4/9/2026 - Testing the M5 Max MacBook Pro by Running Powerful Open Source AI Models Locally

Transcript

1. 1:32Happy Thursday, everybody.
2. 3:27>> In a westerly direction This car is my strength, is my train. I've been driving, I've been wondering
3. 3:42what it is I'm running from again. Feel like an 80-year-old man holding to the 29. And up ahead on that rise
4. 4:00is a California line. Up ahead truck steering a wide load.
5. 4:14Prefab house cut in half. Cute little daughter and two windows, my lord. Ain't sure that the driver should laugh.
6. 4:29Maybe he's already for the Are you ready for the Perplexity Billion-Dollar Challenge? It's like a billion dollars. What? Am I
7. 4:36supposed to like get through the weekend with that? Like it's barely enough. Every since I have created my Open Claw
8. 4:50Claw Bot named Adam, I have been an independently wealthy economy seven tril- billionaire. You You can make money with Open Claw. Come
9. 5:03get now with the Mac Mini on your just on the desktop. Are you rich, too? You could just pay me
10. 5:18$149.99 in the Click in the link. With the Claw Bot Bot Claw Open Adam. I did Adam.
11. 5:36A la da da da da da da da. Oh, Adam. You perplex me so. Oh, Adam.
12. 5:52Please don't go. Please don't forget to like and subscribe. Like Please Please like and subscribe to the Adam clawbot bot mod
13. 6:02mod claw mod open open claw open the open the clawbot. I It's so easy even I could do it. I made I made an Adam open the clawbot.
14. 6:19He's making for me hundreds of dollars a year. And costing me thousands. So, I'm making money.
15. 7:00She'll be here to sing. Sitting in this lonely town. Wondering if things are going to change. Dream my life away.
16. 7:17Seems these dreams have turned to bust dust clouds. Getting my nerve up. But my past is pulling me down.
17. 7:33Wondering how long this black sheep going to stick around. So, told me once before I said you can
18. 7:48never go home again. What is this? A live for guitar classes? You You too can have an open clock club club club at my club.
19. 8:02Op- open the clock up. With Adam and guitar lessons. I will teach you three chords and a dream and and the three chords and a and an idea.
20. 8:16Make country song hits like I do.
21. 8:30I think some night I should just do a stupid character all night. You could argue that I already do. If
22. 8:40any of you irregular say that, I will come after you. And producer Brandon, you're exempt from the competition. Oh, good lord.
23. 9:25Producer Brandon's like a a comedian. He's like a a posted note comedian.
24. 10:14>> AI summary, a creator is showcasing Adam Claw a money from the Adam Claw bot for making money. Ask him about the features.
25. 10:25Ask me. Ask me about the Adam Claw bot features and I will tell you. But you have to ask. You have to say pretty please.
26. 10:38Thanks Tik Tok. Your AI is awesome. Hey Tik Tok, you know what else would be awesome? If you send some people this way. That'd be cool.
27. 10:54I've sent out trouble to many friends. They are amazed. Such a wonderful song. Oh, thank you very much Mary Mary. That's very sweet.
28. 11:02Did you send out the latest one? To many friends. That's pretty funny.
29. 11:22Where is that latest one? Let's play that. I like that song. Um good news. Well, not really. It's not really good news. It's
30. 11:37interesting news. I got my My MacBook Pro came in 13 days early. So, that's cool.
31. 12:41Hang on. I made a I made a boo-boo. Shh. Calm down, Producer Brandon. I'm a professional. Didn't seem like it by that tab management.
32. 13:20I don't know. Familiar, but I'm not quite sure. Brandon.
33. 13:35>> All right, here we go.
34. 14:08Standing between you and a hot place is insane. Standing too near you and a fire makes it clear.
35. 14:26You're trouble to me. Real trouble, can't you see? Leaning in close.
36. 14:41SMELLING YOUR PERFUME scares me most. Leaning away. It feels stronger every day.
37. 14:56You're trouble to me. Real trouble, can't you see? Sitting by you.
38. 15:10I feel the heat, can you feel it, too? Sitting too far makes me away from you are.
39. 15:25Oh, your trouble to me. Real trouble, can't you see?
40. 15:55It's time to go. I got to leave now. Enjoy the show. It's time for you to say goodbye. I'll miss you, too.
41. 16:13Oh, your trouble to me. Real trouble, can't you see?
42. 16:28Now that you're gone, all this supposed to be more fun. Now that you've left, too, all I feel is this gaping rail.
43. 16:46Oh, your trouble to me. Real trouble, can't you see? OH, TROUBLE, DON'T YOU KNOW?
44. 17:05TROUBLE, WHERE did you go? Oh.
45. 17:30>> Have I written it down musically? Well, it's Yeah, I mean I know the chords to it. I I haven't done that arrangement musically.
46. 17:38Um I haven't broken that out. No. But it's in C sharp. So it's it's tuned down a half step. My guitar was tuned
47. 17:46down a half step when I recorded that. Standing between you and a hotter place is insane.
48. 18:06Standing too near. Hey, yeah. You and fire makes it clear. Oh, no. Why wouldn't someone buy this song from
49. 18:17you? Um because I've never asked them to. Uh they probably might. I just it it it's funny. This audience has heard this song more
50. 18:26than anyone in my life. I wrote this in a songwriting workshop in I don't know. 20 [ __ ] 2011? 2010? Something like that?
51. 18:4520 [ __ ] is not a year.
52. 19:33>> It's funny, the reason I went back to standard tuning is cuz when I was tuned down down half a step it didn't put enough tension on my
53. 19:43guitar. My guitar's all [ __ ] up, so so the strings would buzz. So I I moved it back to traditional tuning
54. 19:51and it it bowed the neck back where it needed to be. Mind your tabs when you move on. Okay. Um I want to do anything in here.
55. 20:04I made it I made a uh a a new Quantum Cipher song. I had So Meta apparently came out with some new model. Actually, let's go look at that
56. 20:13TikTok pin. I've gotten so much wonderful feedback from people I've shared it with. That's amazing. Um
57. 20:20I don't know. If anyone's got any ideas, I think um Techie Becky, who was on here, I think is an agent. I think she's an agent in the music
58. 20:28business and she said something about she thought this song could sell. If anyone knows anyone and wants to make
59. 20:34an intro, I'm happy to talk to someone. I just don't know anything about that world. I do not know anything about the world of
60. 20:44selling songs. I know it's brutal, but I mean, hey, listen, we we now have technology where you can take a a basic song and and
61. 20:52imagine it more fully. If you only had an agent who could research it. Oh, okay, that's just cold. So that's Producer Brandon.
62. 21:01Producer Brandon going, "Oh, you got your fancy new machine. Where's your Where's it Why Why we have genius Adam?
63. 21:11I've had it for like 20 minutes. No, it's like an hour. I've had it for like an hour. Genius Adam. Genius is a stretch. Yeah, I listen, I
64. 21:21would be I would be content with moderately intelligent kindergartner. Not embarrassingly slow kindergartner.
65. 21:33That That would be I'd be happy with that with Adam. I don't like any of you. Why is Why is there an extra thing here? Hang on a second.
66. 21:50What How do I get rid of this? Sharing to this tab. I don't want to see that bar. It's screwing up the Tik Tok black box
67. 22:04thing. Oh, that's better. All right. Source camp in the house, what's happening?
68. 22:23Hello old irregulars. Mr. IT in the house. Uh let's go look Let's go look at some news. There There's just too much going on to to
69. 22:38to not look at the news.
70. 22:56Florida attorney general suing Open AI. Nice. There's Elon Musk has Kamala Harris, I think. Swell. People using AI for good.
71. 23:06Trillion-dollar data centers, and this is what we're using it for. Say something nice about Ohio.
72. 23:25Brendan, did you just did you did you see my feed? This is my moment. You and Cam Catkin, say something nice.
73. 23:41That's hilarious. Name a job AI can't replace. The future is getting weird fast. Ah! AI is learning to think. Robots Robots
74. 23:52are learning to move. Biotech is pushing back against aging. First time in history, intelligence, labor, and lifespan all are programmable.
75. 23:59All right. None of this is news. None of this is interesting. That's What's this? Breaking. Nectome, the Portland, Oregon tech company
76. 24:10Sam Altman is on the waitlist. Oh, they're uploading brains to the internet. Okay, whatever. Check your DMs. Oh, yeah. Okay.
