6/24/2026 - Defining Your Digital Self to Keep AI Agents Aligned with Who You Are

Transcript

1. 0:01[music] [music] [music]
2. 0:20[music] [music] Good evening, good people. Why do I have
3. 0:34Monopoly down here? Oh, because I was was a prop. That was a prop. Was proper good fun. Good day, good people.
4. 0:45Welcome to the AI Learning Lab. My name is Kyle Shannon. I'm here in the $20 million production studio,
5. 0:52expertly styled to look like a shitty home office. [laughter] Um, welcome, welcome, welcome. It's Wednesday night, so welcome to the AI
6. 1:04Salon. If you're new here in the AI salon, let me get you oriented. Uh, on this on the side over there is the uh is
7. 1:12the navigation menu for the AI salon. And uh there's all sorts of spaces in there. They're different little think of
8. 1:19them like different little tables in a bar. Um I heard that metaphor today and I I like it. Um so if you're in one
9. 1:25conversation and you want to go hang out with people having, you know, talking about politics, there's an area for
10. 1:29that. So that's what all those areas are. Um the way this uh the live stream works, if you want to have a chat, if
11. 1:37you want to chat to me or to producer Brandon, the chat, well, as far as I see it, it's over here. But how you invoke
12. 1:43it is right here. So, you click on that and then that will bring up the chat menu. Uh, if you're on mobile, it's
13. 1:51slightly different. I think you've got the live stream above and the chat below. Um, but you can chat. So, you can
14. 1:56say hello and type hello and I can see that and we can put stuff on screen. If if you have a question, we can do that.
15. 2:02Okay. Um, if you want to do advanced live stream here in StreamYard, if you're on desktop, right down here,
16. 2:11there's a little button that will pop out this video and you can open a second tab and in that second tab, you can have
17. 2:17the AI salon. So, you can be cruising in the AI salon over here and have the the live stream popped up there. Hey, Lori
18. 2:23Blair, what's happening? What's happening? Silver Fox. Silver Fox in the house. Um, all right. Right. So, what are we
19. 2:32going to talk about tonight? We're going to talk about a lot of different things. Um, I don't know where to begin. So, on this very live
20. 2:50a few months back, four months back maybe, we we were talking about the future of work. Hey, Kim. Ken was shaking.
21. 3:02Welcome, welcome, welcome. Great to see you. Um, [clears throat] we were talking about the future of work and I I was I was I was kind of
22. 3:12lamenting what was happening with developers that even the ones that were keeping their jobs were realizing that their jobs were
23. 3:23completely different because of AI, that they were no longer programming, they were babysitting agents. and just led to
24. 3:30a whole discussion of of what's that going to look like? And talked about this invisible crisis of meaning that
25. 3:40if you if your identity is closely tied to the tasks that you do and the tasks that you do change or are taken away or
26. 3:49just automated, what what is your value? Right? if if my value was I'm a programmer. If you identify as those tasks and now
27. 3:58you're no longer programming, there's this vacuum. And so we came up with the idea of the great repurpose. And so if you go to the
28. 4:06greatrepurpose.com, you can you can see the the basic idea there. And within the AI salon inside the AI mastermind,
29. 4:14there's a whole great repurpose section. And we're populating that section with workshops and conversations and
30. 4:22trainings about how to decouple your identity from your from your work and then reconnect you with your value as a person.
31. 4:37And then once you sort of understand, oh, here's who I am as a person, then you can start to do things like, well,
32. 4:42what do I value in the world? And who do I care about? What can I do? And so we're doing all this work over
33. 4:48there and and I'm lucky enough to get to participate in those. Some of those spaces I lead, some of them I just
34. 4:54participate in. Um we had a particularly powerful one this this Tuesday. Um it was a space that Andy was leading. Um
35. 5:03that was an extension of her four-week workshop on separating your identity from your job. and it's now an ongoing bi-weekly
36. 5:13um mastermind leadership space where we're creating the future of work and it's really inspiring and and this week
37. 5:21it was it was a small cohort but what's remarkable is getting to witness
38. 5:36people as they discover their value like their value independent of the degree they have and their value
39. 5:46independent of the [ __ ] they do and their value independent of the things they make. And it's really moving like I cry a lot.
40. 5:56There's a lot of tears. It's a lot of tears for me. I'm sappy. And when I see people tapping into their humanity,
41. 6:07it's quite powerful. All those sessions, by the way, all the workshops that I just talked about, the the AI salon
42. 6:14practice labs and the great repurpose workshops and the the ongoing conversations are all available on
43. 6:20demand. So, if you're a member of the AI salon mastermind, which is it's a subscription area of the salon. The salon is free.
44. 6:28Claire Jacobs, Dr. Jay's here. Dr. Jay was there on Tuesday. There there wasn't a dry eye in the house. I think there
45. 6:36were a couple of dry eyes in the house, but mine mine were not two of them. It's really powerful. It's really powerful stuff. And
46. 6:50what we're all trying to figure out is okay, I understand that I have I have value independent of my tasks and my work
47. 7:04and I still have to make money, right? So it's like what is the path? What is this path? Right? At some point, five
48. 7:12years down the line, we're going to have universal high income and everyone will have what they want. None of us know
49. 7:18quite how we get there yet. But what what is becoming increasingly clear to everyone that that's in the meetings that I've been in
50. 7:30is that knowing who you are and knowing what you value is going to be an incredibly valuable thing as we move forward. Like
51. 7:41I think it becomes the only differentiating thing for your work versus someone else's work is who you
52. 7:48are. So people are going to hire you not for what you do or the things you produce, but for who you are. So anyway,
53. 7:54that's what the great repurpose is. Um, [clears throat] so that's one half of this thing. You're like, where's this going, Kyle? Are you
54. 8:04ever going to get there? You know, clocks are ticking. I've got some learning to do here. It's the AI
55. 8:11learning lab. I don't think I've learned anything yet. Producer Brandon. Hello, new folks. Chat with us using the chat bug in the upper
56. 8:19right hand corner. That's right there uh under the AI Learning Lab logo. Ask your questions. What should we play with
57. 8:25tonight? Beautiful. And I'm going to show some stuff, too. But yeah, if you've got something you want me to play
58. 8:30with, let me know. Claire Jacob's crying. Yeah, Claire is an amazing person. She's an amazing light
59. 8:46and presence [clears throat] and poet and artist and shrink. Is shrink porative? If it is, I don't mean it to be. She's a therapist.
60. 8:59[laughter] I'm Gen X. Just I say what I was taught. [laughter] Um,
61. 9:14okay. So, one thing that I've been doing a lot of is this work on on the great repurpose and identity and meaning. Then
62. 9:21there's this other piece part of the great repurpose. Stage four was explore AI. Explore what's possible with AI.
63. 9:32And there's something about that that's always felt nebulous to me. And it's like too vague and it's too big.
64. 9:38Like you can't just Okay, Claire Jacobs crying again. Oh, [laughter] you're amazing, Dr. Jay. You really are. Um,
65. 9:49now, now you're gonna make me cry. I can just see your face right now. Uh, and yes, you're a shrink. Okay, good. And a coach.
66. 9:56Perfect. The coaches are going to help. The coaches are going to help. Um, if you've been hanging out with me for a while here,
67. 10:18me understanding what to focus on as as we shift from AI tools to AI agentic platforms, I don't know what to focus on. I don't
68. 10:30know what to learn. I don't know what to teach. There are pieces of that world that I could give two shits about. There are
69. 10:42pieces of that world that seem like voodoo and magic to me and and feel like I could never learn how to do that.
