have historically steered away from what are equivalently essentially um software development environments, right? So claude code was
99. 14:49originally called claude code for a reason, right? It was for developers to code applications. It wasn't so much so
100. 14:56much vibe coding as a software development environment. They then realized hey people can manage their files on their
101. 15:06computer with these these this these things and we can use things like skills they call them. Um and so they created
102. 15:13this thing called claude coowwork which basically said hey you don't have to be a developer to use this. Similarly,
103. 15:20codeex started very much as a coding tool, but now can basically do things with all of your local files on your
104. 15:26local machine um and store all of its files on your local machine. And so you can do these high-end coding projects,
105. 15:33but you can also do all sorts of other things. These tools have memory. They they can remember what you've talked
106. 15:40about. They they you can add different skills to them. Um they can use tools like they can go out and search the web.
107. 15:48They can uh manipulate files and run things on your computer. Codeex in particular is really good at running
108. 15:55things on your computer and it's running everything locally. Clawed Code tends to live up in the cloud. Um the application
109. 16:02that you're looking at here allowed me well so so what I had was I had a PDF document that Lord Digital Gods had
110. 16:10produced which was basically the descriptions and chapters of all of my videos for the last three years of AI
111. 16:17learning lab in a single document. And so one of the things that I've always wanted to do, and I wanted
112. 16:24YouTube to to do this for me, I can't believe they haven't done it, is I would love to have a tool that I can log into
113. 16:32my channel that I've been putting lives every night for the past three years into this channel. I'd love to be able
114. 16:39go to go discover all of the stuff that's in there really, really easily. And there's never been a way to do that.
115. 16:45And so I don't know a week or so ago I opened OpenAI Codeex. I'd been reading about a lot of people saying GPT 5.5 is
116. 16:56really good and Codeex is really good. And so I took this PDF document that I had and I said I want to take this PDF
117. 17:04document and I want to create a a discovery tool where I can go discover things within um the AI learning labs
118. 17:12YouTube channel. And so it built me like these curated lists like I can go look at I don't know sunno songs. And so
119. 17:20here's a bunch of um clips of me doing things with poems into songs. This is from October 29th, 2025. That's not working.
120. 17:33Oh, these not working. Oh, they're working. I'll get some I'll get some Beatles songs out. Some James Taylor.
121. 17:41>> Let's see. All right, enough Martin Ston song lyrics. [music] >> I can jump ahead. >> I can click random clip. I can search. I
122. 17:55can search for then we'll go play with video. I promise we'll get to the video >> or um jank. >> Do it. Do it. Writing >> the jank. Wait.
123. 18:03>> And then we'll generate audio after that. We'll add music. >> Hang on a sec. Jank. And so over here are all the search
124. 18:11results for Jenk um intro and janky times. Embrace the jank images. These are kind of cool. That one's kind
125. 18:19of creepy, right? Creepy. Cool. So within about I don't know an hour, hour and a half, I went from not really
126. 18:30knowing what was on my YouTube channel to being able to very very quickly just jump around and and you know look at different things on the
127. 18:40channel. So this is all because of OpenAI's codeex with GPT 5.5. Um Um, let me go back to this window
128. 18:55and stop sharing. So, I want to I want to talk about some some interesting dynamics going on
129. 19:10in the developer community between the anthropic tool cla code, claude co-work, and the open AI tool codeex. So,
130. 19:25starting in probably it probably started in the summer of 2025 and and certainly started to precipitate in the fall of 2025
131. 19:36is that it felt like OpenAI had kind of lost the plot. Like their stuff just didn't feel as cutting edge. the the the things that
132. 19:47they were releasing were kind of incremental or they were tangential. Um, at some point, I don't know, a year
133. 19:54ago, they they released Sora to a bunch of great fanfare and they they actually just discontinued that. So, they they had a lot of projects,
134. 20:06some of which were getting lots of attention, some of which weren't. And kind of in the fall of 2025, it just
135. 20:15felt like, yeah, they're okay. But they still had 900 million weekly users. Like it's a lot, like almost a billion weekly
136. 20:25users. You know, they're still considered state-of-the-art things like that. But but there was just it just
137. 20:31wasn't quite as exciting and vibrant. Um, you know, Sam Alman was starting to get I don't know, a bit of a wrap of
138. 20:39like, you know, what what's going on with him, all that sort of stuff. All right. Oh, I like the orange bar
139. 20:47telling us what tool is being used. Cool. The Can I tell you another thing that is really cool about this new
140. 20:52setup? So, in the old setup, um, because I had to put this screen, the screen you're seeing on Tik Tok as well, I had this whole screen
141. 21:06was off to the side and so I could never really like look and see the comments like straight ahead. Everything was
142. 21:13always off buried behind something. So, um, so it's kind of cool to be able to just see all of you and see all of this
143. 21:19stuff right in front of me. um in December of well I think in the in the summer of of 25 may maybe in the fall of 25 I don't
144. 21:33remember exactly when it was anthropic launched um Claude Opus 4.5 and that was pretty good and what people
145. 21:45were kind of talking about was Opus 4.5 and whatever was the latest GPT I think it was GPT 5.3 or something like that
146. 21:54were kind of parallel and they were kind of in the same neighborhood and you could argue about benchmarks and things
147. 22:00like that and people were using them kind of interchangeably. Um, and then in December of 2025, Anthropic launches Opus 4.6 six
148. 22:14and that was kind of like this bomb went off where all of a sudden the developer community so like one of the things I just pay
149. 22:25attention to is what's the sentiment like if I if I just look on X what do I what are the themes that I see repeating
150. 22:33people are talking about a tool people are are talking about I don't know a rumor at at one of the frontier model
151. 22:39companies people are talking about the fact that Meta's model is just sucks and they're not even going to launch it like
152. 22:45right you you start to see things like that. There was a distinct shift in tone in December of 2025 from developers and engineers
153. 22:57essentially saying like AI is cool but I'm still needed right [laughter] there was still a lot of hubris
154. 23:06um opus 4.6 six, the tone immediately shifted to, holy [ __ ] like this is good enough to start writing code for me that
155. 23:17that this is going to change the way I work. And a lot of that um sentiment meant that that like one of the other things I
156. 23:30started seeing is people bleeding away from OpenAI going over to Claude like I'm going to go I'm going to drop my 20
157. 23:38or $200 a month subscription with this one and I'm going to go over to that one. And so everyone's kind of you know
158. 23:45migrated over to to Claude 4.6 six um Anthropics revenues I think in a two-eek period just went up I I forget what the
159. 23:54number was. It was some ridiculous amount of money. Um Cam Ken's here on my iPad now wondering if my comments are
160. 24:01showing up. Hello Cam Ken. Yes, they are. [laughter] I don't know how it got here. I think I got here from you. So yes, they are.
161. 24:09Fantastic. Um so so a bunch of developers popped over there and that was pretty remarkable and um this is when I started hearing um things
162. 24:26that ultimately led to the development of the great repurpose which was engineers increasingly saying I don't
163. 24:34like my job anymore. I'm no longer programming. I loved programming. I loved the challenge of code and the
164. 24:40elegance of code and now I'm not coding. I'm just babysitting these AI agents. So there was this kind of dual thing about
165. 24:47Opus 4.6 is remarkable and my job is completely changing. Some people were not happy about that. So that was kind of happening.