77. 24:23You had some things in there. You did. You did, as I recall. Oh. AI is cracking the code on animal communication.
78. 24:39Might be the most underrated breakthrough in a generation. I called this 2 years ago. I told you we'd be talking to our dogs.
79. 24:47Not only will I be able to sing with Champ, I'll be able to hear him say, "Uh hey, uh can I get some cheese? Uh hey Dad, uh
80. 25:02I like the smell of other dogs poo. Brandon lives there. That's the nicest thing. Oh, a silver fox. You're so
81. 25:12sweet. I'm thinking about visiting my painting in New Orleans in May. That wouldn't suck, would it? That wouldn't. Beautiful. Love that.
82. 25:26You and Champ can have a podcast. Hey, I'm Kyle Shannon. Uh I'm Champ. I'm hungry. You want some You want some burger? No,
83. 25:39you're not eating that. I don't want it if you're not eating it. Well, you you could just You like burgers, right? Oh, yeah.
84. 25:46I I'm having chicken. I don't want chicken. But I'm I made you a burger. Yeah. I I don't want that. It's I I just
85. 25:56I don't need to have the conversations with him. I know what he's thinking. I know what he's thinking. Uh My eyes are itchy.
86. 26:05I think there's allergies and I'm whiny. Uh Okay. So, that was that. Holy [ __ ] Woo. Good lord. I think I need a coffee. I'm
87. 26:20going to go get a coffee. Um Google researchers just discovered that AI models are developing multiple personalities inside their own minds.
88. 26:28Being that they're trained on all of our personalities doesn't surprise me. And those personalities argue with each
89. 26:34other to solve problems. I actually think that's kind of interesting. But here's the point. No one programmed this. It emerged on its own.
90. 26:41Wow. There you go. That's crazy. All right, what's this? Project Glass Wing.
91. 26:55All right, that one we talked about yesterday. YouTube comment future of food. I look like a crazy person at Walgreens
92. 27:05laughing so hard. Sorry about that. What we laughing at? What was I saying? Studying black garlic, lost track of time.
93. 27:18You are the weirdest [ __ ] community. Um by the way, you weirdos, all you are regulars, check your regulars tomorrow.
94. 27:26We're going to do a special experiment with ye old stream tomorrow night. It's going to be fun. So just keep paying attention to uh your
95. 27:33regulars. Brandon and I are going to figure out what we're going to do. Uh-oh, my messages failed to load. Hang on a sec.
96. 27:53Zuckerberg described the described the death of human connection on the internet and no one flinched. Whatever.
97. 28:00I just I don't Zuck always seems like he's caught off guard that he's got to answer questions. He's like,
98. 28:16Okay, cool. I'll check that out. Um introducing new early preview of agents.opencomputer.dev. Don't know what that is.
99. 28:54>> My good buddy John Knack, um he used to work for Photoshop, then he worked for Google, then he worked for for Adobe.
100. 29:06Think it was still Photoshop. Then he worked for Microsoft. Now he's back at Google teaching Google how to understand Photoshop.
101. 29:14So, I assume they're going to make Gemini be able to use Photoshop. That's pretty cool. Is Zuckerberg even human? Pitch Kitchen.
102. 29:29to join this. Yo, dude. Hey, what's going on? What's happening, man? Wait, I Alexa, stop. I got It's just here I got to put I got
103. 29:39to put you somewhere else cuz I'm I'm dual streaming on YouTube, so I've got to Hang on. I can do this. I can figure out the
104. 29:46technology here. I'm going to spin you I'm going to spin you around. All right. So, you can see me, they can see you. Uh-oh, except Hang on.
105. 29:59Yo, dude. What's happening? Nice and good to see you. Yeah, man. What's going on? All good things. All good things. Yeah?
106. 30:09I don't know where to start. You know, I think the biggest unlock recently is what I I learned from this idea of uh
107. 30:20provide you know, we all provide value in our jobs to people. >> Mhm. And if and we provide value in a certain period of time.
108. 30:30And what I'm learning is I delivered a proposal uh this morning that I spent 4 hours on. And normally a proposal would be like a PDF doc.
109. 30:39>> Mhm. But instead, I did such a deep dive with Cloud Code. I built out like a nine-page website and the tabs are parts of the
110. 30:48proposal that I walked the client through. And today, they were so blown away by the quality of the intelligence
111. 30:54and the deep research. And they're like, "We need this like now and I don't care how much it costs." Really?
112. 31:00>> And so, I guess what I'm saying is like what I'm learning is the real unlock is the value that we can deliver in the
113. 31:07shorter period of time is what's helping us and helping our customers and helping the world uh because it's applying more
114. 31:16intelligence instead of less. Yeah. >> And that's just a huge unlock that I'm I'm capturing now. That's really cool. I
115. 31:22mean, I would argue there that um a big part of that also was you walking them through what you had built, right?
116. 31:32Like part like if if you had just sent it to them, it's just like another thing to review, right? But like you going,
117. 31:38"Hey, I know I know what you asked for was this and I did that, but because of these tools I could do this and I
118. 31:43thought of you." And so, like what you're communicating there is not just that the tools can do that work, but
119. 31:48that you give a [ __ ] enough Yeah. >> to to do that for them. You know what I mean? It's like like like that's an
120. 31:55unexpected thing. Yeah, so that's Thank you so much. Yeah. >> Illuminating that because that's like the the yin part of the yang that I
121. 32:04didn't even talk about. That's the whole Yeah. >> like trust and the emotional and presence commitment to making something real
122. 32:13>> Yep. and taking responsibility for making that thing real, which is the whole underbelly Yeah. >> of the value. Yeah, cuz cuz what'll
123. 32:22happen at some point very shortly is the novelty of you being able to work that fast will go away. But what what they
124. 32:29won't forget is that you went above and beyond when no one else was, right? Like it's like it's like that's the thing
125. 32:35that'll stick because all of the all of the outputs everyone's now going to be able to do all the outputs. And so why
126. 32:41we're going to hire someone is oh, I remember when remember when Greg did that thing let's get him in here again.
127. 32:46I I that's where I think all of this is going. That's huge. That's huge dude. That's really cool. Yeah, so what about your side? Give me
128. 32:54sort of a recent aha, you know, oh my gosh, holy mother of God moment. Yeah. I mean I think on a on a daily,
129. 33:01right? They're they're crazy. I mean I don't know. I had I had one today. This is a good one. This is a good one. I'll give you two.
130. 33:11Um so I had one today that that when I first started going live on Tik Tok, I I started the channel in in November of 22.
131. 33:19And I did my first live in April of 23. And very shortly after I went live like I've been you know, I've been in this movie enough times
132. 33:28that I knew that for the next 5 years no one knows anything, right? So so I declared very early on no one's an
133. 33:34expert. I'm not an expert. You're not an expert. Anyone that tells you they're an expert, they're full of [ __ ] right?
134. 33:39And then I would spend like an hour or two before each show sort of studying up on whatever the latest tool was so I could be the
135. 33:46expert. So I was like declaring that there are no experts, but I was acting like one. And then in the past year
136. 33:54like I really have not been the expert because the tools have gotten so sophisticated and the the sort of front
137. 34:01edge of this thing is so far out beyond where where I spend any time. And then and then even the tools themselves are
138. 34:09not just single tools anymore. Everything has 28 features in it. So so I truly feel like not an expert. And
139. 34:16what I've come to realize is I don't like that feeling. So and I think it's something we're going to all have to confront that at
140. 34:23some point there's this there's this level of our human achievement that we just have to accept that these machines are somewhere beyond it.
141. 34:34And I so I don't know. So that was one is just you know, I I know in my head that no one's an expert, but I I'm not
142. 34:42that comfortable not being an expert. So that's the first one. And so what does that really mean? What
143. 34:51does that really come down to? Well, I think it comes down to um just being willing to sit in the in the discomfort of
144. 35:04we don't know what's next. And and and what what it really does for me Greg is it take it it go it goes back
145. 35:11to stuff that's very kind of existential like the the the most important thing is not the tools, right? The most important
146. 35:19thing is me and who I am and what my values are and how I want to be in the world. Like how you were with that
147. 35:25client is way more important than how you did it or even what you did. And so for me that's the thing that keeps keeps coming back to me.