70. 10:50the Rick Rubin of random ordinary phrases. But you don't have as nice a beard, I would argue, as Rick Claire. You have lovely hair,
71. 11:03but he's got you in the beard department. [laughter] Oh, you mean in spirit? Yes, in spirit. Yes. Um,
72. 11:26so I've been working a lot with Andy and Brandon on bringing the great repurpose to life, figuring out what we bring into the mastermind,
73. 11:37what can live outside of the AI salon and maybe turn into business stuff, and thinking a lot about And
74. 11:49my inability over the past six months to to really understand what to pay attention to has been very frustrating for me.
75. 12:01And if you come to this channel, that's probably apparent. And it it might be frustrating for you, too. Like, are we
76. 12:07going to ever learn anything again? Kyle, remember when you used to make songs? That was cute. Can we get back to
77. 12:13songs? Maybe we'll make a song tonight. Something I've been I've been talking in circles about for four or five months
78. 12:29now is that our only job moving forward if so here's the way I think about it. If AI is going to automate our tasks,
79. 12:42and it's going to automate all of them, all tasks, all of the dewy stuff, right? The writing of the emails, the making of the
80. 12:51PowerPoints, the just all the [ __ ] it's going to do it. I'm going to show you a thing I built tonight that is a whole new section of a
81. 12:59website that is dynamically updated every single day and I don't have to touch it anymore. And so every little teeny tiny thing
82. 13:09that you used to have to put your attention to is just going to happen. And so what I've talked about for months
83. 13:15is our job is going to be to be Rick Rubin to have taste. Hey Mel, what's happening? Then we get to do the heart-c centered stuff. Exactly.
84. 13:28Our job. Our job, if the machines are going to do the tasks, our job is to figure out, well, what do I want to do?
85. 13:43What does good look like? What does great look like? I'm going to have taste. I'm just going to have taste. That's going to be my job.
86. 13:59And that seems so [ __ ] nebulous, right? But that's what the great repurpose is about. The great repurposes, if your tasks are are
87. 14:07valless and the machines are going to do them, well then who the [ __ ] are you? And then once you understand who you
88. 14:14are, then you can look at things like what do I care about? Who do I care about? What do I value? And then through those lenses,
89. 14:26you can start to look at what does good look like? What is what do I want to do? What change do I want to make? What do I want to create?
90. 14:36But you have to do the work as a human. The human readiness side of this is it's really hard to have a point of view, a
91. 14:44clear creative point of view unless you're in touch with who you are, right? You know those moments where you're like
92. 14:54that sucks and you have real conviction that sucks and that's great. That's because there's this resonant
93. 15:05thing in you that when you see the thing that sucks, it goes you're like, "No, that sucks." And then
94. 15:12you see the thing that's good, it's like you're like, "That's great. If you're not used to doing that on a
95. 15:22regular basis, that's a trainable skill, but it's it's a it's it's a it's a thing you have to develop as a human.
96. 15:31And so I was thinking about that and I was thinking about that. And so what hit me, I think it was yesterday, might have
97. 15:36been the day before, is that the only way to make these agentic tools good
98. 15:50is for the tools themselves to understand who you are. Right? So the human readiness thing says you have to know who you are before you
99. 16:03can have a point of view and and and and understand what's good and what's bad, what's what's you know what resonates
100. 16:10with you and what doesn't, what gives you energy or what takes it away. If if you just want to think of it on pure metaphysical,
101. 16:17right? You got to be in touch with you. We get to do the same thing with OpenClaw and with Hermes and with Codeex and with Claude Code.
102. 16:35And so what struck me is that stage four of the great repurpose isn't just about figuring out what AI makes possible in
103. 16:50general. What hit me was, oh wait a minute, the great repurpose for humans is about finding yourself and defining yourself,
104. 17:02creating yourself. And you can do the exact same thing with technology, with skills. Wait, Mel. Oh, good. This
105. 17:15is good. She's got to push back. I got to push back a little. AI will allow us to do our soul's mission while getting
106. 17:22our adulting taken care of. I I totally agree with that. I totally agree with that. And you could do both, right? One
107. 17:28can be I'm going to I'm going to use these AI things to go do the adulting [ __ ] and make me some money. Go make me
108. 17:36some money tools and I'll pay attention enough to make sure that it's on track. Then I could also use it for this the soul ex uh
109. 17:50words expression soul expression stuff. Andre about a month ago came out with this thing about creating a wiki based memory
110. 18:07and I've been I've been trying to set up in codeex this memory framework that that Andre Karpathy talked about not
111. 18:14really understanding it just having codeex do the activities and I sort of watch what it does and have it ask me
112. 18:21questions and sort of pay attention to it and I've also been thinking conceptually about hey wait a minute all of a sudden
113. 18:30it's starting to make sense to me that the conversations I had in chat GPT and the conversations I had in Claude and
114. 18:37the conversations I had in Perplexity and all of the image training that I did in Midjourney and all of just all of this stuff
115. 18:46are expressions of who I am and what I what I like and who I am and what I value and they're in all these different places
116. 18:57and I don't own them. And so what hit me is that for today anyway,
117. 19:13the thing that I think is the most important thing in AI for us to all pay attention to is creating your digital self, right?
118. 19:27In the great repurpose, we're going to take ourself, we're going to create it. The human side, the human readiness
119. 19:33piece. And on the AI side, all of the tools are changing. All of them. The thing that is not going to change, the common denominator is you
120. 19:48are going to use these tools, but you need to give them enough context so that they understand who you are and what you value.
121. 19:56They don't know that unless you tell them. And so that's going to be the second half of the great repurpose. Now, it's
122. 20:05going to be a very focused thing. It's going to be about designing who you are. What are your tastes? How do you like to
123. 20:18write? How do you hate writing? Um what pictures do you like? What what what what are your turnons? What are your
124. 20:27turnoffs? Um how do you like to do things? Do you like to see bullet points when a thing is created or do you like
125. 20:33to see long rambly big worded paragraphs? None of it's wrong. But if you can train
126. 20:47that core, if you can create that core correctly, so it's in line with who you are as a human, you're like, "Here's who I am. Here's
127. 21:01the thing I value. Here's the people I care about." And then your online presence is some matching version of
128. 21:08that that you can then take these agentic tools and and and sort of plop them into this context and say go do
129. 21:17[ __ ] in the world based on who I am which is based on who you are. It's going to be a flipping superpower.
130. 21:29So the great repurpose there's there's a version of it that is you start out with this one purpose in life. AI [ __ ] it
131. 21:36all up. You get to choose a new one or a new 10. Doesn't matter. But you get to choose who am I? What do I care about?
132. 21:46And then then now we've got this thing we can say we can also take these generic shitty ass tools that hallucinate like jerks
133. 21:57and we can say no no no no you're not going to do it the way 24y old engineers in Silicon Valley wrote you. You're
134. 22:06going to do it on my terms and here's who I am. So, I'm putting together a series
135. 22:20in the mastermind in the in the great repurpose for how to design that thing and how to design that thing in a way
136. 22:31that any of the tools can use it. You get to build it. You get to build you. And you're like, "Well, what could
137. 22:42possibly be in there?" Like, it is a staggering amount of stuff that goes in there. What do you care about? What do you not
138. 22:48care about? What have you done? What jobs have you had? What businesses have you started? What inventions have you
139. 22:53made? What novels have you started six of and not finished? Gotten past chapter one. Put it all [ __ ] in there. Right.