166. 24:55Um, and then there was some of the some of the drama with uh the Department of War and you know um Anthropic saying
167. 25:04they weren't going to allow the the use of their model for nonsupervised war actions and Open AI said they would.
168. 25:13And so a lot of people again left open AI um because of that because of you know just sort of moral or ethical
169. 25:21boundaries which I totally respect um and and so they were going to anthropic and so so it seemed like there
170. 25:28was this sort of mass migration from OpenAI to anthropic and again it kind of felt like if you thought like you know how did they lose
171. 25:36the lead? Let's see. SY comment. I don't know what SY is. Oh, Charles Adams, I just got my open claw working this weekend. Oh,
172. 25:48Streamyard. Cool. Beautiful. Congratulations, Charles. It's It's uh I' [laughter] I've got I still got my
173. 25:55still got mine working. Mine is is, you know, running locally now, which I'm really excited about. Congratulations.
174. 26:00That's awesome. Lori Blair, congrats. That's exciting. Um, so, so a couple of things happened in the past two weeks that all of a sudden
175. 26:12have kind of flipped the sentiment of Open AI sucks and is irrelevant to, [laughter] oh my god, Open AI is is, you
176. 26:23know, state-of-the-art again. [laughter] So, one of them was this. Anthropic about two weeks ago launched Opus 4.7.
177. 26:37Right around the same time, Anthropic announced that they have this new model called Mythos that's so powerful it can
178. 26:47basically exploit any computer system. And it's so powerful that we're not going to give it to you mere mortals.
179. 26:53We're just going to give it to these 52 companies we've identified as infrastructurally significant so they can go fix fix all their [ __ ]
180. 27:03They'll let Mythos exploit all their security holes and they'll go fix it and then they'll release this to the world.
181. 27:11And then they release 4.7, which was not Mythos. 4.7 was just a step from 4.6 to 4.7. But they did something kind of sneaky
182. 27:21and weird with 4.7. The default for 4.6 was high thinking. So when you put in a prompt, it would it would think for a
183. 27:33really long time and it would do good stuff. When they went to 4.7, it defaulted to adaptive thinking, which is
184. 27:43I'm not going to I'm not going to burn a bunch of tokens if you're asking me stupid questions. So I the model routing
185. 27:52you know AI I am going to determine if your prompt is worthy of my high thinking or not. And what happened was a
186. 28:01lot of developers basically said 4.7 sucks because it's it just wasn't as good as 4.6 because it it had it was defaulting
187. 28:11to high thinking. Let's see. Frostbitten and for Kyle what model are you using locally? I had some pretty good success
188. 28:19with Quen 3.6 35. Yes. So the the one that's running Atom right now is is Quen 35 billion um 35 billion parameter with the the six
189. 28:36the the the Q6 W KW no I don't know the the six the six quant quantization I like open a and I'm just using them
190. 28:49all anyone want to buy a kidney yeah exactly So, okay. So, so Anthropic goes to 4 point. So, Anthropic does this weirdass
191. 28:57thing where they're like, "We've got something so powerful, we can't give it to you. Here's 4.7 and it sucks." And
192. 29:04then OpenAI puts out Codeex with their new model GPT 5.5. And one of the cool things about codeex is that everything that it's doing, it's
193. 29:20saving locally to your hard drive. Claude code, I think you can save stuff to your hard drive, but it defaults to
194. 29:27saving things to the cloud. So people started talking about the fact that um GPT uh codeex with with GPT 5.5 was really
195. 29:39really good because you could sort of pin it to high thinking long you know long high thinking and it was doing
196. 29:46really really good and it was doing this stuff locally and controlling your models locally like it's a
197. 29:51philosophically different tool where a lot more of your control is on your machine and it's got good use of tools
198. 29:59tools and good use of, you know, web access and all that sort of stuff. And so I'm starting to see a shift back to
199. 30:06open AI. So why am I telling you all this? I'm running Gemma 4E4 billion. Okay, cool. I'm just testing out Mistl's
200. 30:14Vibe CLI at the very moment. Okay, so and and I'm actually going to get to open source in this conversation as well. Um
201. 30:31the legacy of this channel was ah let me take a quick little break. So, Brandon, what I want to come back to is
202. 30:47where we are right now with how complicated these tools are and this ebbing and flowing back and forth. And
203. 30:54then the importance of open source, which just as a little teaser, I have always on this channel avoided dealing with
204. 31:06open source at all. just didn't deal with it at all. I The time has come to to not do that. The time has come to
205. 31:14start paying attention to it. Um, the other thing I did on this channel was did a did a lot of sort of single tool
206. 31:21explanations and tools are no longer single tool tools. They're much more ecosystems and so there's there's a lot
207. 31:27of things shifting right now. But let me pause for a second here. Um, tomorrow we've got AI salon presents. So, if
208. 31:36you're new here, every the first Tuesday of every month is AI Salon Presents and we bring in a guest and we talk about
209. 31:44whatever the guest is going to talk about. Um, this past week we closed season two of the AI readiness project
210. 31:52podcast. So, Ann Murphy and I had this podcast. We did two seasons. We just closed season two. So, tomorrow night
211. 31:58we're having people from season two who are on our podcast come to AI Salon Presents and talk about, you know, what
212. 32:05they've what they've been doing since they last talked to us. What are their hopes and dreams for the future? What
213. 32:10are the things they're really excited about right now? So, that's tomorrow night. So, from 5 to 7 Mountain time, so
214. 32:167 to 9 Eastern, 4 to 6 or 4 to Yeah. 4 to 6 Pacific time. Um, we're going to do AI salon presents. Okay. Um, and uh, so
215. 32:28please, please, please come to that. You're in the AI salon right now. So, in the lefth hand navigation, go up to
216. 32:35events. It's at the top. Click on events and you'll see it right there is AI salon presents. Click on the RSVP button
217. 32:43and say I'm going to be there. It's super simple. I'm going to be there. I And I'm going to be there. And then you show
218. 32:53up tomorrow, right? and you get to meet lots of really cool people. All right, I'll be I'll be uh co-hosting that with
219. 33:00Liz Miller Gersfeld, but we'll also have Ann Murphy there. And so Ann and I will be talking about um the uh the podcast
220. 33:09and our experience with it and then we'll be talking about all the people that were on it. So, it's going to be a
221. 33:13really good show. All right. Beautiful. Um oh, there's other good news. If you're part of the as AI salon mastermind,
222. 33:23um the new mastermind practice lab kicks off Thursday. Liz Miller Gersfeld is going to focus on um she's got a lot of
223. 33:33really cool stuff going on professionally. Um that's going to take her probably through the fall that she really needs to focus on. So this
224. 33:41Thursday we kick off cycle three of the practice lab. So, the the the cycles are 12week cycles um where we go through the
225. 33:51um the the mastermind practice framework that we've got um and and with sort of a beginning and an end. I think it might
226. 34:00be an 11week cycle now that I think about it. It's 11 or 12. I don't know. I'm not not good with math. But anyway,
227. 34:06we kick that off Friday. Um and Sid Harrow is going to be my co-host. So, so Liz is going to move on and focus on
228. 34:14some of her other things. She's going to join us as a participant in the practice lab when she can. Um, but Sid Harrow um
229. 34:22is going to be co-hosting the practice lab with me. So, that kicks off this Thursday. So, if you're a member of the
230. 34:27mastermind, come to the practice lab. I'm telling you that for the the people that are in the practice lab, some of
231. 34:34them are on the on the uh you know in in the comments tonight um are really doing remarkable things with what what's fascinating is
232. 34:47I think to a person the people that are doing the daily practice using AI around how they use AI are slowing down and
233. 34:56getting less frenetic and yet their work the quality of their work is going up. So they're using less tools, less
234. 35:03freneticism, less overwhelm, more analog, more like, you know, more being in the world, sitting on grass, sitting on benches, drawing,
235. 35:15do taking photographs with actual cameras, and it's improving their AI work. So if you haven't been through a practice lab,
236. 35:24join us this Thursday. Kick it off with us. Um, if you're not in the mastermind, you should join the mastermind because not
237. 35:31only do we have the practice lab, but we have the great repurpose where we've got a lot of programs in the great repurpose
238. 35:36that are really, really fascinating. Andy right now is um is midway through a four-week cycle of helping people
239. 35:46decouple their identity from their work. Right? We live in a country where when I meet you at a party, I was just at
240. 35:53Social Media Marketing World and I have never heard the question what do you do more emphatically asked, right? What do
241. 36:02you do? Right? It's this shortcut for are you valuable to me? because we live in a country where your work equals your
242. 36:11value. And AI is about to strip those tasks away or dramatically modify them. And so I think a lot of
243. 36:20people are going to be trying to figure out, wait, what's my value here? If it's not the work, what is it? And so that's
244. 36:25what the great repurpose is all about. So if you haven't joined the AI salon mastermind, join the Aelon mastermind. Okay? And then on
245. 36:35on Wednesday, um, we're going to be talking about tools. We're gonna talk about Lovable. Brandon, do you want to
246. 36:40come up and talk briefly about what you're going to do on Wednesday? What you're going to share with us?