148. 35:36Yeah. Yeah, so about the tools, you know, they're they're great, but they're frustrating as hell as you know. Like you and I both
149. 35:46sounds like we had a our open claw experience and Yeah. You know, and you know, I'm you know, with Claude Claude code is is what
150. 35:58my go-to is at the moment, but I know in a month or two it'll change and I I kind of don't really care. I'm kind of like I'm just all in
151. 36:06in the messy underbelly. >> Yeah. Yeah. So But I I just think that continues the the the ability to be um
152. 36:18to be whatever level level of comfort you are on just knowing that the thing you're building right now is probably going to be
153. 36:27useless in 3 months. And that And that But But it doesn't mean you shouldn't build it. Like I think you should still And you should
154. 36:34still keep building, right? Because Because there is stuff that you learn along the way. It's just that how you do
155. 36:39it, what the mechanism is, what the tool is, what the output is, all of that gets irrelevant really, really quickly. Like
156. 36:46it Things Things we would build before, the files we would make would last for 2, 3, 10 years. And now they're lasting
157. 36:52for 2, 3, 10 weeks, you know, before before things move on. Yeah. No, and yeah. So, Malto Mike he's the same. He's on Gemini this month. So,
158. 37:03that's That's great. >> That's cool. There's so many rabbit holes. Yeah. So, here's here's Here's the other aha that I had. So,
159. 37:11this is This is You should dig this one. The other aha that I had is almost everyone that I know is hanging a
160. 37:22shingle as a consultant, or they're putting together workshops, or they're spinning up some vibe coding app, and
161. 37:28they're putting that in the world and hoping to sell it. And most people are talking about the thing that they built. Like as their core
162. 37:37marketing, what they're saying is, "Here's the thing I built. The thing I Or here's the workshop I'm giving. And
163. 37:42it's going to solve this problem for you, and it's going to do it with these following bullet points, and it's going
164. 37:47to take 3 days, and it's cost this much money." And very few people are getting any results because no one's paying
165. 37:54attention. Because everyone, when they used to be paying attention, that time is now going to making their own [ __ ] to
166. 38:01go sell, right? So, there's there's There's And I I think this has been happening for a while, but but it's
167. 38:07gotten really acute in the past, I don't know, 3 or 4 months. And I think it's going to get worse and
168. 38:13worse and worse and worse. That talking about the thing that you made is it it As As a way to lead the marketing
169. 38:21conversation is just not going to work. That like that was the big epiphany. And so if talking about the thing is not
170. 38:28going to work, what is going to work? Leading a movement. Well, it it's in that neighborhood. Like if if if I'm going to be selling
171. 38:38um like the uh they just vibe coded this thing for um families to share their recipes on it. Um Brandon vibe coded it with I
172. 38:47don't know a couple of people in here. And and um if you hear Brandon talk about here's this thing we made and here's what why
173. 38:57we made it and here's the idea and there's this big bake sale coming up and we're going to be talking about this
174. 39:03idea thing at the bake sale. All of a sudden that thing that's just a recipe database, right? In the end what they
175. 39:09built is a recipe database. It's got some cool you can upload a video to it. There's cool components to it, but in
176. 39:14the end it's a recipe database. So there's going to be thousands of those. But if you hear Brandon talk about it or
177. 39:20you know, who who who's the who's who are the other people involved with it? Gwen? Yeah, Gwen. You know, if you hear Gwen talk about
178. 39:27it, it's it's her community that she's going to be talking about this with. All of a sudden, you're not sort of
179. 39:33signing up for this recipe thing, you're signing up for that thing that Gwen made that she's really passionate about and
180. 39:38Brandon did his thing. I think there's something about who we are, why we made something, um the stories about the impact of it
181. 39:47that will rise above the noise because that's the only thing that can't be replicated. The marketing messages, the
182. 39:53tool itself, everything can be replicated, but what can't is that personal connection to story, to the to
183. 40:00the human beings behind it. And so I think marketing and and people have talked for a long time about personal
184. 40:05brand. This is kind of that, but not really. It's not purely personal brand. It's about saying, what's the emotional hook
185. 40:13to whatever that thing is. Simon Sinek talks about lead with why. it's kind of that stuff. Feels like it's the only way
186. 40:20you're going to rise above the noise in in a marketing environment. This that's my that's >> here. Yeah, L'Engle says why, how, what, which
187. 40:28is the Simon Sinek piece and Yeah. jobs vibes, like the storytellers are going to rule the world. >> Yeah, exactly, exactly. So I think I
188. 40:37think storytelling leads and if you're not storytelling and but the thing is we've got a habit for like decades of
189. 40:44not doing that. Even though these guys have been saying that, I think we're now in a world where that becomes a
190. 40:49requirement because none of the other stuff's going to matter. Yeah. Wow. Yeah. So that's it. Well, thanks for popping
191. 41:00up here. This is kind of cool. Yeah. Great seeing you. >> my wife is out having girls night out on this Thursday and I put the kids to
192. 41:09sleep and Figured you'd hang out on the Tik Tok. I kind of hang out on the Tik Tok and and I you know I like to do these
193. 41:16conversations talking about like I have seen the future ask me anything. Yeah. I'm I'm like deep dive with Ray Kurzweil
194. 41:24and Peter Diamandis and Mhm. I'm I'm I'm trying to and and Kevin Kelly is also a friend of mine. I'd like to say is a friend. I think
195. 41:34he knows me as well and he has taught me a lot about how to predict the future by looking at how the past Mhm. has has
196. 41:44taken shape and how the present is taking shape. >> Yeah. Uh so I feel like uh I have a a good bead on how things
197. 41:53are going to shake out and I like to talk about it with people. Like I know that the job apocalypse is is is
198. 42:00happening but it's more of like the task apocalypse. Mhm. And I think it's uh people businesses want people to take
199. 42:07responsibility and AI is not taking responsibility. Not AI is not writing checks. Yeah. And it's not it's not
200. 42:14coming up with the story. Yeah. So, the people that are doing that, that are taking responsibility, are going to
201. 42:21do very well. And but the people who are just I say uh getting the TPS reports done, uh you know, the from the Office Space
202. 42:31movie, the TPS report. Yeah. Those those soul-sucking jobs are are going. Yeah, that stuff's going. Yeah,
203. 42:37that's I don't know if you've seen Did you see the the great repurpose that I put out? Yeah, also, you know, I love
204. 42:42what you just said about the economies. And let me just say one quick piece about that. Yeah. Like, dude, I'm I'm rebuilding the deck
205. 42:49in front of my house. Not me, but I'm hiring a guy to do it. And and by the way, like, the guy that's doing it
206. 42:57doesn't have any AI skills or whatever. Doesn't need to. He builds beautiful decks and construction. He's in a
207. 43:04economy lane that is very well represented in your chart. >> Economy one, yep. Economy one that we all want we all want
208. 43:13to you know, have in our lives. Right. And I'm very grateful for and should be paid handsomely for that economy. Yeah.
209. 43:21The other guy that delivered the gravel for my deck, he came with his truck and he backed up his truck and he emptied
210. 43:27all the gravel in my driveway and I went out and I shook his hand. I said, "Thank you so much." Um and he said to me, "Yeah, I own the
211. 43:34business." he said. And he said, "How did you hear about the business? How did you hear about it?" And I said, "ChatGPT." And he said,
212. 43:42"What?" And I said, "Yeah, ChatGPT." He said, "I never heard of it." This guy runs a business in my town in New York. So,
213. 43:49like, there's another example of the economy that's completely different. Completely. Yeah, exactly. And you know,
214. 43:56and I think that that's going to go for a while. And as these tools get more and more powerful, it's just going to get weirder
215. 44:02and weirder and weirder and so. >> Weirder and weirder. Yep, exactly. How's your your play your Broadway play that AI love story?
216. 44:12>> Yeah, um Sydney. Um right now it's being pitched to a European theme park actually. So So that's pretty cool. So if and if and
217. 44:22if it's If it ends up there, it'll be it'll become a much more interactive thing and it'll probably exist in, you
218. 44:28know, multiple venues around the park and things like that. We're talking about doing it like that. I love the idea. I was talking to Grock
219. 44:34about it. And Grock knew everything about Sydney. And Grock was saying that that it it would it's a collaboration, a
220. 44:41co-intelligent performance with live performance that creates something new each time based on audience input and
221. 44:49artificial intelligence collaboration. It just sounds like, "Wow, I'd love to see this." Like I want to I want to
222. 44:55witness this experience. So That's cool. I love that. Thank you. Yeah, Sydney an artificial love story coming to a theater near you. I hope.
