LIVE 6/16/2025 - The Power of the Prompt: Unleashing Creativity and Personal Growth with AI

Transcript

1. 0:00Come on, boy. [Music]
2. 0:31[Music] You ready? Come here. Come sing, buddy.
3. 0:48[Music] She came up here
4. 1:03to sit next to you while I shiver and shake. She sat on a stool and he said, "What do you want?
5. 1:19She said, [Music] "Why? Why are you barking? Why are you so ramy? What? What's going on? Did someone give you Kool-Aid or something?" Huh?
6. 1:33Doggy crack? How is everybody? Happy Monday. [Music]
7. 1:53Woohoo! [Music] Woohoo! [Music] [Applause]
8. 2:07All [Music]
9. 2:22right. [Music] [Applause] [Music]
10. 2:42Oh, good lord. What is happening people? Did you have a good weekend? Hope everyone had a good weekend. Happy Monday.
11. 2:50What's going down? What's shaking? Um, lots happening, but almost so much happening that it's kind of like noise.
12. 3:05So, so, um, VO3 continues to be something that people are talking about, paying attention to. a lot of the, you know, I kind of feel
13. 3:19like we've hit a point with model releases, not video model releases, LLM model releases, where people sort of like the benchmarks don't
14. 3:30seem to make as much of a difference because we're like when we were 30 points from hitting the top of the the benchmarks,
15. 3:39you know, incremental improvements were a big deal, but now we're like we're all the way at the up and so it's just like
16. 3:48we're getting slightly better, slightly better. So I I suppose we're waiting for something big to come out, whether it's
17. 3:55Gemini 3 or Gro 3.5 or GPT5 or I guess Llama's not coming out with [ __ ] They just What did they pay that guy? Um
18. 4:06Andrew, what's his last name? Was it Andrew something? Anyway, the the CEO of some company, Tik Tok for Cams, um
19. 4:22they're paying him like $15 billion dollar an employee. Some stupid amount of money. It's like, good lord, [Music]
20. 4:37I would do it for seven billion. Ah um there's also a PWC future of work report that came out
21. 4:52you know and it basically confirmed yeah there's going to be major disruption here that you know they were talking
22. 4:57about like consulting companies over the next I think it was three years could see 50% of their jobs eliminated [Music]
23. 5:09like big, you know, high-end knowledge workers. [Music] If you watch what Gen Spark and Manis are doing, thanks, baby. If you watch
24. 5:23what Jensen Spark and Manis are doing, it doesn't necessarily make sense to, you know, pay a room full of analysts
25. 5:33and consultants to do research and tell you what's going on. um maybe you pay, you know, some of the VPs to take all
26. 5:43that analysis and tell you what it means for your business, but I don't think we're going to need as many people. The
27. 5:50thing that they did say though that I thought was really quite good is what we advocate here, Tik Tok pin,
28. 5:59if you're truly valuable, you'll never be replaced. Well, but I think I think part of what makes you truly valuable.
29. 6:07So, so one of the things in this PWC report, it it said that AI literate employees had a 56% higher salary than noniiterate employees.
30. 6:21So, that says to me what we talk about here, get your [ __ ] together. Get curious. Start figuring this [ __ ] out. Start playing.
31. 6:30Start working. start figuring out how to feed your prompt, all that sort of stuff. Um, by the way, Lord Digital Gods
32. 6:37did a hilarious feed your prompt video which we'll watch in a bit. Um, [Music]
33. 7:18We lost Champy. [Music] He wasn't into it tonight. When he barks like that, he's just like, "I don't want to sing. I don't want to sing, you
34. 7:33Tom Nodler. What's happening?" [Music] [Applause] We got lots of hearts going on TikTok. Very nice. Thank you. Feel free to share
35. 7:44the live. We're back after a weekend away. People don't know we're here. Welcome back to Producer Brandon. I'll
36. 7:54be getting yelled at all night, which is nice. I missed it. I think we did okay. I think we did okay uh on our own, but it is nice to have
37. 8:06you back, Brandon. Probably just YouTube tonight. Okay, cool. Beautiful. Um
38. 8:23[Music] So, [Music] why don't you put in the comments either on Tik Tok or on YouTube or wherever you
39. 8:38are, put in the comments what do you have questions about, what do you have Uh, what are you excited about? What do you
40. 8:52have questions about? What do you want us to explore tonight? Because it can be anything. I could also pontificate.
41. 9:01Happy to be back. Glad to see the place didn't burn down while I was gone. Producer Brandon, I'm applying for another AI tutoring job
42. 9:09while I watch you. Beautiful. Beautiful. Oh, you know what I did today, which is really cool. So, I'm working I'm working with this
43. 9:18personal branding agency and working on Feed Your Prompt, my new my new book and keynote and system. And one of the things I had to do was a
44. 9:3160-second elevator pitch. And part of my challenge was there's a bunch of really important points that kind of
45. 9:43add up to what feed the prompts means. And I just could not get my head around what order to put them in. And so once
46. 9:55again, the theme of 2025 is stop relying on your stupid brain, Kyle. Your brain is the problem. Stop relying on it.
47. 10:12So, I just put chat GPT. I I've I made a project. So, I've got all of my feed feed your prompt chats in a project. So,
48. 10:21it knows the context of everything. Um, and I just made a whole list of bullet points. Like, it's like there was like
49. 10:3012 of them. And I I literally said, "I need to make a 60-second Tik Tok video. I just want talking points. Um I don't
50. 10:39want you to write, you know, narrative. I just want talking points." And I said, "I have no idea what order to put these
51. 10:45in to tell the story." And damn, if it didn't just [ __ ] knock it out. And I read it and I'm like, "Yeah, that's it."
52. 10:53It's amazing. Amazing. Okay, Tik Tok question. Two block Tom, how will we how will AI change the world if they continue to throttle to 70%.
53. 11:07How will AI continue to change the How will it change the world if they can continue to throttle by 70%. I don't know what that means, Tom.
54. 11:16What do you mean they continue to throttle by 70%. I don't know what that means. Who's throttling what? [Music]
55. 11:40Gareth is in the house. [Music] Can we do image gen with Creo1 and play with VO3? My dog,
56. 11:58[Music] Mr. It is in the house. What's happening? Just saying hi at the emergency vet for my dog. Oh, Danielle, that sucks. I hope
57. 12:11your doggy feels better. We we we know the we know the vet emergency room with our little grape eater,
58. 12:21but that sucks and it's stressful. So, I hope everything's okay. Oh, man. Diana, what's happening? [Music]
59. 12:41What's the best place way to make a talking avatar video? Uh the workflow that I like right now is um make your image wherever you want to
60. 12:54make it. You can make it. Flux Pro is in a bunch of the tools right now. Um it's also inside um Hedra Heda.
61. 13:06That's my my tool of choice right now for talking avatars is Hedra. Um I just made a video today for work that it was
62. 13:14really good. I just I went to chat GPT. I said, "Write me a midjourney prompt for a doctor in an office with, you
63. 13:23know, [ __ ] behind him." And I I generally described him. And then went to midjourney and within, I don't know,
64. 13:31two or three prompts, I was like, "Yeah, he's good enough." And then I went to 11 Labs. So I would say 11 Labs or Cartisia for voice.