140. 23:04I'm now starting to think about like all my midjourney training and creating luras and like is is my visual
141. 23:12personality in there? It should be. And then how do I point these things out in the world to use it? I don't know
142. 23:19yet. That's what we're going to co-discover. Like like in the the human side of the great repurpose, we're
143. 23:26figuring out who we are as humans. We're gonna figure out who we are as digital reflections of those humans. And let me tell you something.
144. 23:37I think there are very few people on the planet right now thinking about these two halves and balance, the human and the AI,
145. 23:49and and how you sync those together and how you leverage AI in a way that it amplifies who you are. [ __ ] fired up.
146. 24:05I used the desktop app and connected it to a database on Google Drive. Yeah. Yeah. Google Drive. Just think about all
147. 24:12the [ __ ] you've put in Google Drive over the years. That is one data source that you can that that you can use to feed this thing.
148. 24:27all the songs you've made, all the powerpoints you've created, the just all of your stuff. You could answer questions. You could
149. 24:38have chat GPT write an interview for you and just record the answers to that. Tell Chat GPT you want it to design a
150. 24:47structured interview to reveal who you are as a human. and then open up your QuickTime player and hit record and just read the
151. 24:59questions and answer them and talk into the camera and then get the transcript of that and put that in your thing and you can say to it,
152. 25:14"This is what this is." And it'll go, "Oh, okay." Okay. Put stuff in places to obsidian. Yes.
153. 25:28Obsidian is this relational notetaking thing that can make connections. It's called a knowledge graph. It can make
154. 25:37connections between all your [ __ ] I've got a musical that I work with with this guy Andrew Watts who I acted with in
155. 25:46this play back in the 2000s or the 90s and and it's all connect. All this shit's connected. And so Obsidian can
156. 25:55keep track of all that. And Obsidian's for you. Obsidian's for you to be able to look at all this [ __ ] But the AI doesn't need all that.
157. 26:05So, one of the things I'm going to talk about tonight is this new thing that Google came out with called OKF Open knowledge framework.
158. 26:22I think so. So, we're going to design this whole thing on Open Knowledge Framework. But
159. 26:37anyway, tonight's a show and tell night, not just a rant night. So, here's what we're going to do. I'm going to show you something.
160. 26:51And hopefully my machine doesn't just keep crapping out on the on the share stuff. Gareth vacation hack. And this is just one use
161. 27:07case because all the data was in Jira. Set up a loop to check every hour for new tickets in Jira from my boss. So
162. 27:15Jira is the a ticket is he's got to do some work for work then if if it was either answer ask for more detail or go and update the code or
163. 27:26build the feature. Yeah. So okay so so Gareth has set up he has created Jarvis right his own personal version of
164. 27:35Jarvis. The thing about Jarvis is he knows it. It knows who Gareth is. It knows where he works. It knows where all his
165. 27:47files are. It knows that he gets messages in Jira. So he said to his Jarvis, "Hey, I'm on vacation every hour. Ping me."
166. 28:01So, so that if I lose track of time and don't check in on on ye old company Slack, [laughter] can you show us how to connect Obsidian
167. 28:14to Codeex or Cloud Code? I will show you what I did to connect it, which is actually different than showing
168. 28:24you how to do it because I don't know. But one of the reasons that I really like codeex is that it can use my computer as if it were me.
169. 28:34Open knowledge format. Yes, format. That's what I said. That's definitely what I said. I basically had my digital clone. It It
170. 28:43knows the code. I could then work it. It It could then work as me while I'm out of town. Yeah. So, so as you build this digital self of you,
171. 28:55it's understanding all of your stuff. It's understanding that you have this job and you have these requirements in
172. 29:01that job and you produce these kind of documents and if you set it up right, it can functionally do your job.
173. 29:09Plot twist. The boss is also using AI to send Gareth messages. Exactly. Exactly. Well, so so here's the funny thing. Like
174. 29:19you joke about that, Brandon, but I do think that's where this goes. I do think that [laughter] that where a
175. 29:28lot of this stuff goes is I'm going to keep Gareth on because I like working with Gareth, but I know Gareth is using
176. 29:35his things to do the thing. And then I know I'm using the thing to check Gareth's thing, but all I know is that
177. 29:41the business needs to produce these outputs to keep working. And then the boss is also going to have finance use
178. 29:48AI to do their things. And so the the AIs will make the [ __ ] move forward in the business, deliver [ __ ] to the
179. 29:55client, the client will be happy. They'll pay money. The money keeps coming in. The stuff will keep moving forward.
180. 30:01And I think we do get to a point where most of the people in the loop are, well, while that's doing all that, do we
181. 30:08want to do something else? Make a new product? If we're midjourney, maybe we reinvent full body scans as a spa event. [laughter]
182. 30:26I haven't allowed them to get that smart yet. Them as in my boss. Oh, that's good. Yeah, Gareth is gatekeeping enough
183. 30:32of the knowledge so that the bosses still have to work and he doesn't. That's pretty good. It's one of the advantages of of being early.
184. 30:41Um, okay. So, what I want to show you, so remind me Brandon to to go talk about open. So, we're going to do we're going
185. 30:52to do a couple of things. I'm going to show lovable. I'm going to show the great repurpose website. I'm going to show codeex.
186. 31:02Uh, and then we're going to talk about open knowledge format from Google. Um, if it's a spa event, I want an ROI. I
187. 31:16don't know if you saw this, Silver Fox, but but go watch the if you go to uh just go to Twitter and search for
188. 31:23MidJourney Medical. And I'm not kidding. It's not an SNL skit. Midjourney Medical. And by the way, I saw someone
189. 31:30post on X. They completely blew it with the branding. It should have been called Medjourney, huh? I didn't think of that either. That's a
190. 31:38good one. Medjourney, MidJourney Medical. And they have created this ring with these little sonic transducers,
191. 31:48like I don't know 4,000 of them on a little plate and then it's put into an array, a ring around and it's sunk in water.
192. 31:57And if those sound waves travel in water, they travel fast enough that they actually penetrate the body and reflect
193. 32:03back all your organs and [ __ ] And so you you stand on a platform in a pool and then the platform like sinks you in
194. 32:12the water. I guess you have to hold your breath at the end and it takes 60 seconds and it's a full body scan. It'll
195. 32:19be like whatever 50 or 100 bucks in a spa and they're opening up a spa in San Francisco in Union Square.
196. 32:26Oh M MRI. Yeah, that's why I said that. Okay, good. um learned last year you don't have to automate yourself out of a job. That's
197. 32:36really smart. So, we're going to go look at all this stuff. So, the first one is I'm going to this is a this is a story of success
198. 32:47and this is a long and winding road um to generating something of value through what I considered largely to be
199. 32:55a failure. my my open claw installation while I got it installed and running, I I would consider it a failure because a
200. 33:04I don't use it and b when I do use it, it's janky. It's just not great. [clears throat] What?
201. 33:25Okay, so here we are. This is lovable. And I did a couple of things. I made a new image for the the great repurpose
202. 33:36website. So, this is where I'm excited to see what's possible with the results of the new midjourney medical scans. Hey, Cam, I saw a thing.
203. 33:48I didn't have time to uh to grab it. I was in the middle of a meeting, but look up in I think it was in South
204. 33:56Korea. There's a South Korean lab that's that's created a bunch of sensors. There's like this three-layered thing
205. 34:03that that can replace the retina. Um, and I like the the way the image was, it didn't look medically accurate,
206. 34:12so I don't know if it's real, [laughter] but but take a look. It was a bunch of sensors that they said could replace a
207. 34:18retina. So, there's really interesting stuff coming with vision. Um, okay. So, we've got the great repurpose.