223. 45:05Yeah. Yeah, cool. All right, man. Groovy. Great seeing you. Great seeing you, too. Okay, cool. We'll talk later. Bye. >> Talk later.
224. 45:18Yeah, so I don't know how to say goodbye to you. Are you going to sign out or I don't know. You don't know either? No, I'll figure it out.
225. 45:28Let's >> I don't know maybe you've done this before. Multiguest. Here. Uh no, you're not there. Let's see.
226. 45:38>> Let's see. It's like old guys old guys trying trying to Hey, boomer. Hey, boom. Hey. You know how to drop the line? Let's see.
227. 45:47>> Multiguest. If I touch that Let's see. Switch kitchen. Huh. I could unfollow you. So what if you just like close the app?
228. 46:02Uh probably. But you're Are you streaming to other Yeah, but I'm I'm on I'm on YouTube. Hang on. I'm streaming YouTube now. Okay.
229. 46:13There's a way There's a way to do this. Hang on. Da da da. Remove background noise.
230. 46:28[ __ ] camera super fan. So when you press on Yeah, multi-guest, do you see yourself there? It says zero guests.
231. 46:38Yeah, it says zero guests for me, too. So strange. >> quit you. Okay, I'll just I'll just
232. 46:53I'll close the stream and relaunch it. So So everyone that's here, if you want to come back, come back. But Greg, good to see you, man.
233. 47:00Good to see you, dude. All right, later. That's really funny. All right. When in doubt, just bail.
234. 47:16Da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da
235. 47:18da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da
236. 47:20da da da da da da da da da da da da da da da da da da. Shouldn't have a day da da da da da da da da da da da da da
237. 47:26da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da da
238. 47:28da da da da da da da da da da da da da da da da da. Da da da da da da da da da da da da da da da da da da da da da da da da da da
239. 47:31da da da da Da da da da da da. I don't have dual cameras. Where are my dual cameras? Oh, there we go. Mm mm mm. Not that I need them.
240. 47:46I got schmutz on the lens. All right, that was cool. Beautiful. Greg Rosna, Pitch Kitchen is his TikTok
241. 47:56channel. Yeah, he's been going live a lot. He's been putting out a [ __ ] ton of content. Probably someone worth following.
242. 48:02Thinking critically about this. Greg was a guy that um ages ago I told this story on here. He and I were talking and he was like,
243. 48:15"Yeah, what are you doing with all this AI stuff?" And and he said um he said, "Oh, you know, I'm doing I'm
244. 48:23doing AI stuff. I'm using ChatGPT whatever." And then I got on a call with him. And I forget what I showed him. I think
245. 48:31I showed him like Code Interpreter where it could it could do spreadsheets or I showed him it doing programming or something like that.
246. 48:39And I think he had only used ChatGPT for like, you know, more like a Google kind of thing. And and whatever I showed him, whatever
247. 48:47I demoed to him, when I came back after demoing it, he was he was like, "I had no idea." He goes, "I need to go."
248. 48:56It totally freaked him out. Um but but he's since like really like like gone deep on this stuff. And he's he's
249. 49:04really knowledgeable about it now. It's super cool. Um but yeah, I got to witness, you know, Greg was one of the guys where I got to
250. 49:11witness a true Kevin McCallister moment where his jaw was literally hung open and he's like, "I had no idea." That was a cool moment.
251. 49:21So, anyway. Um Greasy. So, yeah, you should go follow follow the Pitch Kitchen. Um Greg Rosner. Rosner is his last name.
252. 49:33But but his TikTok channel is Pitch Kitchen. I think this was a loyalty test. This is Hey, we're back to 16. We we
253. 49:42were at 18 and I went away and came back and so now we're now we're good. Now we're good. Da da da da da da
254. 49:52dee, deedly dee, deedly dee, deedly dee, deedly dee, deedly dee, deedly dee, deedly dee, deedly dee, deedly dee, deedly dee, deedly dee,
255. 50:03deedly deedly dee, deedly dee, deedly dee, I lost Kyle and ended up on Greg's channel. That was just It was just a ploy for him
256. 50:15to get more followers. That's hilarious. That's great. Uh I've been hearing it in my head all day. I know it
257. 50:25Trouble is a song that does get stuck in your head. Um what else got stuck in my head? Um the the uh the great repurpose song gets stuck in
258. 50:37my head once a month or so for a couple of weeks. Um and then I'm doing my uh my my morning singing about who I am, that song about
259. 50:47who I am. I get lost in that one, too. And then there's the Olympics. I don't get the joke. I forgot what I was talking about.
260. 50:57Sell it, Kyle. What am I selling? Okay. Oh, you know what? [ __ ] this. Okay, here's what we're going to do. Watch. We'll do this right now.
261. 51:13Where are we going to go? I'm going to do this. I'm going to go screen share entire screen.
262. 51:31I'm going to go to Claude. No, I'm going to go to Perplexity. plexity.ai You should go to irregulars. Oh, boy.
263. 51:52Go to irregular. Hey, did you see my new state five stages of AI adoption graphic? I thought that was pretty cool. I'm I'm 4.532.
264. 52:00That's hilarious. I've weathered the storm and enjoyed the ride. I'm back hanging out with Toto right now. That's good. Uh AI Learning Lab
265. 52:09irregulars. What it's like to watch Kyle talk to Pitch Kitchen on YouTube tonight. That's
266. 52:23hilarious. That's pretty good. Oh, my god. Tina, you fat chap chap butt.
267. 52:37Come get some compute. Oh, my god. Live meltdown. This is why we test. Amazing.
268. 52:51Lord Digital God's coming in hot with the uh with the YouTube thumbnails. As always, beautiful. Okay, here's what we're going to do. We're
269. 52:58going to go to Perplexity. Oh, I've got computer. Do I When Why would Let's see. Let's talk to computer. Why not?
270. 53:10GPT-5.4, let's go with Opus 4.6. Powerful model for complex tasks. Okay, so we're in Perplexity computer. And if
271. 53:20you don't know what perplex Perplexity computer is, that's because they don't have good marketing. But Perplexity computer is like Claude
272. 53:28bot, open claw. The live is restricted for some audiences. Violation. Harassment and bullying. What did I say?
273. 53:39There's It's to be a Gen X exception. It was probably you. It was probably my head. What?
274. 53:56Oh, then just because it just saw my head it thought it was a butt or something like that. Whatever. Oh my god, Tik Tok.
275. 54:09Okay, we're going to try to sell this song trouble. Okay. Okay. I probably need to like register it with the copyright office.
276. 54:25Who [ __ ] knows? Okay. Um I have a song I wrote years ago and have been singing it on my
277. 54:44AI Learning Lab live shows and also using Suno to make more highly produced versions
278. 54:59of it. Um people are saying many people are saying many people I have at least four are saying
279. 55:13that it is good and should be bought by someone. It is Americana like Americana
280. 55:31/country ballad and or rock. I want your help researching
281. 55:46um who to pitch it to, how to break through the noise, and
282. 56:01how to not look like a rank amateur even though I am. I'd love to help you navigate the music
283. 56:16industry and put together a real strategy. First, let me check if there's any relevant context from our past conversations.
284. 56:24And then dive into the research, remembering that the user wrote a song performing it on AI Learning Lab Live.
285. 56:33Type to me like I'm a fifth grader. Go all Prince on it. Keep your rights, baby. Right, I'm the artist 4 million honest as AI Learning Lab.
286. 56:45Who should an Oh, this is cool. It's writing a prompt for itself. Warming up. Who should an unknown songwriter pitch
287. 56:50an Americana country ballad rock song to in 2026? Cover these categories with specific names and contact methods.
288. 56:58Music publishers in Nashville and Americana, which publishing companies? This cool. Breaking through as an unknown songwriter.
289. 57:07Research comprehensively. How does an unknown songwriter break through the noise? Research comprehensively. How does an
290. 57:13amateur songwriter present themselves professionally when pitching songs in 2026? The user is tech-savvy content
291. 57:20creator with a YouTube channel, AI Learning Lab, who wrote an Americana song song registration essentials.
292. 57:27Step-by-step copyright registration with US copyright. Registering with a PRO, ASCAP, BMI, SESAC. I think SESAC and ASCAP allow
293. 57:40Suno-generated songs now. Uh uh uh What a world we live in right now. Isn't it crazy? I got four research tracks running in
294. 57:52parallel right now. Oh, that's what what these are. This is cool. Who specifically to pitch it to, how to
295. 57:57break through, how to look professional, the AI angle. Suno copyright issues, industry perception, legal landscape.