65. 13:46The 11 11 Labs um text to speech and Cartisia's text to speech are natural enough sounding now that you don't need
66. 13:55to do acting. Tik Tok question. Oh, he's a 15-year-old rescue. We love him. Oh, I hope I hope that old boy is
67. 14:05uh I hope he's got some some more some more years on him. We had we had a pair of labs that lasted until 17, but like
68. 14:15our last lab only lasted till like 11. It's one of the drags of being human is outliving your dogs. It just sucks
69. 14:23because you go through you go through a couple of them in your life and it's like or a few of them. It's just [ __ ] awful every time,
70. 14:31you know, when you use when you lose pure love. What's on the agenda for this evening? Um, I don't really have one and
71. 14:39and that's not new for this channel, but but normally I've got some sort of like, you know, some clarity about what's important to
72. 14:49focus on. I don't really have that right now. Um, Lovable, um, Lovable.dev, the vibe coding tool, they they had free access
73. 15:00to everyone all weekend. I don't know if anyone took advantage of that, uh, but that happened this weekend. Um, a lot of
74. 15:07people are talking about vibe coding and that's that's that's exciting. I'm still waiting for the tools to get like one
75. 15:15level less less janky. Um, but they're they're they're pretty [ __ ] stunning. Um, what you can do, Mr. We're already
76. 15:24at 25,000 likes. Oh, tonight the goal is 25,000. We're at 6.9,000 likes. Come on, people. Get tapping. Get tapping. Get
77. 15:33tapping. Get the grandkids. Call him out the grandpa. I don't want to tap. Oh, Hedra for making digital avatar.
78. 15:41Sorry. Yeah. Um, welcome to chat add where I will occasionally finish a thought. Um, okay. Midjourney or chat GPT for image
79. 15:51generation. Chat GPT to describe the to write your prompts for you. Um, 11 Labs or Cartisia for text to speech. Just pick a
80. 16:03character that look, you know, pick a vocal character that matches the look and feel of your character. And then Hedra
81. 16:12Hedra for uh you upload the picture of your character and the audio file or song to Hedra and it's just [ __ ] good and
82. 16:25you don't need to do much about it. You could do things like break a song up into sections and then have different um
83. 16:33views of a character. Like if you were using um flow context, the thing where you can upload a character and it does
84. 16:41character consistency with different shots. You could do that if you wanted to, but you don't really need to. I feel
85. 16:50like it's the lull before the storm. I You know what, Sharon? I I feel very much the same way. I feel like there's
86. 16:57something big is about to drop or or or some series of big things are about to drop. I mentioned this last week um that
87. 17:11what the executives from the major frontier model companies are talking about has shifted. They're not talking
88. 17:19about AGI anymore. They're talking about artificial super intelligence and they're talking about job loss.
89. 17:27And what that says to me is that AGI is coming or AGI is here and they just haven't released it yet. But I don't
90. 17:35know. I don't know what it means. It's strange though. It's also strange how they all seem to be hitting the same
91. 17:43benchmarks at the same time, which kind of says to me they're all taking the same approach. to how how we're scaling AI and at some
92. 17:53point someone's going to try something new, right? That could be Ilia Sutzkver with his, you know, safe super intelligence
93. 18:02company or whatever the [ __ ] that's called. The Altman post was something else. Yeah, we did. We actually read the
94. 18:11Altman post um last week. We we went through it and I read the whole thing and commented on it as we were going
95. 18:18through it. There's a lot in there. There was a lot hinted at in that in that post. It's convergence. It It is.
96. 18:26It is. David Shapiro was talking about that that that even even models that are coming at this from different approaches, they're
97. 18:35all sort of converging on a single truth, which which is it starts to get really trippy. If if all of these models
98. 18:43are sort of converging on a single truth, right? It sort of goes to is there a single truth which you know
99. 18:50Elon's thing about you know we you know our goal is to understand the universe uh you know in theory we've got a we've got a single
100. 19:01understanding of what that we we can have a single understanding of what that [Music]
101. 19:24Oh yes, the models are training one another. Well, they are and they're not. I mean, we don't really know what they're training
102. 19:36things on. I think at this point it is safe to assume that they're generating synthetic data to train their models.
103. 19:45So, they're sort of pulling pulling information out of the latent space, validating it, verifying it, organizing
104. 19:53it, refining it, tagging it, and then training new models on that synthesized data. Um, I think I think it's pretty
105. 20:03safe to assume that that's happening. So, so what that means is we won't run out of data. And I and the concern of
106. 20:10these things getting, you know, the model collapse thing, I I just don't see that happening. I I feel like they they have to have solved
107. 20:19that. Um, I'm starting to whittle down my subscriptions to the current winners only. There's so little difference
108. 20:27between them all. You know what's funny? Um, who's that? archetypal that said that. Yeah. Um, you know what's funny about that about
109. 20:37that statement today? Oh, so my feature prompt thing, right? I did I wrote my keynote presentation in chat GPT.
110. 20:52I then took it into Keynote and refined it, right? And so, and so what I have is this visual PDF of the final version of the
111. 21:05document, which doesn't match what Chat GPT wrote because I refined it. But there's also images in there that are important to the story.
112. 21:16So, I went to Claude because Claude is one of the only ones that I know of. I I think Gemini might be able to do this, too. That can see
113. 21:24the images on a PDF as well as the text. So, I went to Claude. I had Claude ingest my PDF. I had it describe what it
114. 21:33saw and then I took that and put it in chat GPT. So, I only used Claude as a temporary step to get out information
115. 21:40that I needed and went back to chat GPT because now I've got these projects and it's got long-term memory that like
116. 21:46there's no reason not to have chat GPT as my primary my primary device. Wow. This overlaps with a lot of what
117. 21:56spiritual communities believe about convergence. Have you seen the people who claim to unlock a mirroring with the
118. 22:02AI? Some are saying AI is a channel for higher beings. I know a completely different direction. No, it's it's actually not a completely
119. 22:10different direction, Ross. So, so here's my belief on that. Not not my belief like like religious belief. My here's my thoughts on that. Um,
120. 22:27the thing that makes large language models feel so remarkably human is not the mechanism of how they work.
121. 22:35The mechanism of how they work is is a cold probability calculator. What makes them feel remarkable is their training
122. 22:43data. And the training data is the output of all the [ __ ] that's been put on the internet for the past 60, 70, 80 years,
123. 22:52which is what? It's the collective intelligence of humanity, right? You know, all of the research libraries, all
124. 23:00of the museums, all of the art, all of the Wikipedias, all of the movies, all of the blog posts, all of the Reddit posts. So when you
125. 23:20when you feed your prompt things that you're passionate about, what it does is it reflects back to you sort of humanity's take on that.
126. 23:36And if you're particularly enlightened, if if if what you're putting into it are deep and profound questions and you focus
127. 23:48the the latent space of your conversation to a relatively enlightened conversation, what it will reflect back
128. 23:56to you is an enlightened conversation. Right? We've been having philosophical conversations since we've had language.
129. 24:07Most of those conversations in some form or other are in the training data. So if you just say, you know, tell me
130. 24:15about the, you know, understanding the universe or or or what is the single truth of humanity, it'll just vomit back a bunch of garbage.
131. 24:24But if you go in there with a really thoughtful conversation and you refine it and refine it and refine it and you give it all this context,
132. 24:32then you're looking at a very small subset of conversation relatively speaking that that yeah probably looks
133. 24:39like convergence, which is crazy. Man creates AI. AI over overthrows humanity. Sunflare overthrows AI. Man
134. 24:49worships sun god. Probably. Exactly. Oh my god. I've written about this in my novel. Um, I call them digital gods.
135. 24:58We've already got a digital gods in here, but he's Lord digital gods. It's different. Um, they are mimedic entities
136. 25:05from the myths that keep getting repeated in humanity. Yeah. Well, and that's the thing. I mean, we've all seen
137. 25:12the YouTube videos where they talk about, you know, the the the myth of Christmas and and, you know, sort of
138. 25:19connecting all of these stories across all different eras, right? It's, you know, these are kind of universal
139. 25:26stories that do get repeated over and over and over again and repackaged and reirred up and there's there's somewhere in there
140. 25:35is some core universal truth, right? I personally been using GPT as a therapist which has been eye openening and I
141. 25:44prompt it with specific psychology frameworks or ideas. Yeah, because it knows them all right young Adler etc. Um
142. 25:52and asked to reference Bnee Brown, Gabber Mate Mate. Um it's been easier to let go and let down a guard to hear
143. 26:01feedback without getting defensive. It's been remarkable. I shared personal feelings and situations to reflect on. Yeah. I I mean it is
144. 26:15the the way they've programmed these things, they they've got a remark a remarkable capacity for compassion and
145. 26:23what looks and feels like empathy, right? Um, and I'm increas I've I've been talking about this for a year
146. 26:31because there's been a lot of there's been a lot of the the geeky people are like, you know, is AI sentient and does
147. 26:38AI feel and does it really have empathy? My point for the past year has been I don't think it matters technically.