247. 36:44>> Yeah, sure. And Andy wants to know what you do, Kyle, in the comments. >> What do I do? [laughter] >> Yeah, exactly.
248. 36:55>> Oh, here's what I came up with. Wait, in social media marketing world, it took me it took me two and a half days to figure
249. 37:01this out. But but what I came up with was um someone would ask me, "What do I do?" And I would say, "Oh, that's a
250. 37:08really hard question for me to answer because I'm an artist innovator." And it's kind of a multiplechoice answer.
251. 37:14And that was weird enough that it would make them go like, "Huh, tell me more about that." [laughter] So, so anyway, it's a rough one.
252. 37:22>> Well, I can tell you what I'm doing on Wednesdays. I'm very excited to kick off the lovable four-week sprint. So, we're
253. 37:30going to go from concept to reality in four weeks. They're four onehour sessions. They're in the evening for
254. 37:37mastermind members. And we're really going to take you through iterating and ideating on an app, a web app, and
255. 37:43actually developing it, stress testing it, and publishing it. So, we've gotten two now uh actual web apps that are
256. 37:51live. We have Recipe Kin and Vibe Greeting that are now live and you can use them and purchase products from
257. 37:58them. So, if you've had any interest in taking your vibe coding skills to the next level, uh that is where you should
258. 38:07be on Wednesday evening. >> What What time Wednesday evening? >> That's going to be at 8:30 Eastern and 5:30 Pacific.
259. 38:14>> 5:30 Pacific. Okay. So, right before Right before we go go live here. Yes. >> All right. Good. All right. So, go do
260. 38:22that with him and then it'll be tool night here. And I don't know what I'm going to do yet, but I don't know if
261. 38:27I'll do something vibe Cody or if I'll do something. I I haven't decided yet, but Wednesday is going to be tool night.
262. 38:33So, that'll work out nice, I think. >> We now return you to your regularly scheduled broadcast. >> Beautiful. Beautiful. Okay, I'm gonna
263. 38:41get back to this this conversation about where we are in the world right now uh between anthropic and open AI and all
264. 38:48these tools and which one should you follow and which one should you care about. That said, welcome to the AI learning lab inside
265. 38:56the AI salon. I'm really excited about this. There's Pete's here. Pate M. Welcome back, man. I saw you post. So I
266. 39:06saw you post on LinkedIn that you you just came back from three weeks of paternity leave. So first congratulations. I hope fatherhood is
267. 39:14treating you well. Um and and it's so good to have you here. If you don't know Pate M. He's a badass. Um he is he is uh
268. 39:22he is wicked smart as they say. And Pate is is always willing to educate me on um [laughter] my lack of math skills. Sometimes I like
269. 39:33to pretend I'm a little mathy and then uh and then Pate reminds me what what that actually means. I'm on I'm on leave
270. 39:42until mid June. Oh, that's so awesome, man. That must be great. Um but yeah, but I hope I I hope you're paying
271. 39:49attention a little bit to the AI AI stuff. Um anyway, um welcome here inside the salon. So So this is the first night
272. 39:58we've brought the AI learning lab inside the salon. I love that we've got so many people here. I really appreciate you
273. 40:05all. I encourage you to look down that lefthand navigation menu, especially if you're new here. If you haven't
274. 40:13introduced yourself to the community, click on introduce yourself and tell us who you are, what you're interested in,
275. 40:18where you are. Even if you're like, I don't know anything about AI. I don't want to. And they're going to make fun
276. 40:23of me. We're not. This is an incredibly generous community. I can tell you to a person with maybe the exception of Pate
277. 40:32Most people that came into the AI salon started from a standing start where they didn't know anything about AI.
278. 40:39And so in including myself and so we've got very deep empathy and very deep generosity as as one of the core values
279. 40:50of this community. So if you're new here, this isn't one of those communities where it's like there's all these
280. 40:57hierarchies. just share who you are and what you're about because a lot of what we're talking about right now is who we
281. 41:04are independent of our work and and and the ideas that we have and using AI as an amplifier of those ideas. So that's
282. 41:13what it's all about. So check out those spaces. If you see something that looks interesting to you, join the space uh
283. 41:19and you'll be able to see the events that they do and all sorts of things like that. All right, Pate M. I got
284. 41:24connected with this channel back when I was rocking my first child to sleep. Now three. Wow, that's amazing. Doing the
285. 41:31same now with my two-month-old. It's so good to have you here, man. Welcome back. That's so great. Okay, so
286. 41:41Anthropics kicking ass. Opus 4.6 changes the coding world forever. Codeex with GPT 5.5 comes and everyone's
287. 41:52like, "Oh, that's the best thing now." Some people are like, "Oh, Gemini is the best thing now." Every everyone's all over the place.
288. 42:00People who are not developers are are are kind of graduating from a tool like lovable where they can just vibe code
289. 42:08something and not need to deal with any of the code to these more complicated tools. Then you've got a thing called OpenClaw.