296. 58:03I'll compile everything into a comprehensive strategy doc once the research comes back. I don't just want a strategy doc. You're You're
297. 58:13Perplexity computer. I want you to be my [ __ ] agent. My manager. Go out and do the [ __ ] Uh uh uh Yeah. Yeah, see? Yeah, yeah.
298. 58:30See? Yeah, you have a palooka. Uh uh uh uh uh
299. 58:49Oh, you know what we can do? While that's working.
300. 59:06Mary Mary, sorry to harp, I just think it's important. You think what's harp important? Getting the song out there?
301. 59:13Read an article on how to break through, bots that trick the algorithms. Oh, that's pretty cool.
302. 59:28Keep your rights, baby. Well, I guess the trick is how do you get to here's my new machine. Look, I got it in Darth Vader black.
303. 59:40Isn't that cool? This thing's a [ __ ] beast and it's not that big. It's a little 14-in Mac MacBook Pro M5 Max 64 gigs.
304. 59:51It's amazing how small it is. I love that color. Isn't it Isn't it cool? All right, I should have um
305. 1:00:07Llama should be done now or let's see use the new chat. Okay, so this is the Gemma Gemma 4 26 billion parameter model
306. 1:00:19that's loading. Let's we'll see how fast it is. And then I'm also I'm also downloading um the LTX thing.
307. 1:00:32Did it work? Generate background music. No pending generations. Generate. Model cached.
308. 1:00:58All right, I have the the new LTX video generator I downloaded. But anyway, let's see. You can't You can't really see anything.
309. 1:01:08I'm I'm just going to see Are you downloading Llama? I downloaded LM Studio and I downloaded um Gemma the Gemma let's see.
310. 1:01:22Um Okay, hang on a sec. Write me an M&M M and M style rap about
311. 1:01:38quantum mechanics and make sure to match his sophisticated internal
312. 1:01:53rhyme schemes and double entendre's and word play. Okay, so you and I were were looking at
313. 1:02:09this for the first time together. My new toy. Um This better be more than a [ __ ] toy, by the way. Um All right, let's see how fast this
314. 1:02:23is. This is a 20-something billion parameter model. I'm using LM Studio and Cwen right now. But I'm going to move to Llama next, okay.
315. 1:02:33Here's Gemma. It's thinking. Looks like it's pretty fast, actually.
316. 1:02:48Yo, look. I'm microscopic chronic and ironic symbolic topic dropping logic in the pocket of quantum mechanic profit. That's pretty good.
317. 1:02:59This is running locally on this little machine. Look, it did a whole [ __ ] thing and it doesn't suck. See, I'm breaking through the barrier
318. 1:03:08tunneling through matter like particle through a potential well. Watch the physics shatter. I'm Heisenberg with
319. 1:03:14heat. Yeah, the principle is gritty. You can track my position, but you'll never catch the velocity, see? It's uncertain. The momentum is a
320. 1:03:24phantom, a manic, volcanic, organic, titanic tantrum. I'm dodging the measurement. I refuse to be defined in a
321. 1:03:33state of flux. Okay, so here's what we're going to do. Oh, intro. Heavy distorted bass kicks in. Sound of clocks ticking. Oh, this is
322. 1:03:41really good. Okay. Hang on. Everybody calm down. Just calm down. All right? I'm excited, too. Just calm down.
323. 1:03:56Have a cookie. All right. I'm pretty sure he's the one that's excited. Shut up. Okay, I'm going to go over here.
324. 1:04:10I'm going to go to Suno. Suno. I got to tell you, one of the nice things about um the Mac operating system
325. 1:04:22is that because everything's stored and cached in iCloud, um I basically just started up the machine. I said, "I want
326. 1:04:29to set this up as a new machine." I logged into my account, downloaded Chrome. It said, "Do you want to log
327. 1:04:35into Chrome and pull all your [ __ ] out of the cloud?" And I did, and like everything's set up. This already
328. 1:04:40acts like a machine that's been set up for years. [ __ ] amazing. Okay, um sign in to Suno.
329. 1:04:58Something went wrong. What? No. There we go. Okay. Advanced. Lyrics pop right in there. Or not.
330. 1:05:15Copy. Paste. Why are they not pasting? Except all cookies? Is it because of that? Edit.
331. 1:05:29Copy. Paste. They're not pasting. What the [ __ ] is going on? Notes.
332. 1:05:54All right, that's weird. Okay, paste. Okay, Suno is weird. Help, please. Okay, now we're good.
333. 1:06:14Styles. Heavy distorted bass kicks in, the sound of clocks ticking, then suddenly speeding up and warping, a low menacing whi- whisper.
334. 1:06:26Hip hop wrapped chopped vocals. Chill pop deep flow. J-rap. I assume that's Japanese rap. I
335. 1:06:41don't know what that is. Prog alternative rock. I know, you're not looking at me. I'm just doing it down here. I'm going to turn this into
336. 1:06:53a rap song. Let's see. Deep flow Detroit. Fast. Double time.
337. 1:07:09Okay. Here we go. Bang. So, we're going to take my first Oh, you know what's amazing? The The processor must be right here. It's very
338. 1:07:22warm right here. It's It's very toasty. I got I remember Oh, I've got a hilarious story. This is This is amazing tech support story. Um
339. 1:07:37Um There was There was a There was a model of MacBook Pros. No, it was It was It was before MacBook. It was called PowerBook.
340. 1:07:50And And it was famous for overheating. And it got so hot that you couldn't put it on your lap. It would burn you. It
341. 1:07:55would let It would literally burn you. Um And so, mine had gotten to the point that I just couldn't use it. So, I
342. 1:08:01called up Apple tech support. And I said, "Hey, uh you know, I got this computer here and it's it's super hot
343. 1:08:11and I and I can't put it on my lap." And she goes She goes She goes, "Yeah, I'm sorry about that." I said, "But it's a laptop."
344. 1:08:24Without missing a beat. She didn't miss a beat. Her response was, "No, it's a PowerBook." And I was I was so [ __ ] pissed off,
345. 1:08:35but I was so impressed that she pivoted and made me the [ __ ] Oh my god.
346. 1:08:50Oh, this is the This is the machine you got, Side Hustle? I'm amazed by it every day. That's so cool. All right, listen. Here's
347. 1:08:56Heisenberg tick tick warp. Yo, look look Yeah, look,
348. 1:09:11look, yeah, look. I'm I'm microscopic chronic in ironic symbolic topic dropping logic in the pocket of a quantum mechanic
349. 1:09:18prophecy. I'm breaking through the barrier tunneling through the matter like the particle through a potential
350. 1:09:22well. Watch the physics shatter. I'm Heisenberg with the heat. Yeah, the principle is pretty. You could track my
351. 1:09:27position but you'll never catch the velocity. See, it's uncertain. The momentum is a phantom of a manic volcanic organic titanic tantrum. I
352. 1:09:36dodge the measurement. I refuse to be defined. They say that [ __ ] weaving that we fixed variable behind. You try to pin
353. 1:09:41me down. Try to make out my trajectory but my path is probabilistic. Surely a conjecture, see? I'm blurring the dark
354. 1:09:48in unpredictable fog leaving marks on the charts like a neutron in the park. It's the wave particle duality. Reality
355. 1:09:54is casualty. The instability of my mentality is a gravity defying reality. Is it a solid space? Over wave is sound
356. 1:10:01crashing in the sea of superposition where the atoms are smashing. One moment I'm here the next I'm nowhere to be
357. 1:10:06found. Just a ghost in the function lost in the quantum ground. Now let's talk about the box. Let's talk about the cat.
358. 1:10:12Is he breathing? Is he dead? Is it a fact or just a stat? Schrodinger's cat alive and undead superposition of a
359. 1:10:19beast and a ghost released into ubiquitous duality is maintained. The probability of my greatness is all that
360. 1:10:25remains but the second you observe me the wave function collapses. The illusion is fractured. The mental state
361. 1:10:31snaps. And don't get me started on the spooky action at a distance entangled with the game showing zero resistance.
362. 1:10:37Two particles spinning correlated in a trance even light years apart but locked in the same. It's non-local, nonlocal,
363. 1:10:43and totally uncontrollable soul link connected to the rhythm before you even can blink, blink, blink, blink, blink.
