7/29/2026 - Inside Jack Dorsey's Buzz: The Future of Human and AI Workspace Collaboration

Transcript

1. 0:00Well, [music] [music] [music] hey.
2. 0:16[music] [music] >> [music]
3. 0:33>> And we're live. Welcome to the AI Learning Lab, everybody. Welcome to the AI Salon. Um, got a bunch of people in
4. 0:40here. So, here's the deal. If you're new here, welcome. If you're new to the AI salon, welcome. If you're new to the AI
5. 0:46learning lab, that's what this is. Um, in this upper corner right here, if you click underneath the or if you click on
6. 0:51the logo, underneath it is a button that'll pop out the chat. And so you can Oh, Roso Shay is here. How cool. What's
7. 0:59happening, Ross? We're talking buzz tonight. So, that's what we're gonna be talking about in humans 2.0 in our
8. 1:06little our little subre thing. Um, that's exciting. Linda Sherman's here. We got uh Silverf Fox in the house. We got Sharon Crawford.
9. 1:18Um, can't hear anyone else. Can y'all hear me? Are my words making sound? I can hear you. All right. It sounds
10. 1:32like How do I sign it? That must be you, Sharon. Wait. Sh H A R O N. [laughter] Sh
11. 1:46I don't know how to Sharon. Uh uh. Uh, we wait. We can't uh here. Wait. He here. Wait. Here.
12. 2:05Wait. [laughter] Damn it. You. You. [laughter] No, you can't hear us. Okay. Um, so welcome. If you're new to the AI salon, here's what
13. 2:18I suggest that you do when you when you first sign up. So, here's the deal. The way the salon's set up right now, you
14. 2:23don't have to to sign up. You can just see all the stuff down the side. Um, I encourage you to sign up. One of the
15. 2:29things that happens when you sign up, you put in your email or whatever. Um, it gives you a little sixstep
16. 2:34onboarding. Like, here's the six things to do, six or seven things to do. And like one of them is introduce yourself.
17. 2:40Um, the AI salon is is about Okay, Sharon Crawford can hear us now. Um, the AI salon is is about people that are
18. 2:47curious about what's going on with AI and not like, you know, what is it technically and what are the weights and
19. 2:53what are the sciences and this and that. No, more like how do we as human beings use these incredibly powerful tools and
20. 3:00how do we use it in a way that amplifies who we are and what we do. That's what the salon's about. That's what the AI
21. 3:07learning lab's about. Um, and it's a really remarkable group of people. So down the side you'll see all these
22. 3:13different spaces. Think of those as just like little huddle rooms where different people of different interests, you know,
23. 3:19hang out. Um there's all sorts of events. In the upper leftand corner um is is an events tab. You can see all the
24. 3:27events that are there. You can see these lives. You can see workshops. We have these things called LOL's that Vicky
25. 3:32Baptiste runs that are um learn out louds where people from the community are just generously teaching other
26. 3:38people from the community. It's really remarkable. So check it out. Now if you want to there if you want to do like a sophisticated
27. 3:47cool thing tonight down in um this corner you can pop out this video as a pop out window and then you can open a second tab in your browser
28. 4:00and have the AI salon there. So if you want to be checking out the AI salon while you've got this this stream running you can do that. Okay.
29. 4:09Beautiful, beautiful, beautiful. Welcome. So again, everybody up here, if you click up there, it'll pop out the
30. 4:15chat. Please say hi. We've got we've got 38 people in here tonight. So awesome. Welcome. Please say hi. If you haven't
31. 4:22said hi yet, please say hello. Tonight we're going to talk about Buzz. Buzz. If you haven't heard of Buzz, I'm gonna [laughter]
32. 4:33Okay, here a couple of caveats for tonight. When I go deep on a tool, it's probably going to be boring.
33. 4:43I try to make it not boring, but it's but it's when you go deep, you know, when you're not surface level,
34. 4:50um it's probably going to be boring. Well, yeah. Welcome, Sarah. Beautiful. Um, hang on. Got one question.
35. 5:07Um. [laughter] Uh oh. What did I do? Beautiful. Okay, cool. Um, hang on one second. Okay.