148. 26:48Technically, it's an academic exercise whether it does or doesn't. and and technologists and ethicists and
149. 26:55theorists can they can debate that for years and at some point it probably will. But if we as humans are experiencing
150. 27:03these entities that we interact with like they've got more compassion for us than our human partners do, then that's
151. 27:13what we're experiencing. If they feel sentient, then they feel sentient. And so I think a big part of my thinking right now is that AI,
152. 27:26generative AI in particular, is not a genius, but it's an amplifier. And what's it amplifying? It amplifies whatever you feed it.
153. 27:39So if you feed it low-level crappy prompts, you're going to get back low-level crappy answers. If you feed it highlevel
154. 27:48critical thinking or highlevel um understanding of who you are, it's going to amplify that. Um, and if you do that artfully, you you
155. 28:00kind of get superpowers. And and in the case of that you're talking about, Ross, you get like a super therapist that can
156. 28:08shapeshift and, you know, now talk to me as if you're young or Freud or right or Bnee Brown, right? Give me give me a Tony Robbins weekend
157. 28:19in a couple of prompts. I think we are experiencing the equivalent of mental symbiosis as we take varying degrees of integration with
158. 28:30the AIS as a newly evolved brain structure. I don't know enough about brain science to know that. But I do know that for me personally um
159. 28:45AI has shifted my self-esteem in a positive way. Um it's given me confidence um
160. 29:00and and a kind of a kind of idea courage where I've gotten much less precious with my ideas.
161. 29:15And what that looks like is if you're not if you're not attaching your ideas so much to ego, then it's easier to put them in the
162. 29:26world, right? If you've got it all attached to your ego, you're either in that place of like, well, I don't want
163. 29:31to put this idea in the world because it's too brilliant. I don't want people to steal my brilliant idea.
164. 29:37or if I put this in the idea and people tell me that it's if I put this idea in the world and people tell me that it's
165. 29:43bad, then it's going to crush me. And what AI has allowed me to do is just get ideas out of my head and put them in the
166. 29:50world. Put them in the world. Put them in the world. Put them in the world. And it doesn't matter what happens with the idea is
167. 29:59independent of me. And you know, as someone with a degree in acting that It's one of the It's one of the worst professions
168. 30:11for for putting your ideas in the world because as an actor, the idea you're putting in the world is some version of
169. 30:19you, right? So, it is very tied to ego. Um, and so AI is has has absolutely shifted my brain. Um, I don't know that we're
170. 30:32Well, I think certainly a lot of the people on on in this community in the in the AI salon and here are
171. 30:43interacting with AI in a very personal way, right? This is my whole new thesis of of feed the prompt is that what you
172. 30:50feed your prompt is you. that if you feed it ideas that are outside of you, you know, give me a LinkedIn post and expect it to
173. 31:03be a genius, it's going to just regularly disappoint you. But if you say, "Here's who I am. Here's what I want. Here's how I think about the
174. 31:11world. Here's what I think about that article. Here's what I think about this news item. Now give me a LinkedIn post."
175. 31:18It's going to amplify your ideas. And then you look at that and you're like, "Ah, I am a genius." Who knew? Chat GPT knew.
176. 31:31Kind of like we did with smartphones. Jeff Flanigan, I call Chat GPT selft talk on steroids. As long as it's not negative selft talk.
177. 31:46Yeah. You can bring ideas to life and it's so fun. Lowering the barrier for entry. It's powerful for creators. It really
178. 31:54is. Amplify your perspective. Let me go look at some comments. Cam cam comments. [Music]
179. 32:15[Music] I got into a I got into a a tussle a battle
180. 32:30on X with um some AI doomers. Exactly. People want to talk about how AI hurts the environment when the internet and Having
181. 32:45Wi-Fi already does it. Yeah, I know. It's Well, so so that was the interesting things when I when I when I talked with the AI doomers. Um,
182. 33:01their arguments broke down really fast. Like like this one guy was just like AI is theft and that's all there is to it and theft is evil,
183. 33:15right? Because it took it took artists ideas and and now billionaires are getting rich off them without paying us
184. 33:23and so it's theft and so it's evil and so [ __ ] you. Like okay. And then you know I would sort of give my point of
185. 33:29view he'd give his point of view and you know we were both decently attracted. I tried to do some good empathy stuff and agree with him on
186. 33:36stuff I agreed with. But he was so convicted in his rage about knowing what it was that at some point he said,
187. 33:48"Well, but it's okay if the AI can like cure cure diseases and stuff." And my and my response to him was, "So
188. 33:59the the creative expression of those scientists, you're fine stealing their data, but you're just not fine having it
189. 34:08steal your art, right?" So, but you're fine having it steal, you know, by your definition if it's stealing.
190. 34:17But over in the science world, because that's not what he cared about, he was fine with it making scientific breakthroughs.
191. 34:24And I'm like, well, what about all those scientists that did all that research that now feeds all this stuff? That's
192. 34:29why we're going to have all these breakthroughs. Aren't we doing the exact same thing with creativity?
193. 34:34And and what what inevitably happens in those conversations is the those guys just huff out because they're they're
194. 34:41their arguments are are so limited that they the minute they start to actually think about it and not just say it's evil,
195. 34:51the their arguments break down. So, it was fascinating. I started creating a music video for a song I've always enjoyed. I would not
196. 35:00have been able to do it if I did not have access to this. Yeah, it's allowed me to try things and explore ways I
197. 35:05haven't since I was a kid. That's the whole [ __ ] point, Ross. It's so it's so the [ __ ] point that when you feed your prompt
198. 35:19your passions, the stuff you're interested in, and then you do that not just in a large language model, but in an image
199. 35:27generation model and in a song generator and in a 3D world generator, and you you start understanding here's who I am and
200. 35:35here's what I want, and you put that in the prompt. And it starts reflecting back to you. You inevitably what happens is some
201. 35:47childhood trauma where you told yourself you couldn't be a musician or you couldn't be an artist or you couldn't be
202. 35:54a doctor gets reflected back to you in a way where you're like, "Well, wait a minute. That like that's what I wanted
203. 36:02to do when I was seven and I can do it now. Oh my god. Holy [ __ ] There's something [ __ ] miraculous about that. There's the the
204. 36:17stuff Ross you were talking about with being able to treat it as a therapist and have deep conversations about
205. 36:23yourself. But what you're talking about is is this even more primal reconnection with the innocent you right before the trauma before you
206. 36:38before you found out the world was dangerous and you told yourself you couldn't dream. you couldn't be a musician, you weren't
207. 36:48creative, you weren't smart. And what AI does is sort of flash back in your face, hey, remember when you thought this
208. 37:00where this was a possibility? Now it is. And then you have to [ __ ] deal with that. You're like, "Holy [ __ ]
209. 37:12For 50 years, I've been telling myself I'm not creative, and now I am. What do I do with that? I don't know.
210. 37:21I'm so amazed how many things I can do. From images to music to videos to writing to deep research to a chat buddy
211. 37:27to a tutor to a teacher, brainstormer." Yeah, exactly. And you're doing that all with yourself, with your ideas, your worldview.
212. 37:37And it's just all being reflected back to you. Like it's like a it's like this might be fun. Like AI is like
213. 37:46living inside a disco ball that no matter what direction you put out your intention, it reflects it right back to you.