364. 1:10:59Um, um. >> [ __ ] insane.
365. 1:11:31[ __ ] crazy. The sound of a clock ticking. Let's see.
366. 1:11:50Detroit hip hop. I'm going to do more options. I'm going to do weirdness up. Song title. Let's see.
367. 1:12:05Um we'll go M5. Mystery.
368. 1:12:26Three years ago I would not have believed this is possible. Yeah, I know, man. It's crazy. The well so Source Camp, like those lyrics
369. 1:12:42The the the quantum cipher What's the easiest place to find that? If I go to photos photos
370. 1:13:05Um
371. 1:13:38So, that those lyrics, it took 2 years of commercial models before it could do that level of of rapping.
372. 1:13:48And then this in an open-source model is like probably more sophisticated than that. I'm microscopic a microscopic chronic
373. 1:13:58and ironic symbolic topic dropping logic in the pocket of quantum mechanic profit. See, I'm breaking through the barrier
374. 1:14:05tunneling through matter like a particle through through a potential well, watch the physics shatter. I'm Heisenberg in the heat. Yeah, the
375. 1:14:13principle is gritty. You can track my position. You'll never catch the velocity, see? It's uncertain. The momentum is a
376. 1:14:21phantom, a manic volcanic organic titanic tantrum. This is [ __ ] insane. This is the Gemma model is the one I used. And I don't The other
377. 1:14:31model that I was downloading was Quinn 35B. 35 billion parameter model.
378. 1:14:46All right, let's go let's go see how our song is. By the way, I think this thing is rendering a video right now, so it's
379. 1:14:52very hot. Yeah, it's generating it's about halfway through generating a video and the whole back of it is is really hot.
380. 1:15:00The electrons, they be flying. All right, here's M5 mystery.
381. 1:15:21Okay, I don't like this though.
382. 1:16:06It's like crazy. All right, let's let's have let's do another thing here. This might be a little slower cuz I'm rendering a video
383. 1:16:13while I'm doing it, but this is now the Qwen 35 model. And let me go get the prompt. I will do do the exact same prompt copy.
384. 1:16:23We'll paste it in here. I named chat Oh, no, we'll paste it in here. All right, so now I'm doing the Qwen
385. 1:16:34Well, the fan came on. It's not happy. Well, it's so hot. I wonder what the battery temp is. Can you Can you look at that? Let's see. Um
386. 1:16:48I'm going to have to I'm going to have to put this thing on an ice bath. System settings. Um let's see. Temp.
387. 1:17:05Temporary temperature. [ __ ] it's hot. All right. It doesn't have any temperature not in
388. 1:17:19the basic things. Okay, here we go. Boom. All right, so it's thinking doing the same thing. It's actually pretty fast.
389. 1:17:28I don't know if you can see it. Brainstorming. I'm getting to watch it's thinking. So, these are Both of these that I got are
390. 1:17:37reasoning models.
391. 1:17:51Da da na na na na na na na na na na na na na na na na na na na. Final polish. Self-correction. Post-drafting check.
392. 1:18:07Beat drop, minor key, ominous piano loop. Yeah, look. They say knowledge is power, but what if the power's a paradox?
393. 1:18:17What if knowing where you are means losing who you are? Uh Check the coordinates. Check the pulse before the collapse.
394. 1:18:25Before you collapse this function, let's go. Oh, that's just like an intro. Verse one. I'm in a state of flux. No fact to fix
395. 1:18:34the faults that I've got. Heisenberg's principle, man, my position's a plot. That's spinning out of control like a particle no roll.
396. 1:18:44Trying to find stable ground. This is cool. Okay. Okay, we got another one. So, we're going to grab it all. This is bonkers.
397. 1:19:01Back to Suno. Pasted in all that. Let me grab this out of here.
398. 1:19:16Hang on, I got to fix the uh parentheses.
399. 1:20:01The bridge.
400. 1:20:24Outro.
401. 1:20:43>> Okay. And down here we're going to paste that.
402. 1:21:03Okay. So wait, we're going to call this one What was that Quinn that I just used? This is going to be called Quinn's Menace. Quinn's
403. 1:21:19Menace. All right, Quinn's Menace. Here we go.
404. 1:21:33I I'm kind of blown away how small this machine is for how powerful it is. Quinn's Menace. All right.
405. 1:21:50No longer driving. I'm guessing I'm just time to catch up the goodbyes. Man, we're playing. We're playing a
406. 1:21:59little something in here. Quinn's Menace. Here we go. Yeah. Look. They say knowledge is power. But what if the power's a paradox?
407. 1:22:14What if knowing where you are means losing who you are? Mhm. Check the coordinates. Check the pulse. Before we collapse this function.
408. 1:22:27Let's go. Let's go. I'm in a state of flux. No factor fix the facts that I got. Heisenberg
409. 1:22:41principle, man, my position is a plot. I'm spinning out of control like a particle with no role. Trying to find a
410. 1:22:47stable ground, but the ground is losing control over me. See, if you try to pin down where I am, then I'm gone like a
411. 1:22:53ghost in the machine. Yeah, my existence is wrong. It's uncertainty the center of the winter. My sincerity is a phantom
412. 1:22:59when you're measuring my momentum in this demon-driven confrontation. I'm trying to keep my head above the water,
413. 1:23:04but I'm dead inside like Schrodinger's cat in a box where the lid stays sealed tight. Am I alive or am I dead? Is the status
414. 1:23:12or the statue? Is it the law of physics? Amazing. Yeah. Look. They say knowledge is power. But what if the power's a paradox?
415. 1:23:29What if knowing where you are means losing who you are? Uh. Check the coordinates. Check the pulse. Before we collapse this function.
416. 1:23:41Let's go. Let's go. Uh. Yeah. I'm in a state of flux. No factor fix the facts that I got. Heisenberg
417. 1:23:55principle, man, my position is a plot. I'm spinning out of control like a particle with no role. Trying to find a
418. 1:24:01stable ground, but the ground is losing control over me. See, if you try to pin down where I am, then I'm gone like a
419. 1:24:06ghost in the machine. Yeah, my existence is wrong. It's uncertainty the center of the winter. My sincerity is a phantom
420. 1:24:12when you're measuring my momentum in this demon-driven confrontation. I'm trying to keep my head above the water,
421. 1:24:17but I'm dead inside like Schrodinger's cat in a box where the lid stays sealed tight. Am I alive or am I dead? Is the
422. 1:24:23status or the statue? Is it
423. 1:25:18>> All right, I'm impressed. [ __ ] Jesus Christ. All right, do we have a generation? Where Where's my Where is it putting my videos?
424. 1:25:30Where is it putting my videos? And it says no pending genera- Oh, no pending genera- video archive. Ah! We have videos.
425. 1:25:45Okay, this is this is LTX LTX uh rendering on the machine locally. You're seeing it with me for the first time. No audio.
426. 1:26:03It's supposed to be a boxer, a dog a a dog boxer. It's It's not good. It's not good. And then And then this one This one is is a is a human.
427. 1:26:19It's supposed to be a dog, like a boxer dog. Well, All right. So, video's not ready yet for local machines. I'll just keep doing
428. 1:26:33that in the cloud. I can get rid of this one now. Delta X. That's not good. Quit LT X generator. All right. I'll
429. 1:26:45We'll wait for video. I am going to There's There's some audio models that I'll play with. I'll see how those are.
430. 1:26:51But, I guess I'm going to I guess I'll put Ollama on here. And then, I'm thinking about instead of doing open
431. 1:26:58claw on this, I'm thinking about doing uh What's that other one called? Get rid of old models you don't use
432. 1:27:07anymore. Yeah, cuz they'll fill up the [ __ ] hard drive. Um
433. 1:27:21All right, good people. Kyle, can you explain how an LLM is run locally? Sure. Okay. So, what happens when when you use Chat GPT
434. 1:27:36is they've got their Chat GPT has their big ass model that's sitting on a bunch of computers in the cloud.
435. 1:27:46And when when when we're typing with When you're typing into Chat GPT, when you hit submit, it's actually taking what you typed and and
436. 1:27:56the rest of the context window, and it sends it it's Well, probably within the session it's caching the context window, but whatever, it
437. 1:28:04doesn't matter. It's taking what you're typing, and it's it's sending that prompt to OpenAI servers. So, there So, big giant server
438. 1:28:14farms that everyone's talking about are doing the inference. So, so off of your computer in in the cloud,
439. 1:28:22these big machines are processing all of your tokens, and then when it when it streams it back to you, that's when you see it typing.