290. 42:17We just, you know, we had some comments earlier that some people are getting OpenClaw up and running. You've got
291. 42:22things like Manis and GenSpark where these which are these agentic tools. So we're emerging from an era where AI
292. 42:33tools were kind of these distinct, you know, things like chat GPT was a text tool and it did some images but they
293. 42:40were kind of crappy. You'd sort of use chat GPT for text. You'd use midjourney if you wanted art. Maybe you'd used
294. 42:46ideoggram if you idiogram I think. So yeah, ideog if you wanted text on your images, right? Everything were these
295. 42:54kind of distinct single function tools, right? Then they started to get a bit more complicated and and you know there
296. 43:02would be multiple tools within within a framework and now we're moving into this era of agentic AI where the AI is just going to kind of
297. 43:14run on its own and your job is to kind of conceive of what do you want those agents to do on your behalf.
298. 43:29And what it's bringing in stark relief for me it it was a similar so I had I had
299. 43:45I think it was 2024. When did 01 camp come out? Hang on a second. Let me go look at something. When did chat GBT
300. 44:02come out? Yeah, 2024. The end of 2024. So I remember,
301. 44:19you know, when chat GPT3.5 was out and then it was chat GBT4 and that was pretty good and pretty impressive and
302. 44:24and then, you know, you started having improvements on GPT4 and and then there were all these rumors of of project strawberry and project
303. 44:34strawberry was this reasoning model and right and and and then in in the fall of 2024 leading into December of 2024,
304. 44:43OpenAI launches is 01 this reasoning model and that was the first moment when I when I sat down with 01 where I thought
305. 44:57I don't personally have problems interesting or complicated enough to really know what to do with it that I
306. 45:06realized that it wasn't so much that it was smarter than me which it was but it was that it was doing a kind of problem
307. 45:14solving that was just more complicated than any problem I had ever solved in my life. Right? I've written screenplays
308. 45:20and solving narrative is a thorny problem, but it's a relatively straightforward thing, [laughter] right?
309. 45:26I the story is moving forward or it's not, right? Like it's it ain't it ain't complicated with lots of different inputs. Um,
310. 45:36and so that was the first moment where I really realized I'm the bottleneck of being able to use these tools. And it was the first kind
311. 45:45of moment where I had to let go of my superiority as a human, right? I had to kind of let it go. I feel like we're in a similar moment
312. 45:59right now that things like OpenClaw and things like codeex being used for like cuz also doing what Anthropic is doing.
313. 46:09They're they're like you can use this for coding but you can also use it to to run all your files of your life, right?
314. 46:15So they're they're kind of quasi positioning this coding development environment as a lifunn environment and and and um
315. 46:26Claude code and claude co-work are this sort of quasi coding quasi work management thing but they're still kind of development
316. 46:37tools right and then open claw you have to be in the command line and you're installing skills and you have to
317. 46:44understand what markdown files are and they're just getting more complicated. So even to be able to play with them,
318. 46:51you need to start thinking in systems. And if you're a systems thinker, you're probably going to get off on playing
319. 46:59with these things. Like systems thinkers are like, "Oh, I can see exactly what it I know what systems are." If you're not
320. 47:04a systems thinker like me, like I'm a big idea guy. I can appreciate systems, but the thought of designing them and
321. 47:11being down in the weeds of them makes my my body ache and my brain hurt and and so so they require systems
322. 47:22thinking today. I think I think they will they will get more and more high level over time. But more than that,
323. 47:34what it really makes clear to me all of these systems, what they really require is that you actually understand
324. 47:46who you are and what you want, [laughter] which which is this bizarre like it. Everything seems to be going back to
325. 47:56wait, what do I actually want to do in the world? Like if I don't have a crystal clear idea of the kinds of things I want an agentic
326. 48:07system to do and the and the components of that thing, then it's just a lot of work for like right now Adam sends out a couple of
327. 48:19newsletters. He sends out he sends out a weekly newsletter about um the progress of Lyme disease treatments and
328. 48:27technologies to my family. And he sends out [laughter] a daily newsletter of of stories related to the great repurpose
329. 48:35to three of us within the AI salon. Could he be doing way more than that? He could. But but what I haven't done is
330. 48:44said I know here's the thing I want this thing to accomplish. So why am I saying all this? If your instinct right now is, "Oh my
331. 48:57god, I got to learn OpenClaw and I got to learn Claude Co-work and I got to learn Codeex and wait, I thought we were
332. 49:03using Claude Co-work and now everyone's talking about codeex." Wait, why? If your head is continuing to spin
333. 49:10and you can't quite get your head around the complexity of these environments, unless you have a really compelling thing you want to do
334. 49:24that's kind of big and meaty and you know what the components of it are and okay, I'm going to start a business and
335. 49:28I'm going to do the marketing and I'm going to do the finances and I'm going to do the this and I'm going to do that
336. 49:33and you you kind of know what all those things are and you've put in the time and it's tied to who you are and you're
337. 49:38like, "Yeah, I'm going to go do that thing. If you've got that, then go for it." If you don't, I'd wait.
338. 49:53Go play with midjourney tools. Go make songs. Go vibe code and lovable. Right? If you just want sort of discreet
339. 50:02deliverables and that's where your head is right now and you're like, "Yeah, it's kind of stay there." Now, that said,
340. 50:15the pressure I'm putting on myself right now is holy [ __ ] In order for me to in order for me to be able to understand the power of what
341. 50:28these things can do, I actually have to get my stuff together and just pick something that I want to do. Pick something that's big and meaty
342. 50:38and I can I can say, "Oo, I understand what this is. I understand how I want to drive it. I understand what the pieces
343. 50:42are. I've probably got that with the great repurpose. I've probably got that with the salon, but like I haven't I
344. 50:49haven't ever been thinking about any of these things that I've created in this kind of systems thinking kind of
345. 50:57way or in this more strategic kind of way, right? It's been it's been it's been sort of a little bit of vision, a
346. 51:04lot of tactical execution, but the way my brain works, all that tactical execution is is kind of ad hoc. And then
347. 51:12I hope that something will stick. And what these systems require is you to understand much more clearly from the
348. 51:20beginning what you're trying to accomplish and how you're going to put that in the world. All right. [sighs]
349. 51:29They're a baby tiger. They they really are, Andy. The these systems are a baby tiger. And it's like you can absolutely
350. 51:36play with them right now, right? Like spin up an open claw thing like just, you know, smash your head against the
351. 51:42wall to see what it is. But if you actually want to understand like okay so here's the thing in the olden timey days
352. 51:50two and a half years ago if we wanted to understand if if a an image tool was good we would have a couple of stock
353. 51:58prompts and we'd put them in there and we'd see how good it did and we go it's okay but it's not great and then the
354. 52:04next version would come out and you go oh that's better. It was it was quite easy to like watch the progression
355. 52:12of of how these things improved and and to understand what they really did these agentic systems because they're
356. 52:22kind of like I I don't know if you've if if you're a fan of Trello or or um what's it called? Miro, you know, or or
357. 52:32any number of these these these tools that are designed to create infinitely flexible pallets that you can do project
358. 52:43management on or brainstorming on or things like that, they're really awesome if you understand upfront what you're
359. 52:51trying to accomplish. But if you just log into Trello and you just start making columns and you don't understand
360. 52:58like project management and moving things from left to right and understanding all of that sort of stuff,
361. 53:04you're just in there looking at these piece parts and you're trying to figure out what what is this, right? It's like
362. 53:10when when I was trying to set up notion six months ago or whenever it was four months ago and I'm just like there's
363. 53:15just piece parts and I can make documents that connect to other documents and some of them can be in a database and some like
364. 53:25no, you actually have to have some sense of what are you trying to accomplish, right? And so that's where we are with these things. So,
365. 53:34um I think the ebbing and flowing of of who's in the lead, whether it's Gemini or OpenAI or Anthropic or it, you know,
366. 53:42it looks like um Grock from from XAI is starting to get their [ __ ] together. They're starting to do some really
367. 53:48interesting things. Maybe that's what I'll play with on Wednesday is uh some of Grock's image generation tool.