208. 34:29One of the things that I had Adam do semisuccessfully. Let me go to Gmail here. >> Hey Kyle. >> Yeah. >> Your screen share dropped.
209. 34:43>> Really? >> But also a little bit of breaking news from OpenAI uh that they've designed and built their first AI chip. Jalapeno.
210. 34:53>> Oh yeah, I saw that earlier. Sam Alman holding the big old the big old chip, right? Yes, it is the first purpose-built chip
211. 35:01for the LLM workloads powering all of OpenAI's products. >> I'm glad that they designed it. How are they going to manufacture it? Because
212. 35:08I'm pretty sure that all of that stuff still has to go through TMC to produce the machines to be able to produce the
213. 35:15chips. So, the bottleneck remains >> partnering with Broadcom. >> Yeah, but can Broadcom produce chips? Oh, your optic nerves are what are
214. 35:25affected, not the retina. Well, that sounds >> I hope so. [laughter] >> Um, okay. Elon Musk was talking about bypassing
215. 35:38the optic nerves camp. Okay, let's see. Adam. Okay, so you can't see my screen and I am I am not happy.
216. 35:52Okay, you can all see this, right? Oh, that's all Broadcom does is produce chips. Ah, okay. Cool. Okay. So, you can see that. Okay. So, this
217. 36:10um the daily TGR signal Wednesday, June 24th, right? TGR, the great repurpose. So [clears throat and cough] once I got Adam installed
218. 36:23and he could see the web and he had access to my Google Drive or or Google services and that took forever. So if you're get think if
219. 36:34you're thinking of installing um OpenClaw like setting up all the connections to to things like Google
220. 36:42Workspace and and basically making it so that Adam could have his own email address. noticed that Adam has his own
221. 36:48email address. So I gave him his own email address and then got him so he could use email and he could use Google tools.
222. 36:56And so every morning what he was supposed to do was go out and look at RSS feeds of any articles that were related in some way to the great
223. 37:06repurpose. pick the top five of that day of like whatever he found and then write them up as five stories that are related
224. 37:15to the great repurpose. Give it a little synopsis. Pull a picture from the article. Um say the key points and then say which
225. 37:25of the great repurpose stages um is this story related to. So once I got this thing working and this like the the trials and
226. 37:37tribulations of this were one of the things that you'll find out is if you work with openclaw is that there's a there's a huge
227. 37:44difference a huge difference between session memory and ongoing um structured or recorded memory. And here's what I mean by that. So, I
228. 37:59ended up um over probably a 4hour period designing this newsletter and just telling him, "Okay, I want the articles
229. 38:08numbered and I want the the headline to be whatever it is and I want the the synopsis to be this long and I want
230. 38:15there to be three bullet points and then I want you to do this." Like, I I I got it to figure out all of this. I want
231. 38:20there to be an overall pattern of the day and and an overall title of the day, right? I got it to do all that stuff.
232. 38:27And then I I had it send test emails and and I had it send a like a dozen test emails to me and then ultimately I
233. 38:35wanted those emails to go to me and Andy and Brandon. So we're working on the great repurpose. I wanted these things
234. 38:42to come to us and so it sent the first one and we're like that's great. And then I said okay so do the next one
235. 38:49tomorrow morning. It was like awesome. And then tomorrow morning came and there was no email from Adam.
236. 38:58And then I I went and I said, "Hey, can we do those emails that we did yesterday?" And he's like, "I'm not sure
237. 39:03what email you're talking about there, Skippy, but we can design an email." I'm like, "No, we spent like four hours
238. 39:09today." And he had to basically go pull them out of his log files [laughter] to understand what he had done. And then
239. 39:16and then it took me like another two weeks to sort of develop it into a cron job and a skill and a I don't think it
240. 39:22was a skill at that point. I think it was just a cron job with things. And it probably took me a week to debug the
241. 39:28cron job and get it to work. So then for I don't know three weeks Adam was sending these really nice daily emails
242. 39:36and then just one day it stopped. He just stopped. He stopped. stopped. [laughter]
243. 39:52But here's what I did. So, over on my fancy new AI computer, I've got codeex running. And so th this little dude here, this is my little
244. 40:07codeex pet that I I made him look like a little silly version of me. He just hangs out there. Um, but I came over here and
245. 40:20I made a a new chat. Let me go to the top of it here. Oh, no. Let's see. There must have been a
246. 40:35Is this the the big one? No. But anyway, it doesn't really matter.
247. 40:49I'll do a new chat. So, I came over here. So, the way the way codeex is set up, so this is set up on the same computer as my open claw.
248. 41:01So, OpenClaw has a bunch of markdown files. It has a bunch of local files so that the instructions for how to do that
249. 41:09newsletter are already sitting on this computer, right? And this is this is one of the big conceptual unlocks for me
250. 41:17about why use something like codeex or claude co-work that can write files to your computer is because all of the work
251. 41:25that you do with them exists on your computer. you own it, right? And so what I said to Codeex was, I want
252. 41:35you to go into the OpenClaw directory and I want you to find this thing called TGR signal, which was this daily email and I
253. 41:46want you to find it and I want you to resurrect it. And so I've got Codeex. Now, one of the things you can do with
254. 41:53Codeex, it's got these things called plugins, and [clears throat] you can go through and you can connect your plugins to like
255. 42:01here's all the ones I have like Chrome. It can use Chrome. It can use Gmail, right? So, Adam, I've already
256. 42:10got an Adam at atthesalon.com email and codeex can can read and manage Gmail, right? So, it can read and send emails from Gmail.
257. 42:22So, I told Codeex to go find the markdown files that were the instructions for how to do the newsletter and start doing the
258. 42:35newsletter. So, this newsletter that I just showed you Oh, wait. Are you not looking at this? I bet you're not. No,
259. 42:42you're not. Hang on a sec. Um, I got to share different
260. 43:08Yes, please. I just I just shared my whole screen, which which is good, but it's probably going to disappear soon. Okay, so in
261. 43:20codeex, what I was saying before is it's got these plugins. And so here's all the plugins I've got enabled. And if if you
262. 43:25look at this list, I've got um Gmail is one of the one of the skills it's got. Right. So basically, I just told Codeex
263. 43:34to go figure out what I was doing in OpenClaw and bring that back and turn that into a skill that it can publish um
264. 43:44directly to to to the email. And so this this email that I showed you before, this was produced by Codeex. It looks
265. 43:53exactly like the one that came out of OpenClaw because it basically just took the exact markdown file that that gave
266. 44:01the instructions for how to do this in the open claw side and it just replicated it and made its own version
267. 44:08of it. And it's not exactly the same like I think open clause like there were these synopsies were a little longer but this is fine.
268. 44:17So then what I wanted to do was I wanted to add a section in the great repurpose website called the daily signal because one of
269. 44:30the things about this email is that it's actually pretty good. Like it goes out and it finds five news articles that are
270. 44:37related to the future of work and identity being tied to jobs and like you know entry-level AI workers now need
271. 44:45senior level skills. Price Waterhouse Cooper says like that's relevant to the great repurpose, right? Microsoft's AI
272. 44:52education lead says job fear is real. Here's the fix, right? So people are starting to talk about the things in
273. 45:00in the uh in the great repurpose. But this email was only going to three people, me and Andy and Brandon, which
274. 45:08we're swell and all, but it would be nice if other people could see this. So, so I thought, wouldn't it be nice if you could add a dynamic
275. 45:18section to this site? Now, here's the problem. This site was made in lovable. I don't know how it was architected. I don't It
276. 45:30was vibecoded. It Brandon and I have been iterating and modifying this for two months now. And so it is it is
277. 45:40deeply embedded in um lovable. It is hosted on lovable. You know the domain points to our domain but it's it's just
278. 45:49in lovable. But then I wanted codeex to make this you know turn this email into a section on this website. And so I just went to codeex
279. 46:04and I said add the TGR section. So here's this big longass thing.