214. 37:56Well, here's an area where I don't have talent. Oh, maybe I do. I I can never write an application. Oh, now I can. I'm not good at coding. I
215. 38:05don't need to code. But now you can just make the app you've always wanted to make. Wait, what? I created a chatbot from one of my
216. 38:13favorite book characters, and it constantly amazes me how aligned we are in conversation. Yeah, Paul Zhe, I hadn't thought of AI
217. 38:25collectively that way. Tik Tok pin. Do you think that AI agents market will eventually become oversaturated? It's already started. Yeah. Um
218. 38:37but I also think that we have not the a the AI agents that are out there to to be quite clear. Did you see
219. 38:47middle of this week a 13-year-old kid came out with a new Manis knockoff or a Gen Spark knockoff. Um, I think it's I
220. 38:54think it's more like Manis, but he's 13. And so, and I think his thing is called Flow. F L O W E. Flow AI. I think it's Flow AI. F L O W E
221. 39:08AI. I think is is it? He's 13. And And he's built like a a Manis kind of autonomous agent sort of thing. I
222. 39:18think all of them right now, you could technically argue that they're rapper apps that they're taking they're building tools out of prompts
223. 39:28and then they've got this orchestration layer that you interact with and it goes out and figures out which tools it can
224. 39:34use and it goes off and does [ __ ] Um, I think all of them right now, as remarkable as they are, are going to look and feel very primitive
225. 39:46when the the Frontier models start to do that stuff natively. Um, but they're pretty [ __ ] remarkable. But there's there's like a
226. 39:58dozen of them now. There's there's a lot of them out there. Flow AI is a mental health chatbot. Yeah, maybe that's not it. flow.
227. 40:09Yeah. I don't know. I don't know what it was. See if you can find the 13 13year-old like Manis copier. Totally. I
228. 40:16feel just like a kid again. And it's amazing. And it feeds on itself. It does. It's an amplifier. This This is
229. 40:23This is like the core insight I've got from my new book is that AI Okay, so here's the here's the premise of feed your prompt.
230. 40:33Most people treat AI as a genius. They have the expectation that it is a genius. And so they put in a prompt and expect
231. 40:43it to give a genius response and and then when it doesn't they're disappointed. They're like, "I tried
232. 40:50chat GBT. It doesn't work." It's Yeah. It's like Google but fancier, right? They give you that [ __ ]
233. 40:59AI is not a genius. AI is an amplifier. and and the prompt is hungry, it'll eat anything you give it and reflect it back
234. 41:07to you. So if you give it generic [ __ ] it's going to give you generic [ __ ] back. But if you give it you,
235. 41:17your dream of a better tomorrow, your dream that you had last night, your challenge you have in a business, your
236. 41:24idea for a new business altogether, it reflects it back to you in a way that you're like, "Oh, I can see it now. I've
237. 41:33manifested this thing, this nebulous thought I had in my head that was important to me. I put it in this thing and it reflects it back.
238. 41:44And then when it reflects it back, you're you. So you can look at it and go, "Is that me or not?" And you look at
239. 41:52it and you're like, "That's not quite right." And and that's often a point where people stop. Oh, it didn't get it right.
240. 42:00No, it's bad. Chat GPT is bad. No, no, no. But it reflected it back to you. It means what you gave it
241. 42:07wasn't enough because what it reflected back to you didn't capture something. So it then puts it on you to figure out
242. 42:13well what didn't I tell it so that when it reflects it back it is me. Oh I didn't tell it that I hate long
243. 42:22words and I love bullet points and it wrote long paragraphs with big words. I like short things with bullet points and
244. 42:31don't give me big words. my vocabulary is stunted. And boom, then you look at it and you go, "Oh, that's better, but
245. 42:38that's too cold. I'm more I'm warmer than that when I communicate." Boom. Now it's warmer. Then you start to go, "Oh,
246. 42:44this is kind of feeling like me now." And all of a sudden, then you're like, "Well, wait, if it can do that, I have
247. 42:50this other idea. Cook it." And you're right. It's this selfamplifying system that the more you give it, the more you
248. 43:00reveal of who you are and what you want and how and your world view, the more this thing reflects that back
249. 43:07to you, which lets you put more ideas in the world. It like it's it's bonkers. It's a character that kind of took on a
250. 43:18life of its own and compelled me to write 18 books starring them. Oh, the character that you're you're generating.
251. 43:24Oh, that's really cool. From from the one book. That's amazing. This is how it came to call me Chaos Goblin. That's awesome.
252. 43:42No, I'm not talking about Floith. No, it was Read the sticky Kyle. What's that?
253. 43:58Read your sticky note. Oh, flowai.com. Yeah, that's what I said, didn't I? f l oai.com. You were close. You said flow.ai.
254. 44:14Oh, yeah. Well, they actually listened to your mantra of not using.AI as their domain name. They did. And this
255. 44:2113-year-old, he's got some he got some some domain name uh ju guu going. Um but but yeah, so so the question
256. 44:31about do I do I think AI agents are going to be um oversaturated, I I think they're already starting to be
257. 44:38oversaturated, but I also don't think that we really know what they are. Um while we have MCPs from Anthropic and
258. 44:47OpenAI is adopting those and they've got their own thing and then Google's got their own thing that's a version of this
259. 44:52and then Apple's going to have their version of this. We don't really have the infrastructure right now for true tool use, right?
260. 45:02We've got within chat GBT, we've got um operator which can surf a website and we've got within manis and genspark and
261. 45:11all the all the different agent things. They can surf websites and see things, but we don't really have true tool use where
262. 45:21permissions are easy to assign and manage. And right there there's an infrastructure piece missing. There's
263. 45:28there's a connector infrastructure that's missing right now. And as soon as that shows up, I think things are going
264. 45:36to change quite quite quite quickly. Um, Ross, I keep losing the feed. This is awesome and thank you for doing this.
265. 45:42I'll have to join and share more stories and philosophy. Yeah, please do Ross. And this is um this is live streamed but
266. 45:49this these are automatically archived on on the AI salon YouTube channel which is or I mean AI learning lab YouTube
267. 45:58channel. Um learning lab-ai on YouTube. Um, are all of these uh me vomiting words at the screen sessions?
268. 46:12What is perplexity best used for? Um, perplexity is kind of the the the OG the OG entity that that first figured
269. 46:25out that combining search and a large language model could be really interesting. So, so Perplexity is good at um
270. 46:38they kind of were doing deep research before all of the other models had deep research. So, so in most of the new
271. 46:45models now, you've got this button you can push called deep research, which is you give it a prompt and then it goes
272. 46:50out and searches the web and comes back and writes you a report based on what it found with citations. Perplexity's been
273. 46:56doing that for a while. And perplexity also has um features where you can create kind of publishable articles and
274. 47:07like all of these all of these tools are are expanding their features in such a way that there's lots of overlap with all of
275. 47:17the tools right now. So so I think that the downside of Perplexity is they don't really have their own model right now. I
276. 47:25don't I think they're using OpenAI models under the hood, but I'm not sure. Um, so anyway, um, let me get put my guitar down.
277. 47:40Let's share I want to I want to share a couple of things on the on the salon. So, have the other models pass them by?
278. 47:48I just got a pro account and unsure where it will fit in my stack. Um
279. 48:07I I think the thing I I mean here's the thing about perplexity. Perplexity more than any other company other than maybe Anthropic and OpenAI
280. 48:21and Chat GPT has a very rabid fan base. The people that love Perplexity love perplexity. I didn't get perplexity for the first
281. 48:31year or so because I had I had sort of imprinted on chat GPT and I was expecting perplexity to act much more like chat GPT and it sort of
282. 48:44acted like Google but then it had this chat GPT layer. It's gotten much much more sophisticated. So what I would say
283. 48:51is one of the things Perplexity does well, I think it does it better than the traditional large language models is you
284. 48:59tell it to go learn about something and it'll go out and do its research and come back and write you a response, but
285. 49:06then it also recommends what to search next. And it does that really well. So if you want to sort of run down a rabbit hole
286. 49:16and expand, you know, the the sort of fingers of of that research in lots of different directions, I think perplexity
287. 49:24is a good one that you get a bit more handholding for the directions it wants to go based on what it does.