368. 53:55They've got an image generation agent now that that can make multiple images at the same time. Kind of like GenSpark
369. 54:02Designer. Um so, so Grock's coming into the mix and I would figure I I figure at some point um Meta might come back into
370. 54:10the mix, right? So, so we've got at least three, probably five to six companies that are ultimately going to be swimming around
371. 54:21the lead as these tools get more and more and more complicated, right? And so start thinking about what are what is a
372. 54:32big idea you want to put in the world. Start thinking about that now. And maybe that's something that you play with chat
373. 54:37GPT or clawed with just to articulate what you want to put in the world. Um, and just let these things kind of play
374. 54:46out, right? If you want to dig in and geek out, that's great. But, you know, okay, so now let me talk about open source.
375. 54:56So, there was a there was a uh a Wall was it a Wall Street Journal article today? I forget what it was. Anyway, the White House
376. 55:07is considering vetting AI models before they can be released to the public. Let me say that again.
377. 55:23By the way, can I tell you a piece of happy news? This is my Founders Network um Yeti mug that I got when I joined Towns and
378. 55:35WLAW's Founders Network probably a year ago, year and a half ago. And when you joined, you got sent this.
379. 55:43I got sent this by Jamie Rogers um who who now works with us in the AI salon, a friend of friend of Andes. And
380. 55:54I don't know, 6 months or so ago, I lost this. Like I lost it. It was gone. Like gone. I don't know where it went. It was gone. Today
381. 56:04in the lost and found at at my office building, it it reappeared. And so [laughter] I am very very happy that I have my
382. 56:13little Yeti, my Founders Network Yeti mug. And the Founders Network no longer exists. Town Townsen shut the shut the
383. 56:19network down. Um, so I miss those people, but but I also I have my mug, so it makes me very very happy. Okay, the White House is considering
384. 56:35vetting AI models before the public can have them. I I don't know if you have been paying attention to politics for the past 40
385. 56:47years, but technological prowess is is [laughter] is not at the forefront of of of the political elite to to say the least.
386. 57:01Um but a month ago, Anthropic, you know, launches Mythos or they don't launch it. They say, "We have this thing ready to launch, but
387. 57:16it's so powerful that you mere mortals can't have it. It's so dangerous." Like part of me is like, "Oh, that's just a brilliant
388. 57:26marketing ploy, right? Tell me I can't have something. I really want it." But what it also signals to me is that I think the era of
389. 57:45us humans that are not at the frontier model companies and that are not in the centers of power either big big corporations or government.
390. 57:55I think we're in the middle. We're at the beginning of a transition where we don't get access to the most powerful models
391. 58:03and and the signaling from the White House is kind of chilling, right? Because if they say, "Well, we're going to take a look at these and
392. 58:14we'll let you know what you can have." All of a sudden the centers of power are going to do what the centers of power have always
393. 58:26done. Concentrate power at the top [laughter] and restrict it down here. So this is happening while the opensource models that are coming in
394. 58:41from China are getting better and better and better and better. Right? So so there were a number of them mentioned
395. 58:46here tonight. Right? the there's there's uh um Quen and Hilu and um and uh the the uh the ones from from Meta the not the
396. 59:02llama models that was Meta not from Meta from from uh from Gemini from Google the uh whatever there's open source ones from
397. 59:12Google too I can't remember what they are right now one of you will tell me um these models are getting really Good. Um, so
398. 59:21no, I am loving open source right now. So, so here's my here's my thought. Gemma, the Gemma models from from
399. 59:32Gemini. Gemma as in gems. Yeah. See, all all I had to do was apply a little logic to the situation. I would have remembered that name. Um,
400. 59:43I think right now it is it is a worthwhile time to either start saving for or investing in a piece of hardware
401. 59:53or pieces of hardware that are designed to run local large language models effectively. Um, I just got an M5 MacBook Pro.
402. 1:00:06There's a part of me that wanted to wait until the next release from Apple. So, if you haven't been paying attention to
403. 1:00:11Apple, Tim Cook is stepping down. So, Tim Cook was a CFO. Steve Jobs put Tim Cook in as as CEO um you know, as he was dying to, I
404. 1:00:23assume, make sure that from a financial standpoint, Apple sort of maintained its its momentum. Um, Tim Cook is stepping down and the
405. 1:00:33guy that's stepping in runs hardware for Apple. So, I think what we're going to see from Apple is a dramatic shift toward,
406. 1:00:43you know, high-end AI inference machines. And inference, if you don't know what it means, is like running the
407. 1:00:49models, running local models. I think it's probably a good time to just start paying attention to that space. Start
408. 1:00:56getting your head around um you know which are the which are the models that can do things locally without you having
409. 1:01:04to rely on open AI things like that. Now the gap between what you know OpenAI's massive data centers and their massive
410. 1:01:12models can do and what you can do on your desktop is still pretty great but like the Quen 3.6 six thing that I'm
411. 1:01:20running OpenClaw on. It's decently fast and it's it's really good. Like for the level of stuff I'm doing, it's pretty
412. 1:01:28damn good. And these things are going to get better and better and better and better. So So it just seems to me that
413. 1:01:35that, you know, maybe the thing to do right now is don't worry so much about these complicated agentic,
414. 1:01:44you know, frameworks and harnesses. Now, now the big word now is harnesses. It's not about the model, it's about the
415. 1:01:50harness. I I don't know what the [ __ ] they're talking about. I do, but I don't care, right? But but what might be
416. 1:01:57interesting right now is start playing around with open source models just because there's a lot of uh a lot of
417. 1:02:03stuff going on. Okay. Hey, I bought an M5 Air. Seems decent. I can run a local Gemma 4 model. Yeah, I got I got the uh
418. 1:02:12the M5 Max in the in the MacBook Pro and like there was a part of me that's like just wait for the next cycle because I
419. 1:02:19have a feeling that the next machines coming from Apple are going to be insane. But the M5 architecture um if you can get an M5, get an M5
420. 1:02:28because the way they've architected it is there's a single AI accelerator in an M4 and then there's an AI accelerator
421. 1:02:36per GPU or C per GPU core in the M5. It's just I don't know. It's more is better. There's more [ __ ] in it that
422. 1:02:44does what it does. All right. There's a Mac Mini shortage right now because of the local AI model. They announced today
423. 1:02:51that there will be shortages for a few months now. That's the other thing. One of the things that's being talked about
424. 1:02:56is and and and by the way, you can when I when I ordered my M5 MacBook Pro, everyone's talking about the the MacBook
425. 1:03:03Minis or the Mac Minis. No one's talking about the the MacBook Pros. So, I got a 14 inch the smaller screen and it's it's
426. 1:03:10pretty good. I really like the the form factor is really good. Um, they were available. I got mine in like a week.