280. 46:25So the prompt here, okay, you've been generating TGR signal emails consistently. Now I connected my GitHub
281. 46:33to my lovable account where the TGR website is built and published from. I want to be able to create a signal
282. 46:42section of the website where the five articles. So, so I basically described what I wanted and and then I basically asked it, can you
283. 46:52do this? and it said yes, I can do this. and it went and it looked at the GitHub and it found the directory where the
284. 47:01site was and it I don't know it was doing all of its stuff and then it started making this section of the website
285. 47:10but then I didn't know how [laughter] to see it because the website is lo living over here on lovable and codeex
286. 47:19made all this HTML over here in codeex and they weren't in the same place. And so GitHub, the point of GitHub is to be
287. 47:27able to share code between projects. And so, so I went to Lovable and I said, "Hey, I've got this site over here that
288. 47:36I want to see and how do I do it?" So I basically had two different AI agents instructing me how to mediate between
289. 47:45them to have them talk to one another. And then and then I just basically kept iterating and iterating and iterating. I
290. 47:53I got it to the point that I could be in lovable and see there's a daily signal segment now and I could be in Lovable
291. 48:01and preview it. Um but it wasn't live on the website. I would have to publish it, right? And so I just [ __ ] around with
292. 48:07it and [ __ ] around with it and so this is it. Um and then I got it to publish. But but then I said, I don't want to I don't
293. 48:17want to have to every day have to come into Lovable and republish the website. And so I said, how can we do that? And
294. 48:24so Lovable came up with this scheme that did something that Codeex would would basically publish to this service
295. 48:34that would push it from GitHub to the live site. It doesn't [ __ ] matter. Like I don't I don't know any of this. I
296. 48:39don't know how it works. I don't care. But I just had him do it. But if you go to the greatrepurpose.com right now live
297. 48:54and click on the daily signal, you will see all these cards, you know, for for dates. Like here's the 24th,
298. 49:02here's the one for today, and then here's these five articles. And it does really cool stuff. Like if you roll over the stage
299. 49:13the the great repurpose stage that this article is related to it, it reminds you unhook your identity. Separate your
300. 49:19worth worth from your title, craft, task, list or workflow. AI may may be disrupting. You are more than your work
301. 49:26you used to perform. Right? So, and then if you click on one of these, reclaim your value, it'll take you to that
302. 49:31section of the TGR website. Or if you click on the article itself, it'll take you to the source article. I mean, this is this is pretty
303. 49:42sophisticated, though. We can't see it. Don't see it now. TGR, are you sure? Just dropped my screen share. Go to
304. 49:58thegreatreurpose.com. You should be able to see it. It's live.
305. 50:14Go to great repurpose and then at the top there's a a menu called the daily signal. Oh my god. Really?
306. 50:25Hang on. I don't have that many things running. Let me get rid of perplexity. Think I got to keep that running. I'll get rid of this.
307. 50:35Get rid of this. Oh, didn't see it on your screen. Ah, okay. Hang on. That's what I'm trying to do here. Okay, let me go to
308. 50:49Here we go. You should be able to see that now. So, you go to the great repurpose and then up here at the top, the daily
309. 50:56signal. Here's the thing. And if you click on any one of these cards, you get the five stories. So, you can really
310. 51:04quickly scan five stories every day. Are you serious? Oh, no. Can you update your uh post-it note there? [laughter]
311. 51:20Thank you. Um, so that's that. The other thing that I did uh in lovable was I changed out so the what you can do on the great
312. 51:34repurpose website is you can take your find your repurpose profile. So you're one of 10 different types based on sort
313. 51:41of level of you know decoupling your identity from your work, finding your purpose and then using the the AI stuff.
314. 51:48There's sort of 10 different profiles based on where you are. And so one of the things that I did was I I grouped
315. 51:55these into into groups of three and then the final one is is kind of stands on its own. Um so so this is now updated in
316. 52:02the site. Um and the daily signals here and so so what I figured out this morning is basically how to do the full automation.
317. 52:12So now it happens every morning at 5:30 Mountain time, so 7:30 Eastern time. The email is sent to Brandon and Andy and I,
318. 52:22and then it writes JSON, you know, into GitHub and publishes it to this service that then automatically pushes it to the
319. 52:32live site. So starting tomorrow morning, um this is now a daily updated um section of the site. So, if I were you,
320. 52:41I would bookmark this this page, the uh the daily signal page, and every day you'll have like five articles you can
321. 52:49look at that are related to this idea of everything's about to change, and what the [ __ ] are we going to do about it? [laughter] Um,
322. 52:59okay. So, that's that. [clears throat] Let me come back here. So, let me see if there's any questions.
323. 53:08Didn't see it on your screen. We believe you there. Okay. Last night I finally had two agents talk to each other via
324. 53:13voice. Awesome. This may be my main AI could help me make sure the second newer AI was set up correctly. Yeah,
325. 53:26that's pretty cool, Kyle. Great share. Thank you. Um, the other thing that I did with Codeex this week, and there's there's not a ton
326. 53:40to show here. Well, there's a little bit to show. I can go to So, I can go to storyvine.com. Let me share my screen.
327. 53:55Share this tab instead. Oh, stop sharing. Damn it. Okay, so this is our website. This is what it's looked like for for a bunch of
328. 54:11years. Um, we wanted to make this thing visible to large language models. And in order to do that, you have to create
329. 54:21what's called an LL LLMS.txt file. The problem with our website is that our developers 10 years ago, against my better wishes,
330. 54:35against my request, my direct request, instead of making this a site that we could edit, they built it on Cloud
331. 54:43Cannon, which if you're a developer, you might know what that is, but it's basically Cloud Cannon invented its own
332. 54:49kind of HTML. It's sort of quasi HTML that can apparently do [ __ ] that HTML can't. But it's impossible to update without a developer.
333. 55:01And I had had Claude analyze the website and come up with a whole strategy for how to make our site discoverable in an LLM. Um,
334. 55:13and our developers have been busy. We've our clients are asking them to build stuff. So, they don't have time to do
335. 55:18this. And so I thought against my better judgment, I am going to point Codeex at my website and I'm going to give it
336. 55:27all of the strategy that Claude created and all the documents that it made to go do this. Um, and I'm going to have it update my
337. 55:37website. And so it started churning and churning and churning. And then I had to log into Cloud Cannon and I had to log
338. 55:43into Webflow because there's a different part of our sites on Webflow. And then I just started watching it,
339. 55:50pulling up HTML file after HTML file that are uneditable by me and started editing them. And then and then it was
340. 55:59saying, I've made changes to your website. Do you want to do you want to commit those? I'm like, oh god, this is
341. 56:05stressful because I knew that what Codeex was going to do was completely blow up my website. Right? And it just
342. 56:12went it churned and it turned and it turned. And so I am happy to report if I if I type in storyline.comllms.txt that my friends
343. 56:27is a markdown file that tells a large language model what storyvine is. Isn't that cool? How to get in touch
344. 56:38with us. Storybine is a guided video platform that enables anyone anywhere to capture and create polished authentic
345. 56:45videos stories in minutes. And so that's all there now. So it did whatever it needed to do on the HTML
346. 56:53side to make all this [ __ ] discoverable. And uh yeah, so that's pretty groovy, isn't it? Actually, you know what we could go do
347. 57:03is I could go to chat GPT. Chat GPT. We have a song from Claire in the AI
348. 57:18salon. llms.txt. Yes. So basically, Chef Kelly, if you go to So here's here's how I learned to do this. I went to Claude.