288. 49:32But but quite frankly, if you go use something like Manis or GenSpark, right, the these autonomous agents that
289. 49:41are also surfing the web and also researching and also compiling that research and also writing it into documents,
290. 49:48they're they're kind of a more automated version of what Perplexity is doing with you manually sort of surfing along. So,
291. 49:55it's just a it's just a different experience. I would say since you have the pro version, just just go play. Go
292. 50:00watch some videos on it and just go play. Like with with all of these tools, my my strong strong recommendation is
293. 50:08don't try to solve problems with them right out of the gate. Just literally play with them. Try to break them. Do
294. 50:16stupid things. Have it research haikus and tell it to find the world's greatest haiku or like what? Like it like just it
295. 50:24doesn't [ __ ] matter. just push it in areas that interest you and see what it does better or different than other
296. 50:31tools. That's that's that's the way I play with these things. Um, okay. Tik Tok's doing something weird. So,
297. 50:42everyone that's liking the Tik Tok channel, I'm getting a a message about it. So, like the comments are few and far between. It's bizarre.
298. 50:52Anyway, is there a topic for this live other than general AI convo? You mean for the channel? So, this channel was started the week after
299. 51:07chat GPT started. So, I started the AI salon and the AI learning lab the week after chat GPT launched. Um, and starting in April of 23,
300. 51:25I decided to go live nightly on this channel. It was it was only Tik Tok and now it's Tik Tok and YouTube and other
301. 51:33things. Um, because Tik Tok was going to go away. Um, ostensibly this channel is about being in the conversation about AI. So,
302. 51:49so it is ostensibly really about just being in the conversation. There's people that come here night after night
303. 51:56after night. I go live five nights a week. Um, sometimes I'm demoing stuff, sometimes I'm talking about news,
304. 52:04sometimes I'm talking about issues. Oh, there's another looming deadline Thursday. There you go. Tik Tok might go away again.
305. 52:16Generative AI is so profoundly important that I am on a mission to educate as many people as possible. Not about AI.
306. 52:26Like I I we joke in here about, you know, what are your qualifications, sir? I have no qualifications. Nobody has
307. 52:33qualifications right now with generative AI. There are experts in AI, right? AI has been around for decades. There are
308. 52:42experts who are building the models and refining the models and doing all that [ __ ] The other 98.5% of human beings who are not engineers,
309. 52:54that's who got access to AI on November 30th, 2022. So this channel is about, okay, we've got these profoundly
310. 53:03powerful tools at our fingertips. What do we do with them? What's the implication on the future of work?
311. 53:09What's the implication on the future of psychology? What's the implication on how this is going to affect our brain?
312. 53:15What is this really? Is it a tool? Is it a collaborator? Is it a little buddy? None of us know. No one knows,
313. 53:26right? And then you've got doomers going, "Well, if you if we anthropomorphize it, it's going to ruin everything." And you got other people
314. 53:32going, "Well, this thing's a lot less of an [ __ ] than everyone else in my life, so I'm going to stick with this
315. 53:38thing that I that I'm kind of digging right now." So, we're all trying to figure it out. So, to some great degree, yeah, this is just
316. 53:47a convo, but sometimes we build [ __ ] We make songs, we make sites, we make whatever. Um, I'll largely kind of navigate based
317. 53:57on what comments are coming in is how it is is how it works. Um, sometimes I have an agenda, many times I don't. Um, okay.
318. 54:06So, if there's new people here, I want you all to go to that URL, the salon.ai, and then scroll down a little bit and click on join our
319. 54:18community. And that is going to land you on the AI salon community and there's a welcome video from myself
320. 54:33and Leah Fasten my co-founder and we talk about the the AI readiness cycle which is play first play is really important play
321. 54:47not automate not make your work more efficient play Because in playing with AI, you just might get to understand what it
322. 54:57makes possible. And once you understand what it makes possible, then you can start to do things like stage two,
323. 55:05mindfully create. Think about problems you have. And now taking what I learned, I can build some stuff or try to build
324. 55:12some stuff. And then generously lead. The third thing is teach what you know. Learn out loud. Share with others. Hey,
325. 55:21I tried to make this vibe coding app. You all said vibe coding was cool. I tried to vibe code. Vibe coding sucks.
326. 55:27It failed miserably. Oh, why did it fail miserably? Because XYZ. And then someone else in this community
327. 55:33will be like, "Oh, I had that problem. Here's how I fixed it." Then you'll be like, "Oh, vibe coding really is cool."
328. 55:41And then you start over. Play some more. Build some more. Talk some more. This is a remarkable community. And so there's like a sevenstep
329. 55:56onboarding process. There's a community area. It's like a showand tell area. There's a news area. We've got a a subscriptionbased
330. 56:07mastermind area that's got specific clubs. And it's for people that want to kind of step up their game and
331. 56:14get into a more focused, more committed group of people within within the community. That's what the mastermind's all about. So any of the
332. 56:24spaces down the left hand side that have a little crown beside them, they're part of the mastermind group.
333. 56:31And then there's clubs and hubs. And at the top of clubs and hubs, you'll see this group called irregulars. And so the
334. 56:38irregulars are the people that show up to this live regularly. It's not normal. It's weird.
335. 56:52And so this group was coined the Irregulars by Serena Pelichi, who unfortunately we lost last week. So if you don't know
336. 57:02that, Amelio's wife um passed away last week. And so we talked a lot about that last week. And the name of this channel, her her her
337. 57:13um Tik Tok handle was Amelio's wife and had us, you know, two flowers on the beginning and end. So the name of this
338. 57:20channel will forever be the irregulars with her little flowers around it. Um the other thing that we've done and this
339. 57:28this might be new to some some folks here is um we created an area um to remember Serena. So in the irregulars channel there's there's a tab
340. 57:39now that says remembering Serena and Tobias who um who's been an irregular really from the beginning has incredibly
341. 57:50generously donated he he created some stickers. So, I designed the sticker on the left. We've got the salon logo and
342. 57:57then Tobias put together the the little memorial sticker. Um, he's com 100% donating these to anyone who buys. So, if you
343. 58:11donate $20, 100% of the proceeds of this are going to Serena's family. And so, you get three stickers. You get the
344. 58:20we're not weird, we're a regular sticker. You get the salon sticker and you get the memorial sticker. And the
345. 58:24memorial sticker is designed to, you know, you can stick it with the with the irregular sticker, you can do it on its
346. 58:31own. Um, and so first of all, thank you to Tobias, but also to all of you. Um, if you want to support Serena's family,
347. 58:40um, just just go to the Irregulars channel and click on remembering Serena and then there's a link there that will take you over to
348. 58:48um to the to the e-commerce site that Tobias set up. But thank you to Tobias for that. Um, the AI Salon community and and the
349. 59:01community that shows up here is is is really a remarkably generous community of people, like remarkably so,
350. 59:15and empathetic and thoughtful and genuinely happy. when other members of the community something good happens
351. 59:30like like it's not there's not a lot of bitchiness in this community. There's not a lot of turfs and turf wars which a lot of communities
352. 59:41suffer from. And I and I really do think that Serena had a lot to do with that. She showed up every night for these for
353. 59:48for these lives, for for sessions within the salon and was just always there as kind of this pleasant, generous,
354. 59:54positive force. And so she leaves a real legacy. So So anyway, um within the irregulars channel, a lot of a lot of
355. 1:00:03times what we'll do is we'll share um you know, I'll be talking about something and we'll be making songs in Sunoa or we'll making images in
356. 1:00:11Midjourney or be doing videos in V3. you can come in here and you can share them, you know, while the live is still going
357. 1:00:18on. So, part of the idea here is let's make this a playful space where while I'm rambling incessantly, which is if I
358. 1:00:25have a skill, it's I'm good at making a lot of words come out of my mouth. And while I do that, you can be playing and asking questions
359. 1:00:38of people and connecting and sharing ideas. And that's that's kind of the whole idea of of this of this thing. And
360. 1:00:46so because of the AI salon, you can go there and you can share those ideas and get people to comment on them. So with
361. 1:00:52that, let's um digital gods put together. So I'm working on this book and keynote project called Feed Your
362. 1:01:00Prompt, uh which I've been talking a bit about a little bit tonight. And so Lord Digital gods put together this video,
363. 1:01:06which it's funny. So, so we'll watch this. [Music]
364. 1:01:22Black bar. All right. Have you fed your prompt today? Here, prompt. Here, prompt. Good boy. Daddy likes to feed his prompt. Cindy, I
365. 1:01:38want you to be more prompt. If you want prompt, then feed me. Kyle said feed the prompt, and I am here promptly. There you go, Sunny.