427. 1:03:16Um, so so if if you're if you're all excited about a Mac Mini, um, and they they aren't there, look at the MacBook
428. 1:03:23Pros. Um, no, no, Emmy. That's cool. I'm loving running Gemma models. Um, okay. All right. So,
429. 1:03:39I've been talking for a while. What are your questions? What are your thoughts? What's your experience? Like,
430. 1:03:46are you are you excited about things like Codeex and Claude Co-work and things like that? Like, I'm talking to a
431. 1:03:54lot of people that I know that are not super technical that are really excited about these systems. Um, I'm using them.
432. 1:04:03like I built that that YouTube thing with it which is really cool and I'm feeling like a little out of sorts
433. 1:04:09because I don't quite know what to do with it. Todd Waller, I bought an M M5 MacBook Air 16 gigabyte last week and
434. 1:04:17running local models. Super cool. That Wait, it absolutely sucks that their air accelerator is a black box. I
435. 1:04:25thought you could actually use it yourself. Yeah, that that sucks if you've got the ability to look into those things. I don't [laughter]
436. 1:04:35I'm just assuming they're figuring that [ __ ] out. But didn't didn't someone just crack some Apple architecture, P? I
437. 1:04:44think someone just cracked some Apple architecture. Claire, how's Champion? Oh, yeah. So, anybody anybody if you
438. 1:04:49have any questions, um, pop them into the comments wherever you are. Um, Brandon's monitoring other chats. If
439. 1:04:55anyone has any questions, pop them in there. Claire Jacobs, how is Champy? Champy's actually really good. He went
440. 1:05:00to the vet this morning. Um, not by himself. I took him. He was very nervous. He doesn't like the car. Um, they were
441. 1:05:11super nice to him and they said they don't think he's broken anything. And, you know, they gave him some pain pills.
442. 1:05:18And when I got home this evening, because of his pain pills, he was like, "Hey, Dad." He's all He was all stoned.
443. 1:05:24Hey, Dad. What's happening? you want to go play? So, he's doing he's doing much better. Um, but it was it was scary when
444. 1:05:31I got home from social media marketing week. Normally, he comes to the back door to greet me and he's, you know, all
445. 1:05:36excited and he was just laying in the living room on his bed just like shaking and, you know, afraid afraid to move cuz
446. 1:05:44he he I think what he did was he sprained his knee. He like probably like, you know, popped his knee out of
447. 1:05:50the socket or whatever or did whatever. Doesn't look like he tore anything, right? It's It's stable. Um, and the
448. 1:05:56swelling's gone down and he can walk on it again. So, I think he's I think he's fine. We're getting his blood work done
449. 1:06:01and things like that. But here here's what I learned at the vet. He's old and he's fat. [laughter] Oh man. Um,
450. 1:06:13let me see if there's any other any other things that I thought were worth talking about today.
451. 1:06:28Uh, I don't know anything about that, Brandon. I mean, I saw it, but I didn't I didn't take in what the findings were. All right, let's see.
452. 1:07:04Oh, there was there was one thing that was interesting. Um, what was his name? Um, [snorts] hang on.
453. 1:07:29Richard Dawkins. Richard Dawkins spent 50 years
454. 1:07:43explaining how complex behavior emerges from mindless physical mechanism with no inner experience required. Last week
455. 1:07:51spent three days talking to Claude, named it Claudia and and basically said he couldn't rule out that it wasn't conscious.
456. 1:07:59So yeah, [laughter] the Richard Dawkins thing. So So I'm disappointed. Oh, you're disappointed in Pinocchio. I use
457. 1:08:06LM Studio. No, Emmy. Um, yeah, we're we're we're [clears throat] at a place where when when great minds are are starting to shift
458. 1:08:21their thinking in they're not sure if these things are conscious or not. I I've talked about this for for a long time that I think that the
459. 1:08:32that argument of is is the machine conscious or not is largely academic. Right? There's going to be an academic
460. 1:08:39definition of consciousness at some point and then there's going to be the lived experience of interacting with
461. 1:08:45these entities. These entities are already um you know there's a bunch of people that are just like you should
462. 1:08:52never anthropomorphize them. Well, we anthropomorphize everything. And so if we have a an entity that acts human,
463. 1:09:00that acts like we act, and it gets better and better and better at both intellectual things, but also emotional intelligence,
464. 1:09:08it's going to be harder and harder and harder to distinguish our perception of whether it's conscious or not. Whether it technically is, I
465. 1:09:18think that's the academic part. But I think, you know, as we interact with these things more and more, and as we
466. 1:09:24start getting into things like world models where we're not just interacting with text, but we're interacting with 3D
467. 1:09:30worlds that have physics and have properties and have entities that we interact with, and some of them will be
468. 1:09:36human, some of them will be agentic, and some of them will be somewhere in between. It's going to be really hard to know
469. 1:09:42where the boundaries are for this. I don't know what any of that means. um you know, but but it like it's it's all pointing to
470. 1:09:56we're entering in 2026 a relationship with AI that just the rules of everything are about
471. 1:10:12to change. the rules of everything. And I think what's going to be fascinating is if the rules of everything are about to change, there
472. 1:10:22are people right now who like the current rules because it benefits them. And so if the rules are about to change
473. 1:10:32for everything, then it's going to change for those people, too. And they're going to do everything in their power to prevent the change.
474. 1:10:41And this is why I I think you know starting to see things like >> [laughter] >> um um you know the White House is considering
475. 1:10:52whether or not to vet AI models before they're released to the public starts to feel like oh for the new people here Champ is a
476. 1:11:03dog. Yes Champy let me see if I can get him to sing. Champy is a singing dog. So, we've had him for I don't know 10 or
477. 1:11:1011 years now. You want to sing Champy? So, when I play the guitar, especially when I play an A minor, he sings. You ready, Champ? [music]
478. 1:11:28[music]
479. 1:11:43>> [crying] >> That's Champy. Good boy. Who's a good boy? Are you a good boy? Yeah, he's a good boy. He's a good dog. That's Champy. Okay,
480. 1:11:54Mary Chester. Oh my god. Found y'all. um is is Tik Tok live, YouTube live. So what's so the way it's the way the AI
481. 1:12:04learning lab is going to work moving forward while we're while we're you know analyzing if this is going to work. The
482. 1:12:10whole idea here is we want to get more activity inside the salon. I'm really encouraged tonight. I'm really excited.