349. 57:35[laughter] So, so my co-founder Mon'nique sent me a text at 8 o'clock in the morning, a Slack that said, um, "Hey Kyle,
350. 57:45is it possible to make Storyvine show up in like when someone asks for, you know, authentic automated video
351. 57:52in chat GPT? Is it possible to make it show up? And and I said, I think there's a thingy that does that. And so I went
352. 58:01to Claude and I said, "Here's my website and and here's our target audience and here's like I gave it a bunch of context
353. 58:09about our website." And then I said, "I know there's some sort of thingy that's like SEO, but it's for it's for large
354. 58:16language models, but I don't know what it is. Can you tell me what that is?" and it just gave me this [ __ ] baller
355. 58:24explanation of what it was. And then I said, "Can you turn that into um a a basically um a recommendation document
356. 58:33that I can give to my developers?" And so, not only did it give me the strategy, but it wrote all the documents. It went and it found all of
357. 58:41our websites and it found our blog and it found all these issues and it created this 18page document that had the LLM like that that text
358. 58:52file that I just showed you that was written by Claude after researching all of my website. So, I literally didn't
359. 58:58know [ __ ] about it other than that it it existed. And then I went and had Claude just do all the stuff. And then that's
360. 59:05that 18page document along with three of the files that it created are what I uploaded to codeex. And I said, "Okay,
361. 59:14I've got my Cloud Cannon login and I've got my Web Flow login. I don't want to do [ __ ] I want you to do it all. Can
362. 59:22you do it?" And it said, "Yes." And then 45 minutes later it was done. Um, so let's try. Can you still see
363. 59:30this? Is a question. Oh my god, it's a little miracle. Okay. Um, I want a company. How about this? I heard
364. 59:45about a company. company that can do authentic
365. 1:00:00user generated video for pharma companies.
366. 1:00:14Let's see if Let's see if it finds anything. That sounds very likely to be Story Vine, especially given the pharma
367. 1:00:25authentic thing. You're probably thinking of StoryVine. Storyvine helps pharma companies capture real authentic
368. 1:00:31video stories from patient patients, HCPs, reps, employees, or advocates, but in a structured, brand safe, reviewable
369. 1:00:38way. That's our language right there. It sits between expensive produ professional production and raw UGC.
370. 1:00:47Chaotic raw UGC. The public StoryVine language lines up almost exactly. The mechanism is the important part.
371. 1:00:54Storyine uses a guided app that walks people through the recording process to answer specific prompts and then the
372. 1:01:00platform automatically edits the responses into polished videos with graphics, music, transitions, and a
373. 1:01:06dashboard for approval publishing. A Kyle friendly way to say it. This is not random Tik Tok style style video. This
374. 1:01:13is not a $75,000 video shoot. Authenticity with guard rails. So there you go. There's also public evidence of pharma
375. 1:01:25scale. An Oaks Lab case study says Novartis generated more than 5,000 videos in the first six months using Storybind. That's true.
376. 1:01:36So I guess it worked. There's that. Okay, now let's go back to codeex for a second. Change my share.
377. 1:01:47Any questions? Vicky does Tik Toks about LLM's finding your business and recommending. Cool. Now, how does the LLM get you to
378. 1:01:57pay for the boost? Well, they they have advertis I got an email today from OpenAI saying, "Hey, do you want to
379. 1:02:03advertise on OpenAI on chat GPT? Click click here to start to learn about advertising." So you can they're they're doing advertising.
380. 1:02:13I don't know how long that's going to last, but you know. Oh, let me go to the live channel for for a song.
381. 1:02:28Share screen. I'm going to do entire screen. I'm going to turn on audio. Hopefully, this is going to last for
382. 1:02:34more than a minute. I'm going to go to the salon. I'm going to go to Lab Live. Where is it? Chat.
383. 1:02:54Crazy indeed. For all you crazy coders out there, like Kyle, because you're not alone in the code space.
384. 1:03:13All right. What's going on here? So just raised a buttload of money. [music]
385. 1:03:32You are not alone in [singing] the cold base. [music] There's a light on in the corner of a room [singing] nobody sees. [music] And
386. 1:03:46the coffee's gone cold beside you. Like a small abandoned [singing] plea. You've been tracing ghost through functions, [music]
387. 1:03:56leaving comments like a prayer. Trying to fix what keeps on breaking. Wondering if anyone's there. But every line [singing] you're reading
388. 1:04:09was [music] touched by someone's hands. Someone else was [singing] lost here once. Trying hard. You are not alone [music] in the cold.
389. 1:04:22Not alone. [singing] [music] >> That's so good. Claire. Claire coming in hot with the tearjerker for the coders.
390. 1:04:31I like it. That's really good. Um, okay. So, now I want to show you something. Let's see.
391. 1:04:54Did I do the image in here?
392. 1:05:10Okay. Um, let me come back here for a second. Okay.
393. 1:05:28the way these memory systems work. So whether you're talking about Open Claw or Hermes or Codeex or Claude Co-work and Claude
394. 1:05:43Code, Perplexity Computer, Gemini, Anti-gravity, all of them
395. 1:05:58when they're creating either skills or memories or system prompts, they're writing them in MD files. And MD
396. 1:06:11stands for markdown. And a markdown file is simply a text file that's got some specific formatting in it. Like it's got, you know, hashtags
397. 1:06:23in front of text to make it bold or whatever it is. It's It's just got some specific markdown code that allows the
398. 1:06:32document to be structured, but they're just text files. And you can write those text files however you want. And Enthropic came up with
399. 1:06:44these things called skills. And skills are just a specific format for for these text files, these markdown files. and it
400. 1:06:54says a skill, you know, will have a description of what the skill is and then it'll have here's what it can be used for and then it'll have
401. 1:07:00descriptions of what the skill does and then it'll have a sequence of whatever. It's just it's the skills. It's all the [ __ ] that allows
402. 1:07:11the agent to be able to do that skill. So if the skill is write a business email in the format that StoryVine uses,
403. 1:07:21I could create a skill that says, you know, here are example emails, like here's a directory with example emails
404. 1:07:27in it. Here's a directory with logos in it. You know, here's a directory of, you know, um, I don't know, whatever else
405. 1:07:34salutations that we use. And that skill file would be able to point to all those locations and tell the agent exactly
406. 1:07:42what to do to be able to write a good story by email. I'm so late. Sorry. We we've been through a lot. We've been through a lot tonight. Um
407. 1:07:57the good news about that approach is that it's very simple. You can take those text files like the the text file on how to do that email
408. 1:08:11for the TGR signal was just a markdown. It was text file and I told Codeex go read that text file and then make your own skill
409. 1:08:22for that, right? It didn't have to there was no API. It was just it just went and read a text file. That's also the downside because
410. 1:08:34how Gareth might organize and mark his markdown files might be different than how his Jarvis agent does
411. 1:08:43it, might be different than how Codeex does it, might be different than how Claude Code does it. And so about a month and a half ago,
412. 1:08:52month month and a half ago, Andre Karpathy came out with this thing about creating a brain for his agent. And what it basically did was it
413. 1:09:02described a structure like a folder structure. Like the thing that's amazing to me about these AI systems that are
414. 1:09:11doing all this sophisticated work is it's like here's a couple of folders with a couple of text files in it. And
415. 1:09:18what's in those text files is just prompts. It's just [ __ ] that the LLM reads and it basically gives it context to be able to go do [ __ ]
416. 1:09:29So, incredibly sophisticated behavior with just a couple of folders and a couple of text files. What Google did last week
417. 1:09:39was they created version 0.1 of what they're calling OKF open knowledge. What was the F again format?