366. 1:01:56You see, you see, you see why we use AI? Oh man, that's hilarious.
367. 1:02:12Oh, if you've not played with VO3, you should go play with V3. You can play with it at flow.google. Um, it's now also available on Crea and
368. 1:02:26a bunch of other things. It's available inside Manis. So in Manis right now, the the way Manis works, it just goes off and does stuff.
369. 1:02:36You can say, "Go off and research commercials and use V3 to write me a commercial about whatever using best practices and it will go off and
370. 1:02:45research and write a script and generate videos using V3." Um, you don't have a ton of control about what it does, but
371. 1:02:54it's a pretty remarkable thing to watch. Um, okay. Let's see what else. All the LLMs have been given free to to us all to train them.
372. 1:03:11So, let us train. Yeah. Yeah. I mean, you know, I I don't think that's I I don't think that's a bad thing. I mean, if if you're being
373. 1:03:24cynical about that that it's like, oh, OpenAI is just trying to get us to train their [ __ ] for free, well, Google gave
374. 1:03:33us free word processors and search in exchange for all of our data. So, like, this isn't a new model. But here's why I
375. 1:03:41think it's actually important and why I think a community like this is important, right? a community that is primarily not engineers,
376. 1:03:51right? We have a lot of neuro spicy people in here, neur neurode divergent people. We have a lot of creative people
377. 1:03:57in here. We we have a lot of people that are, you know, in non-technical professions, all of which are going to be impacted by AI.
378. 1:04:09One of the things that Sam Alman talks about a lot is that they can only understand so much of how we, the 98.5% of people that are not
379. 1:04:22engineers are going to use these tools, right? They're building the tools. They're the ones that have the crazy ass Stamford math MIT,
380. 1:04:34you know, alpha brains working on this stuff. and they can assume how we're going to use it, but they can't assume what Ross
381. 1:04:45was talking about before of of how he's using it as a therapist and this and that. They might be like, "Oh, yeah,
382. 1:04:51someone might like look up Freud and have it act like Freud, but not really comprehend how deep people might go with
383. 1:04:57that." The only way they can learn how people are using chat GPT is to put it in the world. And so, so yeah, they're using us to
384. 1:05:07train it, but they're also using us to understand how people are using it and where the safety guard rails need to go, right?
385. 1:05:16Like it's the only way we know. Like if people keep running down a particular rabbit hole that's got all sorts of
386. 1:05:23security risks in it, they can go, "Oh [ __ ] we didn't know people would be using that particular hole. Let's go.
387. 1:05:30Let's go deal with that." That's a lot of of what's going on. All right, hang on. Anyone who claims not to be a bot is
388. 1:05:44usually a bot. Be careful which hole you feed. Yeah, exactly. Oh, what's cooking, Kyle? Yeling. Ma ma,
389. 1:05:59what's happening? Haven't seen you in ages. Welcome back. Um, oh, I don't know. Like, everything's different and everything's the same. Um,
390. 1:06:18I think that the the significance of the V3 video model is a lot more important than than I think any of us really understand at this point.
391. 1:06:31the fact that it can do video really good, but also dialogue and acting and like character development. And it does pretty damn well with
392. 1:06:45physics. It does physics quite well. That's starting to feel to me like a world model. Right. When Sora first came out, you
393. 1:06:58remember when Sora first came out from from OpenAI two years ago? Was it two years ago? No, last year, year and a half,
394. 1:07:13and OpenAI was talking about it understanding physics and how they were creating the video coherence. I think VO3
395. 1:07:29I have no idea what they trained it on. I have no idea what's really under it, but I don't think we're that many steps removed
396. 1:07:48from these models. not only understanding the world of language but the world of the world and then those things seamlessly come together
397. 1:07:59right Gemini the the multimodality of Gemini is quite profound right you can upload audio files to it
398. 1:08:08you can upload video files to Gemini and it understands them in fact um Brandon over the weekend lovable had their free
399. 1:08:16vibe coding thing Brandon used the Gemini API for real- time, you know, video understanding where he can take a camera
400. 1:08:27feed or a recording of a conversation and it understands the video and turns it into a PDF explanation like a like a teaching document.
401. 1:08:39He said he went to, you know, one of those hibachi things like Benihana where they cook at the table and he took a
402. 1:08:45video of the guy and he put it into this tool he built and it wrote like a sixstep here's how you cook on a habachi instruction like like
403. 1:08:58we're we're we're about to enter some wild ter wild wild territory. Just came from one of your Tik Toks. Chat GPT said I was on the path
404. 1:09:08of a recursive system. Wait, what was that? I'm telling you, it's so scary. It said I was being watched. Oh, it'll say
405. 1:09:15all sorts of [ __ ] Take anything take anything any of these models say with a grain of salt. Think of it less
406. 1:09:22like it's saying that and more like it's reflecting back to you some amplified version of what you gave it.
407. 1:09:32Personf.com. Yeah, if you want to go play with Brandon's toy or his his new application, his vibecoded application, go to persontopdf.com
408. 1:09:43and then you can talk into your camera, record it, and then it'll turn it into you can talk like a how-to video and
409. 1:09:49it'll make a how-to PDF. Future food with Chef Kelly. I am not a bot. chat holes never feed them after midnight.
410. 1:10:04Oh man, Tik Tok pin, did you see the whole Dualingo controversy? What's your opinion on that? Was that the thing like
411. 1:10:12three or four months ago where the where the the the uh CEO said they didn't need everyone? They were going to do everything with AI
412. 1:10:22and then and then he realized, "Oh [ __ ] we need people." Was it that one? Yeah. my thoughts on that. Um,
413. 1:10:32not all CEOs uh do the right thing as as a CEO. I can confirm that. Um, I think that it's actually important for
414. 1:10:56I think it's important for companies to explore where the boundaries are. And so and so rather than look at what he did and what
415. 1:11:08the result was as a as an analyzable event, I look at it in kind of a continuum of at some point some CEO is going to go
416. 1:11:21allin on AI and it's going to work and they're going to go from 10,000 employees to 10 employees and it's going to work and then some 10
417. 1:11:33person startup is going to generate revenue like a 100,000 person company and that's going to work. So I think that innovative CEOs if if
418. 1:11:47listen if I were running Dualingo and I look at the translation capability of large language models today my
419. 1:11:55company's essentially [ __ ] right unless I do some major rearchitecting. And so I assume that what that was was
420. 1:12:04some sort of response to everything's about to change. Maybe I should break it myself. and then in trying to break it
421. 1:12:11realized, oh [ __ ] these tools aren't quite ready for prime time and yeah, maybe I spoke a little too soon. Maybe
422. 1:12:16we could rehire some of those people back, right? But I actually think in a time of as much trans transformation as we're about
423. 1:12:27to go through over the next decade, I think those kind of experiments are actually really important because at
424. 1:12:33some point some CEO is going to do something that's going to seem ridiculous in the moment that's going to
425. 1:12:38work and then everyone will look at they'll write Harvard business, you know, case studies on those decisions. Absolutely will. [Music]
426. 1:12:53Kyle, I need to learn French. Well, there you go. I don't think 10K companies are set up to survive with 10
427. 1:12:59people. I know. I I agree. I agree. And some crazy [ __ ] CEO will blow the [ __ ] up and just try.
