9/24/2025 - AI Learning Lab: Exploring Generative AI, MCP Servers, and Vibe Coding

Transcript

1. 0:00You ready? You ready for some action? Are you ready for some football? [Music]
2. 0:22Hey. [Music] [Music]
3. 0:50Woohoo. [Music] [Applause] [Music]
4. 1:10There's been something, baby, I've been trying to say [Music] for an age and it seems I don't know Now
5. 1:23with a past and a future now surrounding me. Surrender to whatever cheap can be found. [Applause] There's been a little trouble
6. 1:38since you came to my rescue. [Applause] [Music] And if you were like all of the rest, I would have quit you long ago. But I
7. 1:51couldn't do that. [Music]
8. 2:09Good evening, good people. What's happening? What's going on with the hair? I we have hair management issues
9. 2:15tonight. We're just going to deal with it. It's not as bad as you all make it out to be. All right. I understand there's issues.
10. 2:32[Music] Um [Music] we played with Suno V5 last night. That was pretty impressive. We also did a little vibe coding last night.
11. 2:47Um, we played with Google Mixboard. I don't think anything major came out today, so that's good. Um, so we can do whatever you want to do.
12. 2:59Hair's not that bad. Hey, at least I [ __ ] have hair, right? [Music] [Applause] [Music]
13. 3:16Hey champ, what is it about guitars that you like? What is it? Hey, producer Brandon. [Music]
14. 3:30Producer Brandon going. He said I said at least I have hair. Producer Brandon's like, "Hey, sorry about that. didn't even take that in.
15. 3:41[Music] My dad at this age was was way more bald than I am. So [Music]
16. 4:16[Applause] [Music] [Music]
17. 4:33I there was zone. [Applause] [Music] David played at a place of love. Do you don't really care for music to y
18. 4:51goes like this. The fourth, the fifth man falling majorly. Therefore, king composing. [Music] Hallelujah.
19. 5:08Hallelujah. Hallelujah. [Music] Hallelujah. [Music] Al
20. 5:25[Applause] yeah [Applause] [Music] [Applause] you're feeling strong but you need a proof.
21. 5:41You saw her baby on the roof. Beauty and a moonlight overthrew you. She tied you to the kitchen chair. She
22. 5:58smashed your throne and cut your hair. And from your lips she drew. Hallelujah. Hallelujah.
23. 6:15Hallelujah. [Applause] Hallelujah. [Music] Hallelujah. [Music] Oh yeah.
24. 6:30[Music] Whoa. Wow. That dog is a good dog. All right.
25. 6:48Share the live. Boom. There's that. Do the black bar thing. Boom. There's that. And uh welcome everybody. If you're new
26. 7:00here, this is the AI learning lab. We occasionally learn something about AI. [Music]
27. 7:39[Applause] [Music]
28. 7:55[Applause] [Music]
29. 8:13[Applause] [Music] [Applause] Heat. Heat. [Music] [Applause] [Music]
30. 8:35She came on to him like slow moving cold front. [Music] His beer was warmer than a look in her eye.
31. 8:52All right, enough. guitar. Enough musical instruments. What do you want to talk about tonight? What do you want to talk about tonight?
32. 9:05Let's talk about this AI stuff. Who needs suno when we have Kyle? MCP servers. So, here's the deal with
33. 9:14MCP servers. If you don't know what an MCP server is, stands for model context protocol. It was created by a company called Enthropic.
34. 9:25If you know what an API is, it's similar to an API. It basically allows you to point your large language model like chat GPT or
35. 9:36Claude at a website and the MCP part is where you define what actions you can take on that website. So it might be
36. 9:46Google Maps and you could have an MCP server to Google Maps and you could go find [ __ ] on Google Maps through a chat.
37. 9:53So it lets you use tools through your through your chat. Um I haven't played with them for a couple of reasons. One is they're ve it's very
38. 10:05new. You still need to be fairly geeky to set them up. They're not super intuitive. They're not super supported,
39. 10:12although OpenAI now supports them. Um, Enthropic obviously supports them. They created the protocol. Um,
40. 10:22there they are any anything right now where large language models are using tools are replete with security and privacy
41. 10:33challenges. So, I'm just kind of waiting. I'm just kind of waiting on them to get a little more baked and a
42. 10:39little more easy to use and a little more secure. And they will they'll get there whether it's it's MCPS or some
43. 10:46other flavor of them that comes up. Basically, we need in order for large language models to go truly
44. 10:55agentic. You you hear a lot about agents right now. A lot of people are talking about agents, but everyone has their
45. 11:00head up their ass and everyone's calling everything an agent. An agent is is a is a large language model that can reason
46. 11:10and take autonomous action. So it can do things without you babysitting it. And it can use tools. And so using tools might be I go out to
47. 11:21this website and I do some sort of transaction and then I come back and I tell the thing what I learned or what I
48. 11:27did and then it thinks some more and adds that to the reasoning path and goes out and does something else. Those are
49. 11:34coming. We've got little hints of them right now with things like Gen Spark and Manis. Um, but it's very, very early days. So, I
50. 11:44would say unless you're super geeky, don't worry about it right now. It'll come. It'll come soon enough. And quite
51. 11:52frankly, I wouldn't get too super excited about agentic stuff because at some point uh they're going to get good
52. 11:59enough that those are going to be the things that do your job that get a bunch of people fired. So, so cool out on
53. 12:06wanting those things to get really good. Ann Murphy, what's happening? What is happening? Um, we got Ann Murphy in the house. We
54. 12:15got Dr. Jay ion on China sad hustle mimi shaking it up in here source camp of course Tobias what's shaking Tobias
55. 12:25probably calling in from the hot tub with a wine you know why cuz he knows how to live life I'll tell you that right now
56. 12:37I remember the night what was it you you were doing something and you dropped your phone in the hot tub and kicked you Oh man,
57. 12:52Jetet Nick is in the house. What's happening, Jet Setter? Um Sharon Crawford over on YouTube. How y'all
58. 12:58doing? Um so here's the deal. If you're new here, my name is Kyle Shannon. This is the AI Learning Lab. We're just getting started.
59. 13:06If you have questions about AI, if you have opinions or thoughts about it, um if you want me to demo something, I'm
60. 13:12happy to demo something. Um I tend to focus more on generative AI than than you know sort of building AI or doing coding kind of stuff.
61. 13:21Um the reason for that is because my background is in storytelling and self-expression not in in code. Um I do
62. 13:30have a tech company. Um but I'm not a I'm not a coder. I'm more of a product person than a coder. And uh and I the
63. 13:38thing that excites me the most about AI is the fact that non-engineers are now able to harness the power of this thing.
64. 13:46And so that's kind of the focus of this channel. But I welcome engineers in here. It's just if you're an engineer
65. 13:52and you have engineery kind of questions, I'm probably going to be a little disappointing to you. And I have become one with my with my
66. 14:04disappointment. [Laughter] Oh man. All right. So Vickiy's in the house. Hope you're feeling well.
67. 14:20Yeah, actually that's a good point, Brandon. If you are geeky and you want to you want to um connect with someone who's really
68. 14:28deep into the weeds on this stuff, Pton M um is a member of the AI salon. If you join the AI salon at community.thesalon.ai,
69. 14:37if you could pull up that URL, Brandon. Um, that is um, how is Vicki? Vicki is good. Well, Vicky Vicky had um had some some achy joints.
70. 14:53Deachd achy tonight and tired. Yeah, exactly. to to deach your joints, you have to go through a really achy period.
71. 15:05So, I think she's in the achy period, which probably sucks. So, welcome. I'm glad you're here. Um, so, so speaking of Ann Murphy, since
72. 15:15she's in the house, so if you are a woman in tech, if you are someone who is trying to figure this AI stuff out, the
73. 15:23create conference is happening October 11th to October 14th in Salt Lake City, put on by She Leads AI. So, I do the AI
74. 15:32readiness um project podcast with an on Wednesdays and um that happened today. So, I hope you saw it. Uh, if not, shame
75. 15:41on you. We We are not above shaming people here. We'll do it. We will We will cast stones. We will cast slurs.
76. 15:49We'll do it. We We'll go there. Anyway, um go to sheleadsai.ai and check out the create conference and
77. 15:58uh if you have not signed up for it, you should go do so. So, with that, um what was that? Fortune. Oh, cool.