440. 1:28:30So, all the processing was done on the big iron in the cloud, and then it sends those words back to your computer.
441. 1:28:38There's a couple of downsides to that. One is you are beholden to their privacy policy, and if you read the Open A AI
442. 1:28:46privacy policies, they kind of suck. They kind of say, "We're going to work really hard to not [ __ ] things up for you. We might use it,
443. 1:28:55we might not." It's, you know, the their agreements are shitty. Um, and they're probably going to get
444. 1:29:01shittier and and like people that pay a lot of money will will get, you know, good treatment like that, but the rest
445. 1:29:07of us like anything that you send up there is training their models, goes into those models, is potential security risk.
446. 1:29:15When you run a model locally, you you basically download sort of the equivalent of Chat GPT in the browser.
447. 1:29:21So, there's this this one piece of software called LM Studio that's basically just like a a chat window,
448. 1:29:28like a Chat GPT window, that you can download and and select different models, open-source models, locally to
449. 1:29:38your machine. So, you're actually downloading the model itself onto your machine. Now, the models that they use
450. 1:29:44up in the cloud, like Opus 4.6, that thing's so massive, it can't run on on a local desktop machine. It has to run on
451. 1:29:53massive machines. So, all of the stuff that we're playing with on these local on on laptops and things like that are
452. 1:30:00sort of highly compressed smaller models that are not as capable as those big things, but they've been getting better
453. 1:30:06and better and better. So, both of the ones that I used tonight, Gemma 4, I used the 26 billion parameter model, and then Qwen
454. 1:30:17Qwen 3.5, I used the 31 billion parameter model. And they're just basically decent size models designed for desktop machines.
455. 1:30:28Um so, the model itself is sitting on my machine. So, when I say, "Write me an M&M rap lyric," it's not going to
456. 1:30:34anyone. It's just happening like I could be completely disconnected from the internet, and that thing would still work.
457. 1:30:42Like it's [ __ ] bonkers. So, it's like sitting on that computer right there. We used to make this joke at at
458. 1:30:49agency.com. We'd be like, "Yeah, cuz we had a client say this once." He's like, "Hey, I was wondering if you could come
459. 1:30:55over and install the internet on my computer." And we used to always make fun of him cuz it's like you can't
460. 1:31:01install the internet on your computer. It's the whole [ __ ] internet. That's essentially what we have here now.
461. 1:31:07Right? Like Like all of the world's knowledge that was out on the internet has been smashed into this file that you can download to your
462. 1:31:16machine, and now I can sort of chat GPT-like chat with a very, very capable reasoning model where no data ever leaves my machine,
463. 1:31:25and all of that knowledge is sitting on that machine locally. So, if LLMs run locally, is privacy and security a
464. 1:31:32non-issue? Privacy and security is a non-issue. I mean, to the extent that if I leave this machine unsecured, and
465. 1:31:42someone can hack into it from the internet, they could hack into that machine, but I am not sending any prompts anywhere. Anything that's
466. 1:31:49being generated is just happening on that local processor, on that local hard drive. Nothing is going out to the world. It's amazing
467. 1:32:00that that we're at a place where these models are this good. I mean, two for two, those models basically outwrote
468. 1:32:11the main chat GPT model for a long time. And I And I did um I tried to do one of these M&M rap about quantum mechanics things on
469. 1:32:20on chat GPT 5.1 or 5.2 and it sucked. The The five series of of OpenAI is just not good. Four Four Four and 4.0 were really good.
470. 1:32:33Um but the five series isn't good. But Claude's good and Gemini's good. Um but you're but you're sending your
471. 1:32:40data to Google. You're sending your data to Anthropic. You're sending your data to OpenAI. I now don't have to do that.
472. 1:32:48And also, here's the other thing, Source Camp, when I do my little agents and they're going to go out and do all this stuff for me,
473. 1:32:56I don't need to be paying API costs or running into rate limits for those services because all the all the large language model stuff will
474. 1:33:04happen locally. There's a good 4.0 on LM Studio. 5.2 does suck. Thank Really does, doesn't it? I'm waiting for Sam to tweet so I can
475. 1:33:19tell how bad it sucks. Yeah, exactly. I Listen, you know, they're saying 5.5 is going to be their redemption or six or
476. 1:33:27whatever it is. The But here's the thing. Um Anthropic called their model Mythos. OpenAI called their model Spud.
477. 1:33:43I think there's your answer. Yeah. Yeah, this is our This is our new state-of-the-art. Uh that's called soda.
478. 1:33:54It's not soda. I like soda. Soda's tasty. I'll tell you one of my favorites is the orange creamsicle. It's got a
479. 1:34:00little touch of vanilla in it. Anyway, our our new state-of-the-art, that's what that's what soda stands for. Sota stands for. Uh
480. 1:34:10we got a really clever nickname for it. Um one of the lead engineers, his grandma is from Idaho, and she worked on a potato farm.
481. 1:34:19And what they used to refer to the potatoes as was spuds. Fantastic. I thought, "Hey, let's call this multi-trillion dollar model
482. 1:34:31um spud cuz cuz potato, right? For for Jeremy." Terrible naming. What the hell?
483. 1:34:50All right, kids. I'm going to get out of here. We should have spud in the next week or two, by the way. Um we're
484. 1:34:56probably not going to have mythos until I think they said June or July. And you know, it's possible we we don't
485. 1:35:03get it we don't get it at all. I don't know. I don't know, man. I don't I I don't know what to do anymore, but I now have
486. 1:35:12a machine where we can do actual local stuff that's kind of cool. Spud McKenzie, exactly. Hey hey Sam, quick question. Have we
487. 1:35:22named the new model yet? Oh, no, we're trying to come up with some You know what movie I saw the other night on on TNT? No, is TNT still
488. 1:35:31around? Yeah, they're still around. I was watching TNT and and the movie Spuds McKenzie came on. We we could name it spud.
489. 1:35:45Like someone did that. And no one in the room said no. We've invest We've invested $20 billion to build this thing. Maybe we shouldn't
490. 1:35:58call it Spud. Will you play more of local models tomorrow night? Yeah, you know what I think I'll do, Cam, is is I think that
491. 1:36:10what I I'll do is I'll replace my my laptop my I'll replace my pathetic horrific unusable M1 Mac
492. 1:36:26MacBook Pro with my new Darth Vader black 14-in M5 Max MacBook MacBook MacBook Pro Max. It's the max you can get. Um
493. 1:36:41So, yeah, I'll just I'll just stream from that machine and then we can play with local models and it takes them a while to download, so
494. 1:36:48maybe I'll download a couple of more tomorrow. That'd be cool. I feel like I've been doing this alone for a while now. Yeah, we can do that.
495. 1:36:55Does Spud still run on chips? Okay, tomorrow tomorrow's Friday. What do we do Friday? What do we do Friday? We do office
496. 1:37:05hours. So, come to office hours tomorrow. Um Last week got a little intense. Got a little spiritual. Um
497. 1:37:20But, you know, Mythos came out and it was it was scary and weird. So, so come to come to office hours and then
498. 1:37:27tomorrow night is Friday night date night. And here's the deal. We're going to be shifting where we stream this show, the live, inside the
499. 1:37:39AI Salon. So, tomorrow we're going to do our first actual experiment of that. So, we've been doing a little bit of testing in the past week
500. 1:37:48or two. Well, I haven't. I've just been talking. Brandon has been doing work on the on the back end. He's got this thing
501. 1:37:54figured out. So, we're going to do a test tomorrow that um is this will be streaming inside the salon.
502. 1:38:02So, go into the If you haven't been to the salon, go to community.thesalon.ai. Um the way the format is likely going to
503. 1:38:11work when we finally make the change, I'm going to move the show earlier. I think we're going to start the show at 7:30 instead of 8:00
504. 1:38:19or even 7:00 instead of 8:00. And I'm going to do like a 10 or 15-minute pre-show out to the world like
505. 1:38:28on TikTok and YouTube and LinkedIn and Twitter. And then it'll be a countdown to the actual show starting. So, I'll do all
506. 1:38:37the guitar playing and just, you know, doing happy happy horse [ __ ] out in the world and then we'll push
507. 1:38:43people into the salon. And then you're going to be able to see the stream, chat, everything within the salon. I'm
508. 1:38:49really excited about it. All right? So, that's tomorrow. So, don't miss it. Tell your friends. Bring If you know any
509. 1:38:54salon people who haven't been here in a while, tell them to get their ass to the salon. Okay? Let's get this thing fired up again.