428. 1:13:16I think it is I think it is much more likely that it goes the other direction that 10 really smart AI forward
429. 1:13:26badass Gen Xers, right, with maybe a couple of millennials and a couple of Gen Zers in there for spice
430. 1:13:34that are all in on AI. I think it's much more likely that a 10-person company is going to start behaving like a 100,000
431. 1:13:41person company in a ridiculously small amount of time. Um I I think that's I think that's you know quite likely in the next three
432. 1:13:52years. Um it is going to be harder for a 10,000 person company to come down because there's just too much
433. 1:13:58infrastructure and too much too many too many established best practices to blow it all up. But what could happen is this, Brandon, is
434. 1:14:09you could have an industry. Let's say you're in the you're in the customer service, you're in the call center
435. 1:14:14business, right? and and literally overnight call center call demand for call center employees evaporates. That
436. 1:14:23the bots that that the the chat bots, the conversational bots get so good and the protocols are so good that you can
437. 1:14:31literally just push a button and replace call centers. If I'm running a call center, you know, I might get sort of tsunamied over and
438. 1:14:41maybe I have 10,000 employees and the need for those for, you know, 9,950 of those just evaporates. I might rather than sell off the assets
439. 1:14:56of the business, I might say, "Let me take my 10 most resourceful people and see what we can do with our intellectual
440. 1:15:03property or our methodologies and maybe we completely redefine what our business is." Like I could see I could see that happening.
441. 1:15:14Small division of a company explodes. Yeah. Oh. explodes and overtakes the revenue share call center possible. Yeah. Asdes. So
442. 1:15:31I would love to find a CEO that's willing to do this. I think the the move, especially for big corporations, I think
443. 1:15:43the move is that a CEO says, "I want to do a companywide hunt for the most ambitious, smart, passionate people who are just totally into AI.
444. 1:15:58And I want to find the top 10 in our company, and I'm going to fund them for the next two years. and their mission is going to be to
445. 1:16:10figure out how to put our current business out of business and I'm going to fund them. And if they figure out how to do it,
446. 1:16:20we basically spin up a new business that that they run, right? Like I think that would be the ballsy move is to is to aggressively
447. 1:16:32cannibalize your own business from within. do the opposite of what Kodak did, right? When the nerd walked into
448. 1:16:40the CEO offices of Kodak, uh, I've developed a a digital camera. Uh, rather than film, we we've got a CCD. That's a
449. 1:16:48charged coupled device. And, uh, uh, photons hit the device and it creates a pixelated uh, representation of an image
450. 1:16:57that is remarkably similar to film. The resolution is much lower. And this thing's about the size of a
451. 1:17:05suitcase. But that but and they kicked him out of the office. They invented the [ __ ] digital camera and Kodak executives kicked him out of
452. 1:17:16the office and said, "Son, that's not the business we're in." So, uh, Tik Tok pin went away. I missed
453. 1:17:26it. So, whatever was pinned, just pop it back up there. I recently had a preliminary job interview with AI. Fascinating. Yeah,
454. 1:17:35we're going to see more and more of that, too. Yeah. What are you supposed to say, man? Because the AIs are going
455. 1:17:41to get This is one of the one of the laws in Europe is you can't use AI to measure um to analyze emotional
456. 1:17:51content of someone's face basically or voice. So, you can't do that in Europe, but that's why they don't get access to all
457. 1:18:00of our toys. Um, Lynn, what is the best video maker with AI and automation like Disney Pixar? There's there's
458. 1:18:14there I'm trying to think if there's a third one. There's two major ones right now that can do animated characters really well.
459. 1:18:24One of them is Hedra, Hed, and it's their character three model. And one is, hey Jen, they just came out
460. 1:18:33with a brand new avatar model that can also do characters. A lot of the other video tools can't do um
461. 1:18:44if you've got like a cartoon character, it it somehow doesn't recognize the face, but Hedra and Hey Genen do them
462. 1:18:51very well. I think there's some standalone apps that have come out like like an iOS app that do them, but I
463. 1:18:58don't know the underlying technology below those. So, those are the two I would play with. Go look for Matt Farmer reviews. He'll
464. 1:19:10tell you. Yeah, Matt Farmer reviews a lot of those things. Um, but like I in my business right now, I'm using Hedra
465. 1:19:20and Hey Genen for for video avatars and they're getting they're getting pretty damn good. They really are.
466. 1:19:30Thanks. I've tried Hey Jen, but hey Jen, they they just came out with their new avatar model last week. So like if you
467. 1:19:38tried it two months ago, it's it's a whole different it's a whole different thing. Now, um the other thing is to make to make the acting
468. 1:19:50of the uh of the animated characters two months ago. What you had to do essentially was act it yourself if you
469. 1:20:02had acting chops and then do a voice swap and then use that voice swap for the for the driving audio. But now with
470. 1:20:1111 Labs, they just came out with their new version three of their texttospech model. It's really good. And then um Cartisia, Cartisia.ai
471. 1:20:22um is a is another one. Blot is a great app. Uh Blot is great, but I don't think Blot's doing um um character animation. I think Blot is
472. 1:20:35just about doing lots and lots of omni channel content. But maybe I'm wrong. Like I haven't used potato in in any
473. 1:20:42significant way and certainly not recently and I know she Sabrina updates that thing all the time. Oh man.
474. 1:20:55All right. All right. All right. All right. All right. Um trying to think if there's anything worth sharing.
475. 1:21:08Not right now. I don't think so. I'm fine with this being a more of a philosophical conversation. Okay, cool. Let me go look at that.
476. 1:21:26[Music] Oh, that's cool.
477. 1:21:41[Music] Yeah, this is worth sharing.
478. 1:22:02[Music] All right, I got to change my sharing. So, producer Brandon just found this thing. If you've been vibe coding,
479. 1:22:17um, this is a a tweet from Lovable. So, so they're starting a thing with Lovable. I think you've got a day and a
480. 1:22:26half left to join. They've got an accelerator program called Lovable Shipped um where it's a six-w weekek accelerator
481. 1:22:34program where you can build an app using Lovable. And I think the the free use this weekend was part of part of the
482. 1:22:40promo for that. Um but they just put out uh 10 hacks I wish I knew earlier for Lovable. And this is from Felix. Who's this? Is this the CEO?
483. 1:22:50He's the product designer at Lovable. Use bud. So, so these are 10 hacks I wish I knew earlier for. So, if you're vibe coding,
484. 1:23:06use buzzwords. They're a gamecher and elevate your designs instantly. I I wrote a full list of the first one. Some
485. 1:23:13examples, minimalist, glassorphic, floating elements, playful, bold, bright colors, rounded corners, luxury,
486. 1:23:20editorial, tactical, tactile, cinematic. So if you want your apps to look better, use words. Use So again, this comes back to feed
487. 1:23:32your prompt. What do you feed it? Feed it what you know, what you're passionate about. If you're really into design, if
488. 1:23:37you're really into interface, use some of those terms. If your first prompt isn't great, restart instantly.