78. 16:09So, breaking here tonight, this is an AI learning lab exclusive. If you go to sheleadsai.ai and you want to go to the create
79. 16:21conference, if you put in the code fortune 500, all caps fortune 500, you get 500 bucks off straight from man Murphy. There's a
80. 16:32deal. All right, so go sign up. Let's get some people there. All right. Fantastic. Fantastic. Fantastic.
81. 16:41I tried to join and couldn't find it. Tried to join what the the uh the create conference. Let's see. Let me jump over there right now.
82. 16:56She leads AI.ai. Maybe you're talking about something else. Yeah, it was right there on the right there on the homepage. It was there.
83. 17:09Oh, no. The podcast. Oh. Oh, the podcast. Yes. So, the podcast is now happening on uh a standalone YouTube channel. So, I
84. 17:21think I need to update and I've been a bad boy. I have not updated our uh calendar invites in the salon. That's
85. 17:30that's my problem. Uh Vicki, that's something I've got to fix. Thank you for the reminder for that. Um okay, Groovy,
86. 17:38um can you tell me some advice in building an AI SAS app? Tools and advice on the process. Well, I need to know a bit more about you.
87. 17:51Let's see what's your first name there. I can't really see it with this. Um, someone someone M38 someone M38. Listen, someone listen,
88. 18:04you're never going to be somebody until you're anybody. I am someone. Who's on first? Okay. Um, so I ass I assume by how you asked the
89. 18:18question, you may not be technical. Um, and if you're thinking of how do I vibe code uh a SAS app uh and get that
90. 18:28out there because everybody's posting on X that they vibe coded a SAS app and now they're making $20,000 MR um overnight. Um that you're
91. 18:39>> you can make money with >> you're uh you're you're in that category of of people wanting to want to do
92. 18:46things. Now so a couple of things. If you really know what you want to do, if you if you know what you want to do,
93. 18:53vibe coding is an option. But there's a couple of things you need to know. If you're new to software development, you
94. 19:00need to understand that in order to have a functional application that you're going to charge money for, you need
95. 19:10a number of different components. You need the application itself. That's a relatively straightforward thing to do.
96. 19:17Then you need a database. You need a web server. You need authentication like people need to be to be able to sign up
97. 19:24for it, right? And then you need uh sort of a transaction piece. You need to be able to charge for it and all of those.
98. 19:32And and then the I guess the fifth one is you need to have it that all set up correctly so that you don't get your [ __ ] hacked.
99. 19:40Um and and what I would say is this. you can vibe code together a prototype of a SAS app very very quickly. Going from
100. 19:53prototype level to production level is is a much heavier lift. Um and and you probably need to learn about some of
101. 20:00those other components. Now what I can tell you is some of the vibe coding platforms are getting better at this.
102. 20:08Like Lovable for example has a deal with um Supabbase for the database component of it and they've got an authentication
103. 20:16thing and they're in beta right now of having their own database backend and their own authentication backend. So
104. 20:25Lovable applications are going to handle a lot of the heavy lifting of that stuff. So it's lovable.dev. If you
105. 20:34haven't vibecoded something, you should go do that. Um, you know, some of the other ones are you can do this in Claude, you can do this
106. 20:46in chatgpt in the canvas, you can do it in Gemini in the canvas, you can go to astudio.google.comapps and that's a whole vibe coding area
107. 21:00there. There are a bunch of different places to do it. Replet's got an agent. Um uh uh uh Cursor is an app that that
108. 21:09is a vibe coding app that integrates with with um software development environments. Um there's a bunch of them out there,
109. 21:18but just make sure that you're going into it with eyes wide open to know that you need those components. And if you
110. 21:25don't have those components bulletproof, you're going to get hacked. Because what's happening is that the software development community is
111. 21:37they're they're not quite as cranky as the creative community is with with things like music generation software
112. 21:45and video generation software, but the the people who've been developers for a long time are decently cranky about
113. 21:53people vibe coding apps together. So, so what they're doing is they're letting people vibe code a SAS platform, launch
114. 22:01it, and then they go exploit it and hack it and do things like that and teach them a lesson. So, just just be careful. Be careful out there.
115. 22:13Lovable to do a 2D game. Absolutely. I Well, let's go. Let's go make a game. Let's go. Uh, black screen, please. I
116. 22:21think I do have the the black screen up. Oh, but you know what I need to do? So, I need to get rid of my bookmarks here.
117. 22:29Always show bookmarks. That helps. That gives us more black. And then I can bring this up like this. There. That's better. Is that better? I
118. 22:38think that's a little better. All right, let's go make a 2D game. Um, [Music] do I want to do it in lovable? Yeah,
119. 22:54let's do it in lovable. Well, I haven't played in Lovable in a while. MRR monthly recurring revenue. Yes, ARR
120. 23:06annual recurring revenue. Yeah, I just saw I just saw an ex post today that some guy was like a thing I completely
121. 23:13vibe coded just hit 20 20,000 MRR, which you know that's not bad. It's 240 grand a year, but you know,
122. 23:22it's also it's 20,000 MRR probably, you know, 6 weeks after he made it. So, is that is he going to grow this? Are
123. 23:30people actually using it? Like there's the the thing about creating a SAS app is that your competition is increasing exponentially and
124. 23:44everyone else's ability to create a SAS app is is going to be easier and easier and easier. So your competition is is
125. 23:51going to be accelerating around you. Um so make sure that if you're going to start something like that, don't don't
126. 23:58go into it thinking like it's just going to be easy. like you've got to like market to customers and understand
127. 24:04things like customer acquisition cost and and churn rate and onboarding and just there's all sorts of all sorts of
128. 24:12[ __ ] you got to contend with if you're doing a SAS app. All right. I'm willing to learn cyber security. Wait, this was
129. 24:18good. Um, who do you recommend me doing to protect it? I Oh. Oh. What do you do recommend me doing to protect it? So, I'll tell
130. 24:31you what you could do is you could just go on X. Actually, actually, you know what? Maybe rather than vibe coding right now, let's let's
131. 24:39hop over to X for a second. Let me share this tab. Um, I'm on X right now and I'm going to go to Grock. So, so one of the nice
132. 24:52things about Grock, if you've not used Grock, is that Grock can tell you about Oh, what's a SAS app? Okay, I'll get to that in a second. Um,
133. 25:07one of the nice things about Grock is that because Elon Musk shut down the API to Twitter, Grock is is the only large language
134. 25:18model that can actually search Twitter, right? And and and consolidate what's going on there. So you can go to Grock and you can say
135. 25:28um I am going to vibe code a SAS app and want to launch it commercially.
136. 25:44Um, I want to better understand the security and configuration issues that
137. 25:59beginner vibe coders are overlooking and I want to get
138. 26:13better at my craft. So, my app is bulletproof. Okay. So, that's going to go off and do that. It's going to go think the request
139. 26:29involves vibe coding a SAS app for commercial launch. Uh, check in. What does SAS stand for? SAS stands for software as a service. And so
140. 26:43the the sort of classic the the classic SAS model is Salesforce, right? So you could create your own database of, you
141. 26:53know, uh customer records, right? Say, you know, sales prospects or you can pay Salesforce a monthly fee
142. 27:02for them to build the application for you to, you know, for your CRM. And so that's so the classic model is you build
143. 27:10a website that people log into and then they pay a monthly fee per seat, right? So per user and then you make buckets of
144. 27:19money because you don't really need that many staff because they're just logging into this piece of software and using
145. 27:24it. But everyone needs a different feature. And like one so um Salesforce is a classic SAS app where it's got so
146. 27:34much feature bloat and so much um sort of overkill on features that it's completely unusable at this point, but
147. 27:42no one can leave it because it's got all their customer data in it, right? Like the the the cost of of transitioning
148. 27:50from Salesforce to some other CR CRM is ridiculous. And so they they're basically just printing money. So the
149. 27:58the ideal thing that you want when you create a SAS app is what's called a sticky SAS app where once someone starts
150. 28:06using it, they can't unuse it. Like chat GPT is a good example of that. Like once you've used a large language model, you
151. 28:13don't want to not use one again, right? So uh not to be confused with SAS, which is what I have exactly. All right. Building
152. 28:24a SAS app using no code or low code. Um I I I've structured this table for clarity focusing on key categories of risk.