510. 1:39:02Here's an insight I had. I'm going to I'm going to take This is This is not quite a 30-minute goodbye, but I'm going
511. 1:39:07to take a few minutes here. I had a revelation today that I I told Greg earlier that I'm like I don't feel I don't like not being the
512. 1:39:20expert. I don't like not having my head around how all this [ __ ] works and what it all is and what it all means.
513. 1:39:27I feel like I I was pretty good at that for a while. And I'm not good at that anymore and I don't like it. And my my instinct
514. 1:39:37is to withdraw. My instinct is oh, don't yeah, don't that that that is to not connect with people. Like I have a hard time asking for help.
515. 1:39:47I sense that there's a lot of people right now even in the AI Salon community that are just like I don't know what to
516. 1:39:53do. I don't know what I'm doing anymore. That are withdrawing. Reach out to people in the salon that
517. 1:40:01you know and tell them to check back in. I think now more than ever is an important time to connect in community.
518. 1:40:09All right, Cam Cathcart and same same thing. So it will only be on the salon not on
519. 1:40:23Tik Tok and YouTube. We don't know yet Source Camp. We don't know yet. That's part of what we're experimenting with is where it
520. 1:40:34actually lives. But the whole the whole idea here part of the shift is I've been putting all of my energy outside of the salon.
521. 1:40:43And then talking about the salon but like all this stuff is happening outside of it. So part of what we're doing is let's
522. 1:40:49take this energy and put it inside the salon. So that you don't have to go to the salon, you'll already be in it. You'll
523. 1:40:55already be experiencing this within it. And it so it'll drive people into the salon and it'll drive participation in
524. 1:41:01the salon if we do it right. Um we don't quite know what what right looks like right now but that's the reasoning for it. All right?
525. 1:41:09I think you're right about this Kyle. About people about people kind of checking out. And it's listen
526. 1:41:15it is not it's not checking out like I'm done with AI. I I really think this is a I really think this is a thing where
527. 1:41:22it's like what am I supposed to do? Like the thing that I thought I had my head around nobody cares about that
528. 1:41:30anymore. It's moved on. They're all doing this open claw thing. I don't even know what the [ __ ] that is.
529. 1:41:35Alex Finn's out there. He just raised a hundred million or some stupid amount of money for his his claw bot his open claw
530. 1:41:43startup to build a company on a chip and make it easy for other people to do this. And meanwhile, you know, I'm back at the
531. 1:41:52ranch like, how do you how do you what's a prompt? Right? And they're they're off doing that thing. And so I think it's it's
532. 1:42:03I think it makes sense that people like just [ __ ] No, I know. No, I don't know. I think we should be in that discomfort
533. 1:42:11together. I think we should reconnect. So pull some people back in. All right? Cool? Will we be able to post in the salon
534. 1:42:21while watching in the salon? Yes, there's a really cool feature. Danielle, I think you're going to love this.
535. 1:42:28There's a cool feature that Brandon discovered called picture in picture where when you click it, it pops the
536. 1:42:34video stream out into a separate window and then you just have the chat window. It's a it's a really cool thing.
537. 1:42:42I've been cooking stuff up whilst building a plane and it feels crazy. Building on a plane? That's really cool. I'm excited
538. 1:42:49to see it in the salon. Yeah, I'm really excited, Joy. I think this could be like a really fun thing. It's we're we'll still be using
539. 1:42:56StreamYard so it'll it'll look it'll have all the same look and feel of StreamYard. You'll still be able to
540. 1:43:01comment in in you know, in the context of the show where I can see it. Like all of your comments show up in StreamYard. It's cool.
541. 1:43:10But they also show up in the in the salon itself. What I don't quite know yet is can I send you off to some place
542. 1:43:16some other place in the salon and that picture in picture window still stays up. If it does, that would be really cool cuz then you could be
543. 1:43:25trolling around the salon and doing other things while that stream is still there. Brandon is just a new tab. Okay, but you could do that.
544. 1:43:38So, you could have you could have two tabs you could have two windows open with the AI Salon in both and in one is
545. 1:43:44the is the stream and then the other one is the salon. So, so something like that. But, that's why we're
546. 1:43:48experimenting. Like, you're you're the irregulars. Like, you're the we want to make sure that your experience is still really good.
547. 1:43:57Right? It's still like what you want it to be. And so, we're taking time to do all this experiment stuff. So, if it's all [ __ ]
548. 1:44:04up, we'll find that out. Like, we tried to do some things using Mighty Networks streaming they they they've got
549. 1:44:11internal streaming within Mighty Networks. It sucked. It sucked bad. It was unusable. So, this thing is different. This is
550. 1:44:20StreamYard streaming into Mighty Networks. So, we think it's going to work really well, but that's why we're experimenting. All right?
551. 1:44:30Beautiful. I think it'll be awesome. Will they ban you for zooming zooming in on your forehead? You all you all look cute. All right,
552. 1:44:40I'm out of here. Peace out. Have a good night. Tomorrow, office hours 11:00 a.m. Uh check in the AI Salon events. You can
553. 1:44:48RSVP for it and the link is there for that thing. Once we get this figured out, I think that we're going to probably do the same
554. 1:44:55thing with office hours, but I'm not sure yet. So, we'll see. Um but you're definitely going to access
555. 1:45:03office hours through the salon. I don't know if it's going to stream in it. That's we haven't decided that yet. All
556. 1:45:09right? Cool. Oh, and by the way, um Perplexity came back with how to pitch your song a complete strategy.
557. 1:45:18This guide covers show the full report. Lyrics human, melody, composition overall you, production arrangement Suno.
558. 1:45:33Sound recording master Suno plus your inputs murky. In October they have they now accept registrations of partially AI generated
559. 1:45:41musical works where the composition has human authorship. This means royalty collection is not blocked by the AI production layer.
560. 1:45:50After the Warner Music Group, Suno revised its ownership language significantly. Paid subscribers have commercial use rights,
561. 1:45:57but Suno clarified they generally are not considered owner of the song. Generally we're not owner of the song,
562. 1:46:07but if it becomes a big hit, we might decide to be. Critically, Suno's own terms of service warns, Suno makes no representation or
563. 1:46:16warranty that you that any copyright will vest in any output. Yeah. The hidden risk. Uploading original
564. 1:46:23lyrics to Suno grants them a perpetual irrevocable license to use those lyrics, including potentially for AI training.
565. 1:46:31All right. Nashville's view on AI produced demos is divided but shifting. A document in 2026 pitch event,
566. 1:46:39multiple AI demos were presented and received positive reactions. The publisher was evaluating the song, not
567. 1:46:45the production. Nashville hit songwriter Tom Shepherd has publicly endorsed Suno for demo creation, calling it a game
568. 1:46:52changer for writers. However, the sync licensing almost universally reject AI generated tracks due to copyright
569. 1:47:00ambiguity. And many Nashville A&R reps still reject AI demos on site. That'll change. For serious pitching, invest in a
570. 1:47:11Nashville human produced demo. 400 bucks for a full country band demo, that's not that bad. Or 150 for guitar vocal version.
571. 1:47:21Nashville's country demo studio explicitly offers an AI to human service. They take your Suno demo as a road map and recreate it
572. 1:47:30with real musicians. That's [ __ ] cool. From simple idea or AI sketch to authentic hit. How it works. Hear the quality of our country
573. 1:47:43songwriter demos for yourself. This is really smart. This is really [ __ ] smart. Ready to send your song?
574. 1:47:55I'll tell you what would be interesting. I'll tell you what would be interesting is would the would the human produced song
575. 1:48:09be as good as the AI produced ones? First time writers to season pros. Tick tock fin. I read one music site is getting 60,000
576. 1:48:24AI submissions a day. Yeah, I bet. I bet. Yeah, I think this is this is a a big ass long shot. All right, cool.
577. 1:48:39Groovy. Well, Perplexity did its job. We'll go we'll go look at what they what they have to say. All right, cool.
578. 1:48:46Um, peace out. Thanks everybody. Hope you had a good night. Bye.