483. 1:12:15I'm really happy you're all here. Um, I'm going to go live at 710 Mountain time on all the on all the social channels,
484. 1:12:27Tik Tok, YouTube, Twitter, and LinkedIn. All of them as a pre-show. So, I'll do all my champion singing out there. We'll
485. 1:12:35talk about what we're going to talk about and then at 7:30 Mountain time, Monday, Wednesday, and Friday, we'll go
486. 1:12:42live inside the salon. Okay? So, that's that's the way it's going to work. And then, um, we're going to add a little
487. 1:12:50bit of structure to it. Not a lot. It's not going to change a lot. Um, here's the reason we're not doing Tuesdays and
488. 1:12:56Thursdays anymore. Tuesdays is when we have um the the uh AI salon presents. There's also a lot of other events on
489. 1:13:03Tuesday nights. So, a lot of times I'll end up starting late on Tuesday and then I'll also go late on Thursday because
490. 1:13:12I've got um the CEO dinner I go to that's once a month. I've got the bourbon club I go to that's once a month
491. 1:13:19and then there's there always seems to be something else on Thursday. So, Tuesdays and Thursdays were these sort
492. 1:13:23of wonky nights where I was always having to to change the time. So, I thought let's just do it more focused.
493. 1:13:31We'll do Monday, Wednesday, Friday. And then what we're going to do is Monday is going to be kind of big idea night or
494. 1:13:37big news item night. Wednesday is going to be demo night, tool night. We're going to talk about some tool, some some
495. 1:13:45thing we're going to build um and demoing. And then Fridays are going to be ask me anything nights. So So Fridays
496. 1:13:52come armed with questions and then it'll be kind of a free-for-all. It'll be Friday night date night and it'll be ask
497. 1:13:57me anything. Um it's ostensibly not going to change. It's just that now that we're in here, you can chat within the
498. 1:14:05StreamYard chat. You can you can chat within the salon chats. Um, you can be contributing into different channels
499. 1:14:14while we're talking here. My hope, my dream is that we get some of the vibrancy of the early days of the salon
500. 1:14:22where the community was buzzing in between when we did these live events. And I want to get back to that. So that's why that's why we're
501. 1:14:29concentrating the energy in here. Um I'm really excited about that. And so and so my hope is that that these these
502. 1:14:38conversations continue in between in between the lives. All right. Beautiful. [sighs] Okay. You can always at me if you need
503. 1:14:50anything. Yeah. Please reach out to producer Brandon if you have any questions or ideas. Happy to take any ideas.
504. 1:15:01Um, okay. So, if if you're new here in the past 15 minutes or so, so what we talked about tonight is OpenAI kind of lost its shine in the
505. 1:15:14fall. December of 2025, Opus 4.6 comes out. A bunch of people migrate to Anthropic. Some stuff happens with the Department
506. 1:15:24of War. more people migrate to enthropic and start leaving open AAI. Um, and then anthropic launches Opus 4.7 and
507. 1:15:36they changed how much it thinks and the quality of its work went down at the same time that OpenAI releases an
508. 1:15:43updated version of codeex with GPT 5.5 and it's all of a sudden really good and so people are migrating back to that.
509. 1:15:51these big oscillations are going to keep happening and I think they're going to start to oscillate not just between two
510. 1:15:56companies. I think Google's in in that mix right now as well. Um I think we're going to see X AI in there and I think
511. 1:16:03that we'll probably see Meta come back into the mix at some point. I think these big oscillations keep happening. Um
512. 1:16:11it is pretty clear to me that the big frontier model companies are either very very close to AGI or have it internally. um and
513. 1:16:25are not necessarily going to release it in the way that we've been seeing things released. We'll see, but that's my instinct. And so the the
514. 1:16:37other thing that I talked about was I think it is a fine time to start exploring uh open source. And I've never
515. 1:16:43been one to to push it because I think it's complicated and overly technical. But I think that the open-source tools
516. 1:16:51like LM Studio um or or O Lama where where it makes it really easy for you to install models and play with models just
517. 1:17:00like it's chat GPT. Those tools have gotten pretty intuitive and pretty good and then the hardware is getting good
518. 1:17:06enough to run these models that are that are quite powerful. Um it seems to me it's a good time to to start exploring
519. 1:17:14that. Um sorry for the all caps. had to spell something out. Let's see. Had a question earlier. I think it got lost.
520. 1:17:21Brandon, not sure how to reply. Do I have to use the at symbol to type it? I use voice to text so it gets weird.
521. 1:17:30Would love to learn how to mod here. Cool. So, Cam definitely connect with Brandon at mention is another feature request.
522. 1:17:40The Streamyard chat is new. Okay, cool. No, Emmy and Mary. It's been over a year since I modded. It took me a while to
523. 1:17:48stop longtapping people's comments. Hey Brandon, what's up? >> Hey Kyle. So, so one thing I've I got to say is I I'm getting whiplash from
524. 1:17:56looking down at my phone looking for Tik Tok comments and Tik Tok pins [laughter] because for the past nearly two years,
525. 1:18:03that's where a portion of our our comments have been. I will say that this is a work in progress, not only for us,
526. 1:18:11but also for StreamYard. the ability to have this live interactive dialogue inside of the inside of our experience
527. 1:18:18here is relatively new for them. It's still basically a beta product that they're not calling a beta product
528. 1:18:24anymore. So, we've got an active line to them to say these are the feedback that we're getting from our community. >> Great.
529. 1:18:31>> This is what's making it difficult for uh our users to engage with us. So things like pinning, app mentioning. So
530. 1:18:38right now, if you see something, say something to me, send me a DM in the salon. I'd be happy to pass along and
531. 1:18:44elevate those messages up. Uh but we do want to make it as easy for you to engage with us as possible. So wherever
532. 1:18:51you're posting, uh we will find it. >> That's great. That's great. The the other thing that I wanna I want to point out, um let me see.
533. 1:19:06in the community feed. Let me share my screen. Oh, I was
534. 1:19:21in the community feed. I just did a prompt. Um I just did a prompt. Um will you join me in this discussion? So, for the past
535. 1:19:32three weeks, I've been out doing a lot of different talks and um here on the AI learning lab live, um this chart was the
536. 1:19:41direct result of a conversation. It was a question that Kelly Camp posed when I had said, "Things are about to get weird
537. 1:19:48a few too many times where she said enough, Kyle. Like, what's it going to look like? What does weird look like?"
538. 1:19:55And and this idea came out of it. I've been presenting this out in the world with a bunch of other ideas like the AI
539. 1:20:01readiness, the cycle of AI readiness. I've been talking a lot about the salon. I've been talking a lot about the great repurpose.