153. 28:37Account impersonation and authentication failures. Platforms often default to simple login or use builder credentials
154. 28:44for everything and new devs assume build-in off is good enough without testing edge cases. attackers can hijack
155. 28:52sessions, right? So, here's a big big longass table that Grock just put together. So, so this is good. I mean, what what's nice about Grock
156. 29:05looking at X, what people are talking about on X, X is where all of the the AI and dev communities are sort of
157. 29:14clashing, right? And so, you've got vibe coders going, "Look, ma, look, I made my first app. And then the developers are
158. 29:23going, "Hey, Numnuts, you know, you left the back door open, and by the way, I went in there and got all your, you
159. 29:28know, customers credit card information. You might want to close that back up." So, you have these these conversations going back and forth.
160. 29:36The vibe coding platform community is well aware of all these issues and and there there's a race right now in my opinion for the
161. 29:47vibe coding platforms. There's two basic audience for for AI coding platforms. One is professional engineers, right?
162. 29:55And so things like cursor and codecs and and um and and those sorts of tools are geared at existing engineers doing their
163. 30:05work more efficiently. Then you've got things like lovable um and and now and and the the sort of uh the canvas
164. 30:13features in um Claude and Chatg GBT and now in Gemini are aimed more at think of it as casual developers, vibe coding
165. 30:24developers. So two really distinct audiences. The professionals they know how to do this [ __ ] for the most part. So, so
166. 30:33they're going to do all their configuration, you know, how they do their configuration and they're going to
167. 30:37get that locked down. The casual vibe coders, that's where there's a race right now to make it as simple as
168. 30:44possible to literally just ask for an app, have it build it, and have it automatically set up all of these things
169. 30:52like authentication and databases and commerce and security and create a bulletproof app. there's not one out
170. 30:59there that that does it, you know, super simply right now. Um, but I would say probably within a year there will be.
171. 31:06So, like right now is a really good time to do it because there's good opportunity. Um, because once it's
172. 31:12trivial for anyone to make an app, then you're going to be competing with 98.5% of people that aren't coders. So, hope that helps.
173. 31:21Erica Hana, yeah, also how much ad spend to promote it? Um, people never disclose that. Yeah, check in. What does SAS
174. 31:30stand for? Oh, we did that already. Um, yeah, the the um Erica's point's a good one. So, I talked about customer
175. 31:39acquisition cost or CAC as they call it. If if you're talking to a VC and you're like, "Hey, I got a SAS company."
176. 31:48They're probably going to go, "What's your CAC?" You're going to be like, "Excuse me, did I stutter?
177. 31:55What is CAC?" customer acquisition cost. So, you want to know your customer acquisition cost. So, if if you run let let's say um
178. 32:07you you have a a six-month churn, right? And you get a customer and you're charging them 20 bucks a month and on average your customers last
179. 32:17for six months. Ideally, they last longer than that, but let's say on average they last for six months. the
180. 32:23lifetime value of that customer or the I guess the annual value of that customer will be $120, right? Six months times 20
181. 32:31bucks a month if you're charging 20 bucks a month. So you're going to make $120 on that customer. If you spend $200 in advertising to get
182. 32:42every customer that earns you $120, you're not making money. But if you spend $20 per customer to acquire them and you're making $120,
183. 32:53then you're making $100 per customer and you might be making money. So So the here's the other thing about about
184. 33:03SAS applications and I where did I hear this? I think it was just a Tik Tok channel. And I forget I forget who said it. I'
185. 33:14I'd like to credit them with this, but they basically said, "Listen, one of the challenges with vibe coding is you vibe
186. 33:22code up an app you think is really cool and then you put it out in the world and you wonder why no one comes and and uses
187. 33:28it." Well, does it actually solve a problem that people have? Does it solve a problem that people have that they
188. 33:36there's enough urgency around that they're actually willing to pay for a solution? And then can they find you?
189. 33:42And so all of these things are really important. So I would say a a big thing that you should do before you start vibe coding
190. 33:53is a phase that I'm calling in my new book validation. So you've got your ideiation phase where you're like, "Oh,
191. 34:00I got an idea for a SAS app. Cool." The validation phase before you get the planning and building what you're going
192. 34:08to go do. The validation phase is like go use deep research. Go use GenSpark or Manis or ChatgPT Deep Research or
193. 34:18Perplexity and go say, "Hey, I've got an idea for an app and go find me any competitors. Find out how much they're
194. 34:26charging. Find out if people are actually signing up for this. Find out what the churn rate is for for this
195. 34:31category. Find out what the customer acquisition cost is for this category. Go find out if anyone actually wants
196. 34:40this thing. Right? Very often what you'll find is there are already two or three major players doing what you want
197. 34:47to do. It doesn't mean you can't do it because execution's hard. You might be able to do it better than they do it. But
198. 34:55if you build it and just assume you're like the only one out there, that's just [ __ ] stupid because we now have deep
199. 35:01research, right? We now have agents that will go out and find out all this [ __ ] for you, right? So do that first. All right.
200. 35:15Uh uh uh new book. Um I check out for a week and there's a new book. Let's do a deal for for a launch party.
201. 35:24Yeah, let's do it. Um it's it's uh my new book's called 10person team and it's about if you um learn to feed your
202. 35:35prompt like a producer. You put yourself in the producer role, then you can sort of spin up a 10person team that helps
203. 35:42you take any idea that you have and 10x its impact and 10x its power and things like that. So, I've been thinking a lot
204. 35:50about the the progression of how you move from an idea to really launching something, launching and growing
205. 35:56something. And you know, and the phases are ideation, validation, validate your idea, right? If you're going to put time into this thing,
206. 36:06[ __ ] figure it out, right? Then planning and then building and then launch and growth. So, anyway, yes, yes, yes. I like it. It's a vibe.
207. 36:19Thank you. Yeah, I'm excited about it. I'm really excited about it. And you know, the the thing the thing that that
208. 36:25the 10erson team thing does is one of the things I've been thinking a lot about is every one of us puts limits on oursel. Let me switch my
209. 36:41view here for a second. Every one of us puts limits on oursel and there's self-imposed limits and then there are
210. 36:47actual just limits of you might not have skills, you might not have knowledge. Some of the self-imposed limits are you
211. 36:55might not have confidence to like I'm not really a business person. I don't really know a lot about business but I
212. 37:01kind of have this idea for business. Um, so our limits are defined by our gaps in knowledge and skill and confidence, I
213. 37:15guess, is the third one. What AI allows you to do is to fill in those gaps, right? If you're like, "Well, I can't be an artist. I'm not
214. 37:24creative." Well, now you are. Well, I'm not really a writer. Well, now you are. Well, I don't know how to write
215. 37:30a business plan. Well, now you do. Right? uh I can't do um research and analysis. Oh yeah, actually you can now,
216. 37:38right? You can spin up a team member to do that. So what you can do is you can just put yourself in the role of the
217. 37:45producer where your only job is to have the idea and hold the idea and like you get to you get to define what good looks
218. 37:54like, right? And then AI fills in all these gaps that you don't have and you can just sit there and be like the
219. 38:01orchestra conductor. So that's the basic idea. So I'm super excited. Under and overhyped. Oh, I like that.
220. 38:10Yeah, that's actually a really good point. I say that AI is paradoxically under and overhyped. Yeah, I think
221. 38:17everyone talks about it being overhyped. I think it's actually underhyped. But I will agree with you, Pate. I gotta tell
222. 38:24you, man. I try I took on a video job over the weekend and and like every [ __ ] time I try to take something
223. 38:32beyond the prototyping stage, try to make something professional, like complete. Oh my god, is it a nightmare,
224. 38:42right? Because it's like AI is really good up to about 80 90%. But it's the 10% that makes a project [ __ ] viable.
225. 38:52And it's just like ah it's maddening. So, and you'll find out with vibe coding, too. Like, when you start vibe
226. 39:01coding your SAS app, it it will absolutely blow your [ __ ] mind how quickly you can get that thing to like
227. 39:0980. You're like, "Holy [ __ ] this thing works." And then you're like, "Oh, well, wait. It's creating two things here, and
228. 39:15it it should only create one. Let me just go fix that." And then you go try to fix that, and it breaks everything else.