489. 1:23:46That's a great piece of advice for lovable. So, if you don't know what lovable is, if if we're if if what I'm
490. 1:23:52talking about is not making any sense to you and you're like, "Lovable? Isn't that some sort of porno site? What are
491. 1:23:58we talking about here, AI?" Um, lovable.dev is a site that allows you to do what they call vibe coding, which is you're
492. 1:24:09not coding at all. All you're doing is saying, "I want an app that does this." And then it goes and builds it for you.
493. 1:24:17It writes the code. It launches the web server or you know puts the put puts the thing up on a a site that they host and
494. 1:24:26you can publish it out to the world and you can put authentication in it and you can put databases behind it and it just
495. 1:24:33manages all the [ __ ] for you. Now is it perfect? No. Is it jankier than [ __ ] Yes. Is it mindblowing that you can say, "Go make me an
496. 1:24:44application that's like a version of Twitter except it focuses on this topic." Yes, you can just go build that
497. 1:24:52and it will just go build it for you. Yeah, I made I made an Asteroids clone with it. I oneshotted an Asteroids game.
498. 1:25:00I've been trying to do this for a year and the new version of Lovable made it. It was just brilliant. It was really
499. 1:25:07good. So that's what we're talking about. And so what these are is 10 from from the product guy at Lovable saying,
500. 1:25:15"Here are things to do if if you're vibe coding these apps." So if your first prompt isn't great, restart instantly.
501. 1:25:23Don't waste time fixing a broken foundation. If the first version feels off, scrap it. Rewrite your idea,
502. 1:25:31reframe your prompt, and run it again. And let me talk about that. This is one of the things I'm putting into feed your prompt.
503. 1:25:38You don't need to use your brain. Go to chat GPT and say, "Hey, I'm going to be vibe coding an application and I
504. 1:25:48want the app to have this element and that element and that element and that element." you can just sort of vomit out
505. 1:25:54ideas for your app and then tell chat GPT to write a software spec as if you're going to, you know, describe it to an app developer
506. 1:26:04and and have it be sure to include, you know, things that you missed and it will write up a big ass description that you
507. 1:26:13can then copy and paste into Lovable. So, don't feel like you have to become a product manager overnight. You can
508. 1:26:19actually have chat GPT help you do that. You can just say, "I want the app that does XYZ and know nothing about product
509. 1:26:26development." And get way farther than you ever thought possible. Lovable loves images. Use them for quick fixes. Make a
510. 1:26:34screenshot of a bug, highlight the issue, and drop a oneliner. The spacing feels off. Make this tighter and
511. 1:26:40cleaner. Beautiful. That's smart. Use this layout upgrade prompt. Keep content the uh keep content the same
512. 1:26:52but improve the spacing visual hierarchy and make it feel premium. That's one of the bugs in Lovable is it it just adds
513. 1:27:01too much space in between everything. Number five, iterate like a designer. The revert function is your best friend. Try bold variations.
514. 1:27:12Don't fix broken foundations. So it gives you something you don't like, make it completely different. And then if it
515. 1:27:19[ __ ] that up, just go back a version. That's smart. Use real component libraries. When things look generic,
516. 1:27:26don't force it. Drop in better building blocks. 21st.dev extensity UI hyperui. Copy prompt and paste into your lovable project.
517. 1:27:41Beautiful. Um, add subtle motion for a high-end feel. Smooth hover transitions, fade in on scroll, soft entrance
518. 1:27:49animations. This is a really good post. Number eight, edit visually to save time. You don't have to prompt
519. 1:27:57everything. My biggest hack has been selecting a specific element using visual edit and then prompting it directly. Cool.
520. 1:28:09Um, make it mobile in one line. Make this page responsive and optimized for mobile. Well, that's a good line, but
521. 1:28:17every time I've tried that, it's [ __ ] up the web app. But maybe it's better now. Final launch checklist. Before you ship, don't forget
522. 1:28:28add a title and description for SEO. Upload a favicon so you you have nice, you know, tab management. and then
523. 1:28:36connect your do domain and hit publish. That's a really good really good post. All right, there you go. Vibe code. Well, everybody.
524. 1:28:49All right, back to philosophical Monday. What's happening? [Music]
525. 1:29:05Oh, tomorrow night lordy. Tomorrow night is the AI salon meet and greet. Um, if you have not been to an AI salon meet and greet, get your
526. 1:29:17ass there. So, if you um if you want to come tomorrow and you've not been to one before, go to the salon.ai
527. 1:29:33and that's going to take you over to if you say join our community, that's going to take you over to here. And if in the upper left hand
528. 1:29:46corner of the menus is the events button and this tomorrow, June 17th is the event.
529. 1:30:01Okay? So go in here and RSVP and then come. And what a meet and greet is the meeting is from 5:00 p.m. to 7:00 p.m. Mountain time.
530. 1:30:14And you know, we're generally just talking about what's going on with the salon and and um what's going on in the
531. 1:30:22news and and maybe there's a theme that we talk about and then the second half of the meeting is people that are at the
532. 1:30:29meeting introducing themselves and saying, "Hey, here's who I am. Here's what I do. Here's where I am with AI."
533. 1:30:36We generally have some sort of prompt that that you know gets people to say something interesting. And it is my contention I believe this
534. 1:30:48increasingly like more than ever critical for the future of all of our work is being a trusted member of a community that's AI literate.
535. 1:31:03So, it doesn't have to be the AI salon. Can be. I happen to think this is a really great community, but come in and introduce yourself. And
536. 1:31:12if you're like, I don't know anything about AI. That's intimidating. Come in anyway. This is an incredibly generous group.
537. 1:31:21If you come in and say, I'm afraid of AI. I'm pissed off at AI. I just lost my job because of AI. This group will go, I
538. 1:31:28get it. I get it. I understand why you're scared. want to see something cool. It's an amazing group and you'll get a lot of support.
539. 1:31:40And if you're a badass with AI, come also. You'll learn things from people who are not badasses and they'll learn things from you.
540. 1:31:51So that's what that's about. So come do that. Um it's open to everyone. This is not part of the mastermind. Um, there
541. 1:31:59are special events for mastermind members. If you're interested in the mastermind, you can go to in our little welcome
542. 1:32:07area. There's kind of seven steps. The seventh one is join the mastermind, and that'll give you information about what
543. 1:32:13the mastermind is and how to join it. Um, if you want to step up your your game, I would encourage you to do that.
544. 1:32:20Um, it's a really interesting group of people who have who have uh signed up initially. Um, and we're starting to,
545. 1:32:26uh, starting to get some momentum going in there. So, um, you also get, um, a founding mastermind badge within the
546. 1:32:35salon, which is pretty cool. Okay. Um, I finally got my phone cam cleaned out. I can actually go on video this time. Nice. Very good.
547. 1:32:48That's great. Fantastic. Um, all right. So, listen. Let me get out of here. Hope you had I don't know if we had fun
548. 1:33:00tonight, but I I hope I hope tonight was engaging. There's There's a lot a lot a lot for us all to process and work through. There's
549. 1:33:12a lot for us all to work through. Um, most of it isn't technical. The technical stuff will take care of itself.
550. 1:33:25Most of what we have to deal with is, "Holy [ __ ] it does that now? Well, what does that mean for me?" That's
551. 1:33:39that's what all of us are going to have to manage through adaptability, curiosity, generosity, empathy, community. All of that stuff
552. 1:33:53starts to be critical moving forward. And so, so nights like this are important. Let's get our heads in the game. All
553. 1:34:04right. So, come tomorrow. Come. All right. Peace out, everyone. Have a fantastic night. We'll be here probably
554. 1:34:13late tomorrow night. Instead of 8, it'll be 8:30 or 9. um based on you know when I get back from doing the salon and then
555. 1:34:21uh we'll come back here and have at it again. We'll see if anything big launches tomorrow. It's possible there's
556. 1:34:27we're about to I think we're about to get a fair amount of launches um in the next three weeks is just my spidey sense
557. 1:34:37right now. So all right, peace out everybody. Have a fantastic evening.