36. 5:22So, it's probably going to be a little boring, but what I'm going to try to do tonight is tell you enough about the architecture
37. 5:32and the technological pieces of it so that you understand why it's different technologically, but more important, what I'm going to
38. 5:41try to do is talk about why I think it's important. And so I think what I'm going to do tonight is I'm going to start with just
39. 5:50talking a little bit about what I think what I think is coming with the future of work. Right now in 2026, in mid 2026,
40. 6:03work essentially looks like what work has always looked like. There's some decent swath of people have been laid
41. 6:13off because of AI, but what it looks like is kind of what it's looked like, right? You get a job, you give a resume,
42. 6:21someone hires you for a specific role, you go in there, you do that. I think within five years and potentially three, work is going to look
43. 6:31very, very different. And the reason for that is that AI is going to increasingly automate the tasks of what we do.
44. 6:44Right? In today's world, you get hired because you can do tasks that generate outputs and those outputs make the business, right?
45. 7:00[laughter] Be like Brandon. We got a bunch of new people in here. By the way, if you're new here, go up here. Click click right
46. 7:06here. It'll pop out the the chat and just in in the chat, please say hello. We'll bring your name up there.
47. 7:13Mindware, the new term I learned in the past few days. Hey. Hey, Foxy. Okay. So, so to your point, Ross, so right now,
48. 7:23it's kind of independent of AI. What we get paid for is trading our time for generating outputs. Right? Time equals
49. 7:32money. Punch in, punch out. Right? We're living in a post-industrial revolution model of economic value. You give us your time, you do these
50. 7:46tasks, you generate these outputs, and we'll give you money. Right? Tasks equal money. And in America certainly I think
51. 7:56this is less true you know in different parts of the world but certainly in America when we meet each other at a party what
52. 8:04we ask is what do you do what are your tasks what's your degree what have you perfected what outputs do you make I make powerpoints
53. 8:16at Coca-Cola that's a big company I bet they pay you good money for those poweroints points, right? As AI automates more and more of the
54. 8:28tasks, people are going to be left with like, wait a minute, what's my value? I went to school for this thing that is now valueless.
55. 8:37Yes. Is the answer. And so what's going to happen over the next three to five years? The shift is going to go from valuing the tasks that
56. 8:48you used to do to valuing who you are. And our ability to understand and articulate who we are and what we bring
57. 9:01to the party is going to be the primary reason why people want to work with you. Right? It used to be I'd work with Jim,
58. 9:11who's a flaming but he was really good at at After Effects, and we really need good motion graphics. And
59. 9:18Jim's great at After Effects, but he's a flaming Well, now I can just make motion graphics, and I can choose not to work
60. 9:28with Jim because the only thing that Jim brings to the party is arrogance and conceit and condescension.
61. 9:35And so in this new future, I don't have to work with assho. It's a reckoning for is what AI is doing. So we're
62. 9:44gonna have to be on our best behavior. And you're going to end up working with people that you trust.
63. 9:50And like I'm going to work with, I don't know, Valerie Cox or Silver Fox. Like I know that Silver Fox is just she's got a
64. 9:58a sharp wit. She always shows up. She's really reliable. She's incredibly sweet. She's incredibly caring. She's compassionate.
65. 10:07I'm going to assume if Silver Fox is AI literate, which she is, she hangs out here a lot. She understands what this stuff makes
66. 10:16possible. But I'm going to work with her because of who she is. We then get to build things
67. 10:33because AI can do so much of the execution. We can just build things that kind of reflect who we are. Like like there's this guy um his name
68. 10:47is Price and I talked about him I think I talked to him about him on Monday. I'm going to try to get him to come speak at
69. 10:53the salon. And Price is a roofer and he loves the game Railroad Tycoon. And he thought, wouldn't it be cool if I
70. 11:03could turn my CRM of my roofing business into a tycoon simulator where I could actually play my business, right?
71. 11:15That's a really cool thing. And so we're entering a world where we can just say, "Hey, I have an idea." and go build that idea.