229. 39:22It's a [ __ ] nightmare. Archetypo, what's happening? Oh, man. All right. Um, [Music]
230. 39:38is there an AI bubble? Au. Oh, yes, definitely. Um, well, so, so, so there's a couple of things going on. I think we're gonna see waves
231. 39:51of AI bubbles. Um, the fact that Meta was hiring 24 year old developers with $200 million signing bonuses, there's a sign.
232. 40:06there's a sign that we might be in a bubble that if you're that desperate for talent that you're willing to pay $200
233. 40:14million to get an engineer um for your AI initiative um that might be an indication we're in a bubble and
234. 40:24I think as the tools get better and better and better and as more and more compute comes online TPUs in Google's
235. 40:31case and GPUs in Nvidia's case Um, we're going to see new capabilities coming and we're going to see as as um,
236. 40:42agentic AI gets stronger and stronger, I think we'll see waves of bubbles where we're like, it's absolutely ridiculous.
237. 40:49It's going to crash and it's going to sort of it's going to build ridiculously, then it's going to plateau
238. 40:55and and the theory is that at some point the bubble will crash and will come crashing back down. I think what's going
239. 41:00to happen is it's it's going to build ridiculously plateau and then some new breakthrough is going to happen. It's
240. 41:06going to build again and then that'll plateau and then it's going to build again. I think we're probably five years
241. 41:12out from this thing getting really frothy. Um, metaphorically, if I compare this to the early days of the worldwide web,
242. 41:21we're probably more like 96 97 than we are 1999. And the bubble burst in 2000. So 1997, 1998, 1999, it the worldwide web got
243. 41:36absolutely [ __ ] ridiculous. There were things being funded that shouldn't have been funded. It was just stupid.
244. 41:43There was just stupid money in it and it needed to crash. I don't think we're there yet with AI. I think I think there
245. 41:50are some areas of AI that are like that, but I think it's just going to keep rising. So anyway, um it's like the.com bubble.
246. 42:00A lot of businesses are going to die soon. Yep, I agree with that. And then and then new ones are going to roll out.
247. 42:05The difference between this and the.com boom is that the core feature of the technology of the worldwide web. It was a single feature.
248. 42:17It was the hyperlink was the single thing that was the foundation for all of the worldwide web and all of the internet
249. 42:28and all of us being connected all the time and commerce all that sort of stuff. There were things that were
250. 42:34layered on top of that, but the core breakthrough technology was the hyperlink. Very simple technology.
251. 42:40AI is much more complicated, much more sophisticated, much more capable than just that. So, so I think where you're
252. 42:49going to see growth and evolution is just going to be it's going to just be way more complicated than than the worldwide web. Um,
253. 43:00YouTube comment, David, but then again, Zuck wasted hundreds of million dollars on metaverse and didn't get anything
254. 43:06from it really. Well, no, that's that's my point is that I I listen, I know there's some talented 24 year olds out
255. 43:13there. Um I don't I can't imagine one that's worth $200 million. I mean, if you back a a trusted founder and they
256. 43:23have a $200 million exit 10 years from now, that's a that's a good startup. So to invest $200 million just as a signing
257. 43:32bonus to a 24 year old just as an employee that they better be doing some pretty [ __ ] remarkable work. Um yeah,
258. 43:42that's that's some of the bubble behavior that's out there right now. [ __ ] sucks. Exactly. Hopefully they'll shop at the AI salon
259. 43:54mall. That's great. the smart VCs will stop will start shopping and consolidating talent. Yeah, I'll tell you. I'll tell you here's one that
260. 44:06I I think is if if you want an area to play in um where I think there's massive opportunity. This is sort of Gary Vee
261. 44:17level. You know, you wanna you want to make a name for yourself right now? Get yourself turn yourself into a content creator, an
262. 44:29AI content creator. Make AI videos, AI images, AI songs. Get really good at content creation. And don't just, and I
263. 44:37don't just mean like I can prompt and make a cute image. I mean put together projects. Put together two, three minute
264. 44:46short films. Put together music videos. Put together um art projects that that take images and and do something with them. 3D space
265. 44:56explorable 3D spaces. If you can establish yourself as a talented content creator right now, one of the things that I'm seeing and I'm
266. 45:06seeing it more and more and more and more is that the content creation platforms are turning into media
267. 45:13companies. They're turning into content companies. They're turning into film studios. And who they're going to hire first and
268. 45:22who they're going to feature first on their TV channels. Like Runway um Runway ML right now just launched a TV channel.
269. 45:30What's it called, Brandon? Do you remember what it's called? They have a they have a a channel now. Sunno has a
270. 45:38radio station and they've now got this thing called Hooks, which is like Tik Tok. Yeah, Flume. Let's Let's go check out Flume. Um,
271. 45:51runwayml.com. And if you're old enough to remember, get started. Where's Flume? Is it runwayml.comflume?
272. 46:08and go to dashboard now. Let's see. RunwayML.comflume.
273. 46:25This page does not exist. Take me home. [Music] URL. Anybody? A Sununo creator in
274. 46:40Oh, yeah. Auno creator in Mississippi just got a $3 million recording contract. Yeah. Yeah. She created the
275. 46:47virtual her the like the the voice, the look and feel. She's a producer. She's created the songs using Suno. She's got
276. 46:55a $3 million a $3 million deal. Um, it's going to seem weird right now, but I'll tell you what, when MTV first came
277. 47:07out, it seemed weird. People are like, "Wait, what? You're watching songs. This is a channel about watching songs."
278. 47:18And it became a juggernaut when Netflix went from delivering physical DVDs to streaming to then producing original
279. 47:28content. People were like, "Wait, a streaming service is making content." And then they won some Emmys and some
280. 47:34Oscars and holy [ __ ] all of a sudden they were taken seriously. That's going to start happening with these content
281. 47:41creation. Boom. But what's the URL? I guess I could go Google it, couldn't I? Runway. Boom. [Music] LA.
282. 47:55[Music] Where is it? I just don't see it. I don't I don't see it in in Google search. Or did I not say Yeah, I said
283. 48:10fume. Runway fume. Runway mlfoom. App dot watchfume.com. Okay, here we go. watchfume.com.
284. 48:31For kind of context, I was working on a project in the Netherlands where I wanted to kind of generate a tulip using
285. 48:39a GAN. So, I took 10,000 photographs of tulips, stripping the tulips, holding the tulips, photographing the tulips.
286. 48:51But after I made this data set, I realized that it was important for me to present it as an artwork in and of itself. [Music]
287. 49:05And this is now a massive installation, >> right? So, here's >> you understand. >> Here's a documentary about someone
288. 49:14making data sets, generating content, and then there's other things here that are runway ML created um films,
289. 49:25right? And these things are going to go from curiosities and weirdnesses to all of a sudden Paramount is going to buy one of these
290. 49:37media properties, right? or they're going to acquire them or something like that or these things are going to spin
291. 49:42up and become as powerful as Netflix. Um, so like and and that's going to happen quickly. That's going to happen
292. 49:49probably over the next three years because what's going to happen is right now the technology is still bad enough
293. 49:57that professional filmmakers can point to it and go, "See, it's still shitty." And they and and they'll point at how
294. 50:08amateur-ish anyone who's making films with this is and how bad the physics are and how bad the editing is and how bad
295. 50:14the continuity is and there's no character consistency. And in the past three months, character consistency has gone from unachievable
296. 50:23to it's pretty close and three months from now it'll be really close, right? And so what's going to happen is you're
297. 50:30going to have existing story makers that ignore AI and resist it. You're going to have new people that embrace AI and get
298. 50:39really good at it, but they don't have storytelling chops. And then you're going to have existing filmmakers that
299. 50:43embrace AI and take their really good storytelling chops. and and those latter two categories, pure AI storytellers or
300. 50:52existing AI storytellers that level up with AI, they're both going to start rising above the noise and the
301. 51:00traditional filmmakers are are going to be [ __ ] in a world of hurt and it sucks be trying to catch up and there's
302. 51:08there's there's a huge opportunity right now. Huge opportunity. All right. Continuity is the biggest thing. Where
303. 51:20are you watching this? The doc the documentary is at watchfoom.com. W Ah foom m. And that's runway ML's new like MTV, I guess.