418. 1:10:20I'll share my screen in a second here. Okay. So,
419. 1:10:54okay. So you can go find if you
420. 1:11:11where can I put this? I can put this. We're gonna put a hyperlink. Brandon format. Well, there's the there's the link. Oh,
421. 1:11:26an error occurred in the chat. Maybe it didn't like my URL. Just go search for Google open knowledge
422. 1:11:40format. Um, okay. And so here's basically what it is.
423. 1:11:56Open knowledge format is like giving every important piece of knowledge a baseball card. On the front of the card,
424. 1:12:03there's YAML something markup language. And I don't care what what it stands for, but it basically says here's the type of
425. 1:12:15thing it is. Here's the title of it. Here's the description. Here's the tags. And here's the timestamp. So this little
426. 1:12:25they call it front matter. Basically what Google has done is they took Andre Karpathy's idea and they said hey instead of just
427. 1:12:35creating randomly named markdown files. Why don't we come up with a standard format that at the top of every markdown
428. 1:12:45file that's going to be used by an agent, we'll put a little bit of metadata that explains to all the other
429. 1:12:52systems what this document actually is. Um, and then there's the content of the card. And let's see, back of the card, markdown, definition,
430. 1:13:04source of truth, related concepts. So like here's connections to other markdown documents, right? So that's the
431. 1:13:11whole vibe. So basically all the open markup, all the open knowledge format is is basically saying, "Hey, we're going
432. 1:13:20to set a standard for how you tag your markdown files." And so what I did was I went into
433. 1:13:41I went into Codeex. This is Obsidian. I went into Codeex. And so I've been working with with Codeex to create a memory system based
434. 1:13:57on Andre Carpathy's thing. And it it Codeex named it agent brain. And that's it's fine by me. But basically what I
435. 1:14:05said was, I want you to go read the open knowledge format documentation and see if you think that's something
436. 1:14:13that would be good to add to what we've already built or should should we replace what we've already built. And so
437. 1:14:19it basically, you know, went and analyzed it and said, "Yeah, I think we should do it." And it came up with, you
438. 1:14:26know, this this is the little tag the the format of the tags that it's going to put in all my markdown files. Um,
439. 1:14:36and then it's telling me, you know, I've got all these different markdown files. I've got a readme file, an agents, a
440. 1:14:42context, an access policy. This is all stuff that I've been developing with it. [clears throat] Um,
441. 1:14:58so like agent brain. So this is where my memory lives. And then underneath it, I've got agents.md. That's an existing
442. 1:15:05agent bootloadader. The bootloadader is here's the basic information about Kyle Shannon. When when an agent wakes up, it
443. 1:15:13knows nothing. And the agents.md in the case of codeex, the first thing it will do is go look at that agents.md file and
444. 1:15:21it'll say here's who you're working for. Here's all the things you need to know about Kyle. Here's his memory. Here's
445. 1:15:26his skills. Here's all the the things, right? Um, it put all this information in Obsidian for me. It's got a global memory. Um, I
446. 1:15:39don't know. It's got a bunch of other [ __ ] it does. Okay. And then it it basically tells me where everything is. I asked it to teach me
447. 1:15:49how to use how to use obsidian, and I said, I don't think I like obsidian. It said it doesn't really matter if you
448. 1:15:55like it or not. It doesn't need obsidian. The only thing obsidian does, if you want me to show you obsidian. Um,
449. 1:16:05where's Obsidian? Here it is. So, I can Can I do command plus on this? Yeah. So, what Obsidian is is like this graph
450. 1:16:20of all of my ideas and how they interrelate to one another. And so, I can click on any one of these and it'll take me to that document.
451. 1:16:33And I can go back out to the big map. Command minus. I don't know why it's zooming infinitely. Huh. Something's wrong here.
452. 1:16:51Well, whatever. It's broken. Um, Obsidian is Notion for people who like markdown files. There you heard it from producer Brandon first.
453. 1:17:04Why can I not see? Oh, there we go. So like there's the AI learning lab wiki and then it says what it is. So this is
454. 1:17:16the thing that Lord digital gods made and this is now a source inside my codeex right in inside the memory
455. 1:17:28system. Like one of the things that's in there is hey there's this cool resource out there that's all your transcripts in
456. 1:17:34this really accessible wiki. like that's one piece of information in my otherwise, you know, crazy crazy straw brain for my agent to be
457. 1:17:47able to use. Um, like I could say, um, hey, go look at the AI salon
458. 1:18:02wiki and find the times where I said technology was wait that I said
459. 1:18:18we should embrace the jank. And in theory, this is going to go off, look at the wiki, and come back with some some links.
460. 1:18:38SOS, share your screen.
461. 1:18:55searching the web. There's a salonwiki.com. So, it's finding some things. I found three episodes.
462. 1:19:13So, anyway, it's off doing that. Um, I'm just going to stop it there because that's kind of boring to watch
463. 1:19:18it search. But let me come back over here. We have a few minutes left. We have 10 minutes left. Why am I sharing all this?
464. 1:19:31Makes a chart like Alice's on the Lord. Yeah, it does. You can make money with Chad GBT. Okay, here's why this is important.
465. 1:19:56The thing that's keeping me up at night with where AI is going and where the future of work is going
466. 1:20:11is that there is very little there's very little that I can see that is digital that's not going to be completely commoditized
467. 1:20:27meaning anything you produce digitally anything a book a powerpoint a workshop outline a song a picture whatever if they're at all any
468. 1:20:43good, they're going to be copied instantly by agents. Agents are going to take over the knowledge part of the
469. 1:20:49internet. And so it's going to be almost impossible to get your outputs to stand above the noise. And
470. 1:21:04it's it's this weird thing to process. But what I keep coming back to is the thing that will be of value is one of
471. 1:21:14those outputs that you can connect to you, right? You are the only thing in the entire [ __ ] chain that is unique.
472. 1:21:32And so part of repurposing yourself is to to really understand, well, who am I? What do I care about? What am I passionate about? Because if
473. 1:21:44you can get in touch with what you're passionate about, other people will be attracted to that energy, right?
474. 1:21:56like it's exciting to be in the presence of someone who's into something. And so if we can understand that piece and then we can say, "Hey,
475. 1:22:10there's this collection of text files that we can actually have these agentic things help us build." I didn't have to
476. 1:22:18write any of that [ __ ] Codeex has produced all of that stuff I showed. I'm just telling Codeex, go find this thing.
477. 1:22:26Go find that thing. You figure out how to store it so that you understand all of what I'm doing. You understand what
478. 1:22:33Sydney is. You understand where the files are on SoundCloud. You understand that my writer is Andrew Watts and my
479. 1:22:40producers are Amy and Kathleen. And you understand here's the people working on the AI salon and here's my co-founder
480. 1:22:47for StoryVine and here's what StoryVine is. All I have to do is tell it who I am and what I'm working on and I say, "You
481. 1:22:54go figure the rest out and write it in these files." And the reason that I went down the the OKF model, the Google
482. 1:23:03knowledge format model was ideally because this is probably going to take me months to really put together a formidable
483. 1:23:16knowledge graph of who I am. in a way that makes these agents work like I want them to work. I only want to do that once.