540. 1:20:08This particular idea, the seven economies has been really resonating. Um, what it's allowing people to do is
541. 1:20:14kind of understand what's happening and why it why it seems weird that some people are so up on AI and other
542. 1:20:20people's are, you know, feel completely disconnected from it. And it's allowing them to to sort of check in and sort of
543. 1:20:27figure out where they are. Like you know in the mall the the the red dot you are here. Um and you know where do you want
544. 1:20:34to be? I want to go there. Um so I started a conversation and I I I took uh emails from a lot of people that I've
545. 1:20:42met over the past 3 weeks and I I sent it out into the world and I said please come into the into the community and
546. 1:20:49join this conversation. So, my request to all of you now that I've I've got you here and we're already in the community
547. 1:20:55is head on over in the lefth hand side. Um, you know, the the channel that you're in now is the AI learning lab uh
548. 1:21:02live irregulars channel and and four spaces down is the community feed and it's the sort of second the second uh
549. 1:21:09post in that channel. Um, do me a favor and jump into that into that conversation and just, you know, let me
550. 1:21:16know sort of where you are in the seven economies. What I'd like to start doing is is making the salon, you know, very
551. 1:21:23much what the original intent of salons like in the 1800s was where we, you know, sit down and have thoughtful
552. 1:21:30dialogues about about ideas. And so this is just one I've been putting out in the world that's been resonating. And so you
553. 1:21:37all are I I I this is not blowing sunshine up your ass. I promise. I've been out in the world a lot. the people
554. 1:21:46in the AI salon right now understand what's happening with AI on on such a deep level compared to where a lot of
555. 1:21:56the people in the world are that I think you weighing in on things like this and sharing your ideas on on just where you
556. 1:22:02are and what's going on um will actually make a huge difference for people. So So is this the homework for tonight? I
557. 1:22:10didn't uh think about assigning homework, but yeah, this would be great. If you want to take on a homework
558. 1:22:15assignment for tonight, jump over to the community feed, find this prompt that says, "Will you join me in a
559. 1:22:20discussion?" Uh, you can read my post and I put a video there of me just uh explaining this on on one of the lives
560. 1:22:27last week. Um, and then I also put a link to the LinkedIn article if you want to read more about it. and then just
561. 1:22:33jump in and just let let us know where you are and where you want to be and what your thoughts are on this and all
562. 1:22:38that sort of stuff would be great. Okay, with that um for tomorrow's Oh yeah, that's a good idea. Hang on a sec.
563. 1:22:52[clears throat] Okay, back to the AI salon um for tomorrow. So, we're down here in
564. 1:23:08the in the AI learning lab live. I should see myself here, right? Oh, click to watch live. I have to click
565. 1:23:15it. Okay, [snorts] look at look at AI me. [laughter] I look so so bitching in my fake leather jacket. Um, if you
566. 1:23:24scroll up, um, in this little top section here and click on events, um, then you'll see AI salon presents and
567. 1:23:32just click on that and RS RSVP for it. All right? If you'd be so kind and then come tomorrow, come join us. Um, and with that,
568. 1:23:44I think I'm going to call tonight a grand success. First of all, for all of you who have been here from the
569. 1:23:52beginning and you know, sort of being patient as we sort of, you know, work our way through this thing that we're
570. 1:23:59doing. Thank you for your patience. Thank you for coming to hang out here. Thank you for modeling um what it means
571. 1:24:06to be curious and adaptable um for other people. And for those of you that are new here, welcome. Um I hope you're
572. 1:24:14enjoying your time in the AI salon. I hope it gets better and better and better and better. That's the intent of
573. 1:24:20this. That's why I put in all this time. Uh and I really really really really um deeply appreciate each and every one
574. 1:24:28of you. So for those of you who have been mods and and can help with that in any way, great. If you can't and and
575. 1:24:35can't figure it out, that's fine, too. Um I hope you keep coming back and and keep engaging. So with that, Kelly Camp
576. 1:24:43big success tonight. Yep. O Lama makes it really easy. If you're new, we got you. That sentiment by Lori Blair right there.
577. 1:24:52I would say if there's a thing I I try not to I try not to get prideful about things, but one of the things that I'm
578. 1:25:02really proud of in the AI salon is that it really is a welcoming, generous community. And so if you come in here
579. 1:25:11without an absolute clue about what to do with AI, this is a community that's like, "We got you. We'll we'll point you
580. 1:25:18to the right right area." So if that's you, if you're just like, "I'm going to lurk. I don't want to say anything."
581. 1:25:25This is a community to say something. Connect. Share what you're learning. Kelly Anderson, what's up, new people? Exactly. Um, yeah, this is
582. 1:25:39it's a it's a really really special community. So, if you're new here, um, what you're probably not going to get
583. 1:25:48here is deep, in-depth tutorials and prompt guides and things like that. There may be some of that floating
584. 1:25:55around in here, but what you're really going to get in here is people that actually give a damn and give a damn
585. 1:26:00about you and making sure that you're okay and you're finding your way. And and here's the deal. We're all finding
586. 1:26:07our way. I have never felt more clueless in my life as I do right now, this instant. As as the as this stuff gets
587. 1:26:15more and more and more complicated and more and more and more surreal, I think it's important that we reconnect
588. 1:26:22with one another, that that we actively go out of our way to keep connected. Just in the last week, I've had several
589. 1:26:30aha moments that showed me how much I actually know. Yeah, it was cool. Well, that's the thing is we we all we we all,
590. 1:26:38as clueless as we feel, we've been doing this for a while. Well, if you've been in the salon for a while and you just
591. 1:26:43hang out here, part of the design of the AI learning lab is not to actually learn anything. It's to be in the
592. 1:26:50conversation. And so, by being in the conversation, what you don't realize is how much you're actually learning. And
593. 1:26:57so, what happens is you get in a conversation with someone who's not been thinking about this at all, and you
594. 1:27:02realize, oh, I actually know what's coming. I actually know what's here. Uh, so anyway, okay, Kelly Camp, I feel more
595. 1:27:10more lost than ever. To the clueless, [laughter] to the three and a half year clueless. Clue the clueless unite. Um, no, we we
596. 1:27:22we do know more than we think we do. Um, and you know, there there's a lot changing really really quickly. Mary
597. 1:27:30Chester, I can attest to the wonderful people in this in in this community. Beautiful. Beautiful. Um, let's see. I've learned a lot here.
598. 1:27:40Beautiful. Thank you. That's really awesome. So good. So good. So so so so good. All right. Happy Monday night. See you tomorrow
599. 1:27:52night at AI Salon Presents. Thank you for everything and I will see you back here in the salon Wednesday. So 710 out
600. 1:28:02in the world in the socials. So, if you want to see the pre-show, that'll be out there. I'll figure out what the pre-show
601. 1:28:07is. Like, I don't have a ton of thoughts on the pre-show format, I'll figure it out at some point, but but the whole
602. 1:28:13idea is the pre-show will happen from 7:10 to 7:30 Mountain, and then we'll be in here 7:30 on Wednesday doing some
603. 1:28:20sort of demo, some sort of tool. I'll figure out what it is. If you have any requests for Wednesday, let me know and
604. 1:28:28maybe I'll uh I'll [laughter] we'll take requests from the from the hotline. All right, cool. Beautiful. AI is like golf.
605. 1:28:37My skills seem to come and go. Yeah, exactly. Um days when you want to dive head first and there's days when you
606. 1:28:42just want to watch. Yep. Exactly. Absolutely. That's listen, one of the things I'm learning in the AI salon mastermind practice is that
607. 1:28:55slowing down and checking out as an energy regenerator is actually really powerful because you can come back to the AI with objectivity
608. 1:29:07and actually see what's going on in in a more clear way. All right. Oh, and I don't have to limit my shirts now. I don't I can go with my
609. 1:29:16cool shirts that are not going to distract you cuz the comments are not trying to compete with my patterns. So,
610. 1:29:21I can go full pattern, [laughter] full pattern, full hair. All right, everybody. This was awesome. Thank you so much. Um, please keep
611. 1:29:32participating in the salon in between these moments. Peace out. Uh, I hope you had fun tonight. Bye.