304. 51:34It's crazy. Absolutely crazy. Um Bam. Let's go make music.
305. 51:48Um, so last night, so if you don't know it, SNO, which is the new or not the new um, which is the music generation tool
306. 52:02that's kind of the industry leader. It used to be Sunno and Yo were duking it out and UDIO's just sort of fallen off
307. 52:09the face of the planet. Um, 11 Labs now has a music generation tool, but 11 Labs music generation tool is probably like
308. 52:18Sunno's 3.5, which was a year ago, year and a half ago. Sunno just dropped um V5 and we made a we made a song last night um
309. 52:33and I'll play it for you. It's really good. [Music]
310. 52:51A cafe door, a fleeting face, a matchbook offered. Time misplaced. She slipped it deep. Her coat held tight. No flame was born,
311. 53:09just endless night. Years roll by like tides on stone. But sun lit still chills the batch
312. 53:24of memory never struck a spark that faithful got to pluck. A stranger's gift a fleeting glance a fire lost to
313. 53:42circumstance. Ooh, [Music]
314. 54:00the scent of rain. >> A thousand lives. She's lost in one. Her pocket holds a quiet plea.
315. 54:16a relic from what I could not be. What if the mag kiss the whip? What if the clock had passed it tick? [Music]
316. 54:33A tiny blaze of fleeting glow. Will she have stayed? We'll never know. book memory and never struck a smart
317. 54:50faithful God. A stranger's gift a bleeding glance a fight a lost circumstances.
318. 55:04[Music] Hey. [Applause] [Music] [Applause]
319. 55:21[Music] Hey.
320. 55:39[Music] Nice. Um, interesting comments while that was playing. Um, if that if that elementary school student makes a real
321. 55:50artist feel scared, it says more about the artist than the medium. Well, I mean, listen, here's the thing.
322. 55:56I don't trivialize the fact that this that this stuff is scary. Like if you've been a musician for 20 years or 30 years and all of a sudden
323. 56:06this thing comes along that can do what you do better than you can do it or or even approaching the level that you can
324. 56:12do it. That's [ __ ] terrifying. Like that I get. Um Ashu asked the question, "What do you say to the cohort of people
325. 56:19that want to see AI for the public at least crash and burn?" Um what what I say to them is I absolutely understand.
326. 56:35I understand why you want it to crash and burn and I would encourage you to to hold the hate of AI in parallel with the fact that you
327. 56:50actually try it. like you can you don't Ann Murphy made a video about this today that I thought was really good. So someone said how how
328. 57:00do how can you help me not hate AI and and I her her response was something in the neighborhood of you know what you
329. 57:08can hate AI but I want you to try it. You can you can keep hating it. What what what my experience has been is that
330. 57:17the people that vehemently hate AI have never actually [ __ ] used it. Because the minute you use it and the
331. 57:28minute you shift your mindset from AI is this tool that's going to replace me to AI is this tool that I can strap on
332. 57:37like a jetpack and augment my ideas and amplify my ideas and create this 10person AI super team that can fill in
333. 57:46the gaps where where I don't have knowledge and I don't have skills. All of a sudden AI starts to get pretty [ __ ] exciting.
334. 57:56So, so hating AI in the absence of actually understanding it, that's the thing that is it's the ironic thing is
335. 58:06it's going to hurt them the most, right? It's a self-fulfilling prophecy that if you just if if you use your hate
336. 58:14of AI as the justification for sitting on the sidelines going, I'll never use it. then AI happens to you
337. 58:28and it's going to be [ __ ] miserable. But if you're like, I hate AI. I hate everything it stands for, but you know
338. 58:36what? I'm pretty sure it's not going away. So, I'm going to go hang out at the AI learning lab or I'm going to go
339. 58:43to join the AI salon or I'm going to go to the she leads AI create conference and I'm going to sit there pissed off
340. 58:49and hating it, but I'm going to learn about it and I'm going to try it and I'm going to see what it's like. There's
341. 58:56we're we're trying to get we're trying to arrange right now. There's a musician named Gabriel. I forget his last name. I
342. 59:02don't know if you remember it, Brandon. Um, he's been like a a songwriting producer for I don't know 25 years. Like
343. 59:12he's an older Gen X dude. And someone told him about Sunno that you can upload your songs into Souno and
344. 59:20it it will produce them and you're you can just like record like a song sketch into Sunno and it'll produce it. And
345. 59:27there's these videos of him where he's like he's pissed off and intrigued at the same time. Hang on. Got to do a Tik Tok physical dexterity
346. 59:43challenge, which I failed. Damn it. Um. Uh oh. What happened? All right, there we go. Um, and and he made he he thankfully made
347. 59:58these Tik Tok videos of himself playing his song into Sunno and then hitting the generate button and and like 10 seconds later,
348. 1:00:10two versions of his song come out that are fully produced. And you can watch his face melting and you can watch him
349. 1:00:16trying to process what this actually means. And then over the next week or two, he takes his most complicated song and then
350. 1:00:25his his his favorite song and then a a song that he wrote for a teacher when he was 17 and he puts them all in. And you just witness him
351. 1:00:36co-processing how pissed off he is that everything he's done in his life is now up for grabs and going, "These tools are remarkable
352. 1:00:47and I'm going to learn how to use them because he's smart enough to recognize that this shit's not going away." So
353. 1:00:53that if he wants to be a relevant music producer, he's at least got to understand what he's [ __ ] competing with, right?
354. 1:01:01That that kind of attitude is what I would say to the haters. Hate it all you want, but [ __ ] understand it.
355. 1:01:11Like understand it from the inside. Understand it from using it. Because what you may find is that once you
356. 1:01:19actually understand it, you're like, "Huh, I actually kind of like this. This can actually make a difference for me.
357. 1:01:25It can make a difference for other people." Um Jeff Flanigan on YouTube spotlight. Um hate it but firsthand. Well, but what
358. 1:01:36I would say is hold hold the hate, right? Just put the hate on pause. You can come back to the hate. Just pause it for a second. Try on AI.
359. 1:01:47Oh, I still hate it. I hate the way it was trained. I hate the way it was trained, too. And I also know that we
360. 1:01:53would not have the AI we have today if they had asked permission to do that. I know that, right? And they're paying the price.
361. 1:02:02Anthropic just got slapped with a $ 1.5 billion settlement. Now, luckily, they raised $10 billion, so they're going to spend 1.5 of that
362. 1:02:12paying book authors. Great. And now that it's trained, we have this remarkable [ __ ] tool, right? And so
363. 1:02:22just pause the hate for a moment, try it on, see where it is, and then think critically about where where you stand with it. You
364. 1:02:32can return to the hate. You can return and say, "Hey, I did it. I went in. I had a couple of Kevin Mallister moments.
365. 1:02:38I'm going to choose to never use it again. I'm getting a cabin in the woods. Peace out. I'm checking out
366. 1:02:45um Erica, the same people hated email um and didn't trust the internet. But uh so this is the other thing. So so the one
367. 1:02:53thing is I don't think that you can I don't think it's realistic for AI optimists like myself to evangelize
368. 1:03:04people that hate AI not to hate it. But what I can evang evangelize is give it a [ __ ] shot. Don't sit on the sidelines. Right.
369. 1:03:16Here's the thing about the haters. The haters have always hated. Every single every single technological
370. 1:03:26revolution in human history has been met with people that say this new technology will ruin the world.
371. 1:03:36It happened with the steam powered loom. It happened with the printing press. I'm sure it happened with fire.
372. 1:03:43You realize I could burn down the village. Don't you dare bring fire near me, you heathen. Right. Wheels. Can you imagine how many people
373. 1:03:57got rolled over by those big giant stone wheels and got crushed to death? I'm sure there were the equivalent of
374. 1:04:04posters, anti-wheel posters in the wheel community. Let's go back to dragging things. [Laughter]
375. 1:04:20Right. The [ __ ] steam engine. Think about this. Over a 40 or 50 year period from when the steam engine was invented, 80% 80%
376. 1:04:35of the jobs that people had were eliminated. 80%. Now it took 40 or 50 years. We are likely going to live through
377. 1:04:52something like that happening over five or 10 years. Probably more like 10 cuz [ __ ] takes longer than you think it's
378. 1:04:58going to. But probably over 10 years, 80% of the current jobs are going to be gone. And and then you're like, well,
379. 1:05:06wait, what happened to those 80% of the people? Well, those 80% of the people all worked on farms. We were an agrarian
380. 1:05:13society. The steam engine came along and it it had better muscles than men and horses did, right? 10 horsepower. Wait, we could get rid of
381. 1:05:2510 horses, right? How many people power is that? Oh, that's like 50. Wait, 50 people are going to lose their job? Yep.