484. 1:23:28And what the what the knowledger graph format has done that Google has done is it's basically making a standard so that
485. 1:23:35you can take that repository of data and any AI tool any AI agent if it follows that format will be able to use it.
486. 1:23:47So I won't have to redo my memory system every time I use a different tool, which is effectively what we have to do now. Right?
487. 1:23:57Why I don't use Claude is because I've been using Chat GPT for three and a half years and I don't feel like retraining
488. 1:24:02Claude unless there's some really specific thing I know Claude can do better than Chat GPT, then I'll go use it.
489. 1:24:10But I don't just freely use it because my my memory is over here. So like one of the things I want to do is I want to
490. 1:24:17download my chat GPT history and give that to codeex and say figure out what in here is valuable and put that in the thing
491. 1:24:31right and then once that thing is built that data that representation of me is not copyable because it's me.
492. 1:24:43So when this thing produces something, I can look at it and go, "Yeah, that's mine. Here's what that means. Here's why
493. 1:24:52that's exciting to me." And if you're excited about what I'm up to, then you'll be excited about my output. My output then doesn't have to
494. 1:25:01compete with all the other outputs, which are just valueless because they're just outputs. But if this one is Kyle's output and
495. 1:25:09Kyle talked about it, oh, that one's different. That output's different because it's connected to this guy that I trust.
496. 1:25:17Make sense? But it won't replace community. I think it I think what this doing is doing, Mel. Yeah, it it can
497. 1:25:24help us connect deeper. I I think what this does is it it it actually connects us. It allows me to use AI tools to
498. 1:25:32produce stuff that I can go be in the world with. But in the end, it's about me connecting with all of you and you
499. 1:25:40connecting with each other and us all connecting with each other. And then at some point, Gareth says, "I made this
500. 1:25:46cool thing." Or Lord Digital God says, "I made this cool wiki. You want to see it?" Well, yeah. If LDG made it, I want to see it because
501. 1:25:58I value his code. No, because LDG is awesome and I I know that if he did it, it will be awesome. Like I know that
502. 1:26:12he doesn't have to market it, right? So that's where the great repurpose is going. I'm going to lead the charge
503. 1:26:27blindly. I'm likely I I think I have LDG and Gareth on this call right now. I will likely recruit them to come teach a a guest session.
504. 1:26:41But anyone who's in the great repurpose mastermind, we're going to start building these things together.
505. 1:26:54We're going to figure out how do I go get the source materials? How do I organize them? What do I do? I use Claude Co-work. I use Codeex. I
506. 1:27:02use OpenClaw. I use Hermes. I don't [ __ ] know. We're going to figure it out together. There are people in this
507. 1:27:08community that know parts of this. Some people know deeply technical parts. Some people are going to know philosophical
508. 1:27:15parts. Cam Ken might understand, you know, digital entities, right? And sovereignty of of these digital entities. And that's got
509. 1:27:26implications for what these things are. ChatgPT or your main AI will be your orchestrator from your knowledge base to
510. 1:27:34your other AIS. Yeah, there there will we will have some central orchestrator. I'm I'm betting on codeex. The reason
511. 1:27:43I'm using codeex is that I think chat GPT is ultimately going to go away and codeex is ultimately going to go away
512. 1:27:50and it's going to turn into chat GPT 2.0 know is going to be this orchestrator layer and then we'll connect all of our
513. 1:27:58data to it and we'll have this digital Kyle. In my case, I'll have this digital Kyle. Here's the exciting thing also.
514. 1:28:05How we doing on time? Two minutes. If I've got a digital Kyle that's well put together and uses this standard and
515. 1:28:14Kelly Camp has her Kelly Camp and it's put together. We can connect those two and we can say, "Hey, let's have those two entities
516. 1:28:25collaborate and make a thing that takes the best of your thinking and the best of my thinking and mashes that together." Right?
517. 1:28:35There's all sorts of implications of if we get this digital self right. So, the great repurpose is now going to be
518. 1:28:41these, it's always been these two halves, but the AI half was always really nebulous. And I could not get my
519. 1:28:47head around what the [ __ ] do you teach if the tools keep changing. The only thing we need to teach is what
520. 1:28:57is the what are the digital bits pieces that can represent who you are. And that's about your history and your
521. 1:29:06opinions and where you are now and where you were then. All that sort of stuff. Anyway, all right. Tomorrow is Thursday. We've got the AI
522. 1:29:16salon mastermind practice lab. So, if you're in the mastermind, get your butt to the practice lab and we will talk
523. 1:29:23about how you're doing your AI stuff and how we're doing with it, how we're all doing with it. Um, Friday, we've got office hours.
524. 1:29:32We've got AI for all minds. Um, and then we got Friday night date night, which is an ask me anything. So, I vomited a lot
525. 1:29:40of stuff tonight. I showed a little bit. I didn't show a ton. Hopefully, it was a little bit of value. It's hard to show
526. 1:29:47these agents because the stuff that I showed was just kind of the output of it churning for 45 minutes with me
527. 1:29:55occasionally going, "No, go that way. No, do that." There's not a lot to show, right? You can't. There's even if I
528. 1:30:03played the time-lapse movie of it, it would be boring as [ __ ] So, but I but I hope you can you're you're starting to
529. 1:30:09get that these things are these sophisticated multi-layered systems. So, anyway, so tomorrow practice lab, Friday office
530. 1:30:20hours at 11:00 a.m. Mountain time and then Friday night date night, ask me anything. So, think about some questions
531. 1:30:26for Friday. Um, it does look like Claude Fable came back to AWS in some form. someone said tonight. So, I don't know
532. 1:30:37if we're going to get Fable back. I kind of doubt it. I doubt that we're going to get anything from OpenAI tomorrow, but
533. 1:30:43who the [ __ ] knows? Um, and I don't really care anymore. Like, like this this thing about creating your digital self.
534. 1:30:54This feels really important. Like, it feels as important as those early days figuring out what the [ __ ] these tools
535. 1:31:00were. figuring out this digital self piece in a way that's aligned with who you are as a human. I feel like if we can crack
536. 1:31:10that, if you're in the mastermind and you want to actually [ __ ] nail that, that I I just like I can't overstate how
537. 1:31:17important I think it is. Like I actually think this is really important for all of us to understand because it is the ultimate way to claim
538. 1:31:28your agency. Right? One way to claim your agency is say, "I am not my job. And I don't give a [ __ ] if you take the
539. 1:31:34tasks away. I am who I am. Here's my value. Here's why I matter." And if we can do the same [ __ ] thing with the
540. 1:31:40digital side and marry those two together, that's an unstoppable force. And it's independent of any of the
541. 1:31:48technology. So, [ __ ] Fable. [ __ ] OpenAI 5.6. None of it matters. The only thing that matters is you and how you present
542. 1:31:57yourself to these agents. So they can go do the [ __ ] work for you. So you can go grow blueberries and hang out at the beach. Okay, good.
543. 1:32:11[laughter] All right, everybody. Think I blew a blood vessel in my temple. Okay, [laughter] that's where we're headed.
544. 1:32:28You matter. You matter. You all matter. I matter. And we're going to figure out how to reclaim our agency and let all this
545. 1:32:38chaos re around us and and we're going to be able to sit in the pocket of the storm, the eye of the hurricane,
546. 1:32:45and just pull in technologies as necessary that will allow us to do what we want to do and then do whatever we want to do.
547. 1:32:53And we'll be able to do it sooner than anyone else. and we'll bring them along. All right, cool. I'll see you Friday. See you
548. 1:33:01tomorrow. If you're in the mastermind, come. Come. Even if you haven't been in a while, come, please. All right, bye.