382. 1:05:36What happened to those people? Did they just dig a hole and crawl in it and die? No. We're human beings. We adapt.
383. 1:05:44They they left the farm. Where did they go? They went to New York City. They went to Philadelphia. They went to
384. 1:05:51Cleveland. They went to cities. They created cities. They created the service economy. They created restaurants. That they
385. 1:05:59created jobs that when they worked on a farm, you couldn't imagine that that would be a job. Wait a minute. You can
386. 1:06:06you can put a cart full of meat on the street and people will walk by and they will give you money to cook them
387. 1:06:17meat on the street. Yeah. How How do you do that? I I made a cart and I have hot water in it and I stick the hot dogs in
388. 1:06:26the water and then I have mustard in a little thing here. What happens when it gets sunny? Oh, I have an umbrella.
389. 1:06:35And I painted my umbrella yellow. And people like my yellow umbrella, so they come buy my hot dogs. That's a job.
390. 1:06:43Yeah. Yeah. That's a job. That's still a job today, right? So, we don't know what the next jobs are going to be,
391. 1:06:56but but the the fear is still very real and the steer the fear is still very palpable. What what makes it particularly
392. 1:07:11um pointed right now is that all of the well-educated people, the
393. 1:07:26knowledge workers assumed that they were immune from this change. And they're not, you know, frankly, I don't think
394. 1:07:41anyone's immune. I think if you're a plumber and if you're paying attention to what's going on with humanoid robots,
395. 1:07:48at some point we're going to have humanoid robots coming over to the house to fix the plumbing. Now, is that, you know, this year next?
396. 1:07:57No. Is it five years from now? Probably. What the [ __ ] What are the what the [ __ ] are we supposed to do with that? I
397. 1:08:05I don't know. I don't know. But but what I do know is if you're sitting on the sidelines just watching the spectacle,
398. 1:08:16it's going to suck for you. It's going to suck bad. So, you might as well [ __ ] get, you know, grab a surfboard. I hate the
399. 1:08:29water. Okay. How do you like waves smashing you in the face? Oh, that's not good. All right. Well, paddle, pick up a surfboard, paddle out.
400. 1:08:42Two choices. Deal with it or not. Anyway, who will be the first to go?
401. 1:08:57Selling hot dogs is a safer job than being a parallegal. It is. It is. And scene. Yeah, exactly. Um, let's go. Let's go make some songs.
402. 1:09:08All right, we're in create. Okay, so we're going to do this. We're going to do this interactive style. Um,
403. 1:09:18so I want you can either shout out on Wait, creatives outside of the ones that use see themselves on the side of
404. 1:09:27soul versus soulless. Yeah, but Ashu, here's here's the Okay, as as someone listen, I have a degree in [ __ ]
405. 1:09:34acting. Sorry. Sorry. Back on the soap box, Brandon. Thank you. I have a degree in [ __ ] acting, right? I've been a
406. 1:09:41creative a creative professional my whole life. The trope that creatives have right now that AI can't be creative is simply wrong
407. 1:09:55because they're missing a huge part of the equation. These AI tools are not prompting themselves. There's a human being there sitting in
408. 1:10:06the producer role. Now, how these tools produce content is completely different than how it's produced today. And it may feel like
409. 1:10:18cheating, and it may feel um unfair, and it might feel like I spent 20 years learning music theory just to have some
410. 1:10:26stupid [ __ ] machine be able to do this. It's where we are. And what what the creatives are discounting, the soul versus soulless
411. 1:10:36thing. I agree that the technology on its own is soulless, but the technology when you when you strap AI on and you you sit in the
412. 1:10:46producer role and I say, "Hey, I've got an idea for this song called Weird Mary." That was not my idea. I sort of said, "I wanted some
413. 1:10:58non-traditional ideas." Chat GPT came up with that idea and it sounded weird and I liked just how it sounded. And then I stepped into the
414. 1:11:08producer role and I said, I'm going to take this idea of weird Mary and I'm going to turn this into something that I like.
415. 1:11:14And then over the next three hours on this live, we crafted the lyrics. We co-wrote them. The irregulars, the people that were
416. 1:11:22here co-wrote them with me. We put them into Sunno. I created, I don't know, 15, 20, 30 variations of the song until we
417. 1:11:31got one that was like, "Oh, that's weird. That's kind of cool." And then we tweaked it and tweaked it
418. 1:11:37and tweaked it. And over three hours, we generated this thing that I was super happy with. That's the soul is me.
419. 1:11:49The AI was just the tool that I used to take some nebulous idea and bring it to [ __ ] life. So, it's not soulless. It's a human
420. 1:12:01being using a tool that's incredibly powerful. But again, if you're sitting on the outside of it and you assume that AI is this,
421. 1:12:11>> you can make money with >> you push a button and out squirts a a musical, then yeah, of course you're going to
422. 1:12:18resent it. Of course, you're going to say it can never be creative. It's soulless. It will never have soul.
423. 1:12:25They're absolutely right in their conviction, but they're not looking at the actual reality of how these tools are being used. So,
424. 1:12:39you want to go listen to Weird Mary? Have you all heard Weird Mary? I know the irregulars like, "Oh, Weird Mary, Kyle, could we could we not
425. 1:12:49um share this tab instead?" Weird Mary from Cedar Hill.
426. 1:13:09[Music]
427. 1:13:27Can't be singing. [Music]
428. 1:13:42Cedar Hill is a quiet town where the oak trees sway. >> The visuals here, Kelly Bosch did the visuals for this. She used Chat GPT to
429. 1:13:55make the images and then I think she used Lumal Labs to animate them. >> The whistle of the train echoes through the day.
430. 1:14:08People greet you with a nod. Nothing ever seems to change, but the stories still unfold. Each unique and strange.
431. 1:14:23[Music] She grew up tough in the town we called home. Folks never took to her, no matter her tone.
432. 1:14:40longing to be on the inside. Longing for love just a little. She found herself as a guest in the nervous hospital.
433. 1:14:54>> Like like that that line right there. She found herself as a guest in the nervous hospital. Nervous hospital was a
434. 1:15:01line from Joker. Like throughout this song, there are things that people people humans with souls contributed to this creative process,
435. 1:15:12right? Shadow lighting in the direction of these visuals is chef's kiss, isn't it? And this is like a year and a half
436. 1:15:18old. That song might benefit. You know what's funny? Archetypal, I was just thinking that. Um, let's go. Let's go
437. 1:15:25find Weird Mary. Did we do Weird Mary and Sunno? I think we did. Oh, no. We did it in Udo. That's okay. Um,
438. 1:15:35all right. Let me see if I can go find it. Hang on a sec. Weird Mary. We'll we'll upload it into Sunno and remaster it. We did this in
439. 1:15:45UIO because Oh, that's right. Because when we did this UDIO, so when it when it was in version three had
440. 1:15:57really bad um compression artifacts. Everything sounded like autotune. It was really bad. And Udo was clean. So when
441. 1:16:06when we made Weird Mary, Udo was my tool of choice. And then when Sunno 4 came out, it got better. And
442. 1:16:14then when 4.5 came out, it got good. And then 4.5 plus came out, it was really good. And now five's great. Um, let me
443. 1:16:24see if I can find Weird Mary. Weird Mary Wave. Okay, there it is.
444. 1:16:39That's it, right? Yeah. Okay. So, let's go to uh uh uh uh uh uh uh. So, we're going to go plus audio. We're going to upload
445. 1:16:53some audio. Am I sharing this? I'm not. Hang on. There you go. Grab weird Mary. Weird Mary.
446. 1:17:19All right, that's Flack. I guess I can open Can it Can it open that? Yeah. All right. Save. All right. So, it's uploading that clip.
447. 1:17:30Let me share this tab. Okay. So, here we are. So, we're taking So, here's what we're going to do. Oh,
448. 1:17:36Erica's out of here. I got to jump. See y'all tomorrow. All right. Peace out, Erica. Good to see you. Um
449. 1:17:46All right. So, we're So, we're uploading the Weird Mary clip, and then we're going to do a We're going to do a remaster of this,
450. 1:17:58which should be interesting. I've made thousands on my soulless writing. Yeah, there you go. I think that you are doing the job.
451. 1:18:12Wait. Oh, thinking that you are doing the job then realizing AI is mowing and how you create believers in AI. Oh, how wait how it's moving.
452. 1:18:27Oh, how it's mowing and how you create believers in AI. Listen, I I am I am I am personally a technology optimist and I am personally
453. 1:18:41like one of my passions in life, the through line in my career is finding ways to use new technology to
454. 1:18:51as a tool of self-expression for human beings. Oh, wait. Continue. Okay. Um, so it's something I've been super
455. 1:19:01passionate about my whole life, but I don't actually have a I don't love or hate AI. I'm excited to under
456. 1:19:40Um, this thing's not uploading right. Something's broken here. We're losing you. Oh, yeah. I thought so. All right. Um,
457. 1:19:54let me try something here. Open with QuickTime player and then let me export this as audio only. We'll do it to the desktop.
458. 1:20:13Okay, coming back now. Yeah, something was weird with that upload. It It was a flack file and that's I think that was bad.
459. 1:20:24Okay. So now now I've got it as a M4A file. Save. Continue. I don't know. Maybe it's maybe it's you
460. 1:20:41might lose me again here in a second. was searching for this comment I left on LinkedIn recently. Uh I thought it was me. How are Okay.
461. 1:20:55Did that upload? Yeah, it did. Okay, great. So now we're going to go create remaster variation strength normal or high.
462. 1:21:15Um, we'll go normal. Remastering clip. The The other thing that's amazing about this is how quick it does this. So, we
463. 1:21:27uploaded Weird Mary, and now here we are. We're done. There are only so many letters, but it's how you arrange them. There are only so
464. 1:21:39so many chords, but it's how you play them. That picture of the sunset you took, it's a replica. It's not real, right? Yeah. Exactly. Like
465. 1:21:51one of one of the one of the questions that bugs the [ __ ] [ __ ] out of me right now is, "Oh, is that AI?"
466. 1:22:01Because if your answer is, "Yes, I used AI for that," then they completely dismiss the work. Even if you only used AI for some small
467. 1:22:11portion of it, it's it just it just drives me [ __ ] crazy. Anyway, let's listen to let's listen to AI. How AI ruined this AI song. [Music]
468. 1:22:28That's cleaner. Well, that's wild. [Music]
469. 1:22:46[Applause] [Music] That's bad. That muddied it up. [Music] [Applause] It's weird. It's like taking the
470. 1:23:01artifacts that were like bad audio and turning those into bad instruments. [Music]
471. 1:23:16Yo, Andy, what's happening? You're watching the live. We're having I don't know how long you've been watching. I've
472. 1:23:21been ranting for about an hour. [Music] [Applause] [Music]
473. 1:23:46Cedar Hill. [Music] >> Yeah, I don't like it. Okay, so let's let's go back to this one. Let's Let's experiment here. So, we're going to go
474. 1:24:00remaster. I'm going to go I'm going to make this one subtle. So, the variation strength here, I'm going to make it low.
475. 1:24:08And then I'm going to go down and I'm going to remaster it. I'm going to make it high. And then we're going to do a cover of it.
476. 1:24:19So, I'm going to go down and I'm going to take the remix edit. I'm going to do a cover. And then the cover I'm going to do,
477. 1:24:30let's see, acoustic folk rock style in a minor key with a male vocalist. Oh, that's interesting. So, when it imported this,
478. 1:24:44it wrote a description of the music. Huh. All right. Right. So, I'm going to do a version where I just leave that the same cover.
479. 1:25:00And then I'm going to do a version where I just put um Americana um rock with um
480. 1:25:20acoustic and electric guitars. Um, a solid beat. Wait, a solid beat
481. 1:25:34and uh, let's see. And, um, three-part harmonies and we'll create that. Okay. So, let's go see what the [ __ ] we've just created.
482. 1:25:50Okay. So, we had two versions that sucked. So, these are the these are the two. The next two here are um low low variation.
483. 1:26:03So this should sound like what we had. [Music] Yeah, that sounds more like it. [Music]
484. 1:26:37Cedar Hill is a quiet town where the oak trees. >> That's not bad. That's that. So there's so tomorrow, starting tomorrow, there's a new
485. 1:26:54um there's a new thing coming from UDI or from Sunno called Sunno Studio. And I think premier members I think I'm a
486. 1:27:02premier premier member. Premiere members get to play with this new studio. So, what I'm going to be able to do is like
487. 1:27:09right now I feel like the mix in that that early instrumental thing is is weird, but I'm going to be able to go
488. 1:27:15into it and take like the guitar levels and turn them down a bit. So, like this, we might be able to go in
489. 1:27:23and actually edit these songs like you know the way you edit a real song. So, this is the second version of that that low variation. [Music]
490. 1:27:55[Applause] [Music] to the Cedar Hill is a quiet town
491. 1:28:10where the oak trees sway. The whistle of the train. [Music] >> All right. It's I like I can hear it. It's good, but it's like
492. 1:28:24like I I feel like the original is just fine. But I'm going to give that one a thumbs up. Like that's that's a decent
493. 1:28:30one. Okay. Okay, so these next two are with high variations. So these these might get weird, but they might get better. [Music]
494. 1:28:50[Applause] [Music] That's cool. [Music]
495. 1:29:16>> That's [ __ ] awful. [Music] Oh, that's bad. [Applause] [Music]
496. 1:29:39All right. Hate it. So now these are the covers. So the first two the first two are the covers where we left the
497. 1:29:46description what it interpreted out of the song. [Music]
498. 1:30:02Okay. So, so archetypal. Yeah. Kind of feels like the mediocre quality was part of the charm. It felt more vintage. I
499. 1:30:09agree with you. But listen to this one. So, what we were doing with that other ones was remastering it. So, it was
500. 1:30:16taking that effectively taking that existing recording and trying to recreate it. These two things are
501. 1:30:23covers. So, like I can already hear this one. This one feels much cleaner to me. And this feels like someone actually
502. 1:30:30covered covered the song. [Music]
503. 1:30:53Cedar Hill is a quiet town. >> This is good. >> Where the oak trees sway. The whistle of the train echo through the day.
504. 1:31:08People greet you with a All right, let's hear this other one. [Music]
505. 1:31:48Cedar Hill is a quiet town. >> I love that melody variation. Cedar Hill is a quiet town. Where the oak trees sway,
506. 1:32:02the whistle of the train echoes through the day. >> Yeah, this one's good. >> People greet you with a knot. Nothing ever seems to change,
507. 1:32:18but the story still unfolds. Each unique and strange. >> This is really good. She grew up tough in the town we call home.
508. 1:32:34Folks never took to her no matter. Longing to be on the inside, longing for love just a little.
509. 1:32:50She found herself as a guest in the nervous hospital. This one's really good. This one's really good. This is a really good
510. 1:32:59That's a really good version of the song. All right. Now, let's hear. So, so those two were both It basically took
511. 1:33:08the it analysis of the musical style and applied it. So, this next one is I just did a much more simple Americana rock
512. 1:33:17with acoustic and electric guitars, solid beat, three-part harmonies. So, these should be even more different. And
513. 1:33:24then we could also do this as a funk song that maybe we'll actually let's do that. We'll do um we'll do um uh we'll do 70s funk. Um uh
514. 1:33:40just 70s funk. We'll just do it as 70s funk. We'll do it really simple. All right. So here's the here's the Americana version. Listen
515. 1:33:55[Music]
516. 1:34:10to the drum. That That drum is just That's a much more like straightforward [Music] Cedar Hill is a quiet town
517. 1:34:28where the oak trees sway. The whistle of the train echoes through the day. [Music] People greet you with a knot.
518. 1:34:43Nothing seems to change. But the story still unfolds, each unique and strange. She grew up tough
519. 1:34:57in the town we called home. Folks never took to her, no mattering. [Applause] [Music]
520. 1:35:12>> There's there's a three-part harmony. [Music] That's two-part harmony, but that's cool. All right, here's the other one. [Music]
521. 1:35:24>> I like this. I like this a lot. [Music] [Applause]
522. 1:35:39[Music]
523. 1:35:55Cedar Hill is a quiet town. >> This one I like a lot. Where the oak trees sway, the whistle of the train echoes through the day.
524. 1:36:13People greet you with a knot. Nothing ever seems to change, but the story still unfolds. Each unique and strange. [Music]
525. 1:36:29She grew up tough in the town we call home. Folks never took to her no matter her tone.
526. 1:36:44>> All right. All right. That one's cool. Um All right. Let's listen to the funk one. And then I also did one while that
527. 1:36:50other one was playing with a female vocal. So that should be fun. So this is 70s funk, right? Yeah. [Music]
528. 1:37:35That's not 70s funk, but whatever. [Music]
529. 1:37:52Yeah, neither of those really worked. All right, let me let me try something here. I'm going to go We'll do rap jazz plus
530. 1:38:02uh studio recorded plus [Music] R&B. All right, so there's that one. All right. Now, let's go listen to I did two with female female leads.
531. 1:38:19[Music] And this was the the uh style here is just Americana rock. That Americana thing I wrote.
532. 1:38:34[Music] Now, I'm I'm not remixing right now, Gareth. I'm the These are all covers. I found I didn't like the remixes. I The
533. 1:38:46the original recording of it had too much like sonic mud in it that that the remixes are trying to recreate and just
534. 1:38:55made it messy. So, the cover the covers I like better. [Music]
535. 1:39:13Cedar Hill is a quiet town. >> I like it. [Music] >> She sounds like Chrissy Hines.
536. 1:39:30People greet you with a nod. Nothing seems to change, but the story still unfolds. >> It's not Chrissy Hines. That's not how
537. 1:39:46these things work. But anyway, [Music] Oh yeah. Love that.
538. 1:40:03[Music] [Music]
539. 1:40:22Cedar Hill. [Music] This sounds like Cheryl Crow. [Music]
540. 1:40:40>> People greet you with a nod. Nothing ever seems to change, but the story still unfolds. [Music]
541. 1:40:56She grew up. >> Oh yeah. >> In the town we call home. >> Folks never took to her no matter her to
542. 1:41:12longing to be on the inside. Longing for love just a little. She found a guest. >> That's a really good That's a good
543. 1:41:27version of that song. All right, two more and then we'll call it a night. Um, this is rap jazz studio recorded R&B.
544. 1:41:44[Music] like this. >> Oh, love that.
545. 1:41:59[Music] [Music] City Hill is a quiet town
546. 1:42:16where the oak trees sway. [Music] The whistle of the train echoes through the day. People greet you whether or not.
547. 1:42:31Nothing never seems to change, but the story still unfolds. [Music] >> That's [ __ ] hot. I love this.
548. 1:42:47[Music] >> See, it [ __ ] that up. It [ __ ] up the timing there. [Music] Yeah, that's not usable. I mean, you get
549. 1:43:02what's cool about Sununo is you can go in and you can just rerender that section of audio, which might be, but let's let's hear
550. 1:43:10what the other one sounds like. Like that. [Music]
551. 1:43:45Dude, it did a sack solo of the of the [ __ ] melody. Oh my god. [Music]
552. 1:44:03Sist [Music] good side hustle, Mimi. Okay. So, wait.
553. 1:44:19So, we're going to we're going to keep all of these because I like this overall sound. So, rap jazz studio recorded R&B,
554. 1:44:25but then we're going to put in Appalachian influence. That's really good. Appalachian. I'm glad you spelled that influence.
555. 1:44:38Okay. Appalachian. Oh, I spelled it wrong still. There. Oh, I needed a capital. Okay. Um, so so we'll do two more. All right. Two more.
556. 1:44:47But let's keep listening to this one because this this one's really tight. [Music] >> Nothing seems to change,
557. 1:45:01but the story still unfolds each unique and strange. >> She grew up in the town called home.
558. 1:45:15>> Yeah. folks. >> Listen to her. She's in the [ __ ] pocket here. Longing to be on the inside,
559. 1:45:31longing for love just a little. She found herself as a guest in the nervous hospital [Music] with Mary from Cedar Hill.
560. 1:45:48She's blue sadness and shame. All right. I think that one's my favorite one so far. That one's [ __ ] sick. All right, that one's not
561. 1:46:01rendering. Oh, yeah. Rendered. It's just interface problems. Let's try that one. No. No, it's dead. All right, that one's dead.
562. 1:46:15This one's alive. All right, this is this is the one where we added Appalachian Influence at the end. So, rap jazz studio recorded R&B
563. 1:46:29Appalachian influence. Let's see how it does. [Music]
564. 1:46:47Oh, nice horn section. [Music] Hill.
565. 1:47:01>> Damn. [Music] >> That just had a weird bad chord. [Music]
566. 1:47:17>> That's bad. [Music] All right, I'm already getting the thumbs up.
567. 1:47:31[Music] [Applause] [Music]
568. 1:47:50This one's making champ. [Music]
569. 1:48:08>> Damn. People greet you with a seems to change. >> This is kind of dirty. I like it.
570. 1:48:26>> She going to break your [ __ ] heart. [Music] >> Yeah. The these two are both the ones with Appalachian influence. home.
571. 1:48:45>> Yeah, [ __ ] that. That that [ __ ] up the timing pretty bad. [Music] >> Just a little strange. She grew up in the town called home.
572. 1:49:03Folks >> Yeah, that like that one's not usable. That that's I'm gonna give that a thumbs down actually.
573. 1:49:10Well, which made it disappear. That's fascinating. Okay, [Music] this is the one with Appalachian influence. I think this one's pretty good.
574. 1:49:21[Music] Oh, yeah. That's one of the other things you can do in Sunno. And I haven't
575. 1:49:37played with this much. side hustle. Mimi said she she liked the Oh, bye Cam Katkin. Um, she liked that voice. You can take any
576. 1:49:47of these voices and turn them into a character, a persona, and then use that voice in other songs. So, that might be
577. 1:49:54something to play with, too. [Music] Love that. [Music]
578. 1:50:11Hill is a quiet town where the oak trees [Music] >> one's got that weird chord. All right. So, which was our favorite? I think it
579. 1:50:24was this one. [Music] [Applause]
580. 1:50:38But the story still unfolds. The story still unfolds. [Music]
581. 1:51:08[Music] Cedar Hill is a quiet town where the oak trees sway.
582. 1:51:23[Music] The whistle of the train. >> All right. Well, we could do this all night. Anyway, um yeah, this is
583. 1:51:37this is fascinating. And you know, again, it's like I go back to, you could absolutely argue that this is
584. 1:51:45soulless music production, but I'm going to go back to, well, we took a song we put a lot of time into, we produced,
585. 1:51:53there was a lot of ideas and inputs in it. It's a song we actually know and love because it was a good song
586. 1:52:00independent of how it was made. And now we're tweaking it, modifying it, pushing it into different areas.
587. 1:52:08this is music production. Like I don't the fact that we don't have musicians, you know, running loops in a DAW doesn't
588. 1:52:18mean it's not music. So anyway, whatever. That's a that's again there's like the theory of what this is you can be
589. 1:52:31enraged at when you actually use these tools. The fact that it [ __ ] up some of the chords and it [ __ ] up the timing of
590. 1:52:39the things. Like that's like that's just like a musician [ __ ] that up. It's like were we gonna save the recording? I
591. 1:52:46don't know. Was it good enough for us to save? Maybe, maybe not. Maybe we go try something else. Anyway, crazy. Yeah, we
592. 1:52:52have no no union scale. We just have my time. Uh and all of yours. Okay, cool. Um why aren't musicians cloning their voice to
593. 1:53:01prototype songs? I'm sure some are. I mean, listen, some, you know, some people are jumping on this stuff. There
594. 1:53:08there's producers out there. Timberland is a producer that's all over this right now. Um, so anyway, all right, I got to
595. 1:53:15get out of here. It is Wednesday. Tomorrow's Thursday. I don't think anything's going on. I'll see you at 8
596. 1:53:20o'clock tomorrow night. Have a fantastic evening. Peace out. [Music]