72. 11:25But but typical corporate structures are going to start to crumble and get weak and feel weird and like what jobs you
73. 11:32do, right? And a lot of people are going to be laid off. And then those people that are laid off are going to be like,
74. 11:38"Well, wait, I've do this AI stuff and and there's so much to learn." And so one of the visions that I've had
75. 11:46is that imagine being able to take So you're in the AI salon, right? And you're meeting people and and Rasho Shay
76. 11:53comes in and Mr. It is here and and Steo is here. Hey, Steo. Um, lucky for me, I don't work anymore. Yeah, Steo, he he
77. 12:03just wants to do cool images and he doesn't work anymore, but he's just he's just a a fun guy to be around. and he's
78. 12:08got a great sense of humor and he's willing to just jump in and do silly stuff, right? So imagine a future where rather than
79. 12:16saying I have to work with someone because they've got these skills or I need to hire for those skills, imagine
80. 12:21just being able to get together with people that you know and trust and say, "Hey, what if we got together and it
81. 12:29doesn't matter where we live physically, but we get together in a space where we can collaborate on projects
82. 12:39and we can learn how to use AI agents in conjunction with our ideas. We can just say to the agents, "Hey, agent
83. 12:49that's really good at building software, go build us some software." Hey, agent that's really good at research, go off
84. 12:55and research something for us. Hey, agent that's really good at marketing, now that we've built this thing, can you help us market it?
85. 13:03And imagine that group being able to just do project after project after project. And there's individual projects
86. 13:10and there's collaborative projects and there's group projects. And imagine that sitting in an infrastructure
87. 13:21where you can actually start making money with that thing. It could start to look like the future of business. It could start to look like
88. 13:31the future of work, right? You could start out like a typical business. You could say, "Okay, we want to do a gardening
89. 13:41business and we're going to get into this space and we're going to talk about this gardening business." But you could
90. 13:47also get in there and just say, "I want to get into a space with a group of people that I trust and respect
91. 13:55and I want to start building stuff." What are you going to build? We don't know. But what if the goal was we want to have these agentic
92. 14:06things take our ideas that we're passionate about and go off and build them and go off and make us money. Like maybe you're Cindy
93. 14:22and you've got this publishing what's it called? uh uh not label labels music publishing a publishing house and
94. 14:32she's got all these book published and some of them she wrote and some of them other people wrote and they're just
95. 14:37sitting there and she can put in as much time as she can put in but she's got other things she's doing let's see
96. 14:48daily I'm amazed at the projects I'm working on and the people I'm working with four years ago none of this would
97. 14:53have been fathomable to me yeah So imagine there's 52 people here. So welcome to the new people. If you're new
98. 15:00to the AI salon, welcome. This is an amazing amazing group of people. If you want to chat in the in the comments, if
99. 15:07you want to if you want to comment and chat um up in this corner, if you click on that, it'll pop out the the chat
100. 15:13window. So please say hello. Hey Danielle. Um and then and then in between listening this, go check out the
101. 15:20AI salon, check out the different spaces and things there. And I would encourage you to join. Um, go ahead and join and
102. 15:26it'll give you a little we'll give you a little onboarding sort of a six or sevenst step thing to to onboard and get
103. 15:32up to speed on what the salon's about. But imagine that. So about a month ago, I was thinking um I bet with the talent in the AI salon,
104. 15:49we could probably get some people together that understand business and understand technology and understand
105. 15:54this and understand that and we could probably brainstorm what that infrastructure might look like. And then three days ago, Jack Dorsey
106. 16:07launched Buzz And with one major exception, Buzz is everything I was thinking about building. And I'll talk about the
107. 16:17exception because I'm going to go build it. On its surface, Buzz looks like Slack. It looks like a Slack competitor. If you
108. 16:28don't know Slack, Slack's just it's where people in business communicate with one another. You've it's kind of
109. 16:33like the AI salon. You have little different channels down the side. So, like in our StoryVine Slack, we've got,
110. 16:40you know, account and delivery channel. We've got a development channel. We've got a water cooler where you can talk
111. 16:45about nonwork stuff. I've got individual chats with my co-founder and with our sales guy, right? You have all these
112. 16:52different channels and you go in those channels and some people can see some channels and they can't see others,
113. 16:57right? So it's a permissionbased conversation space where you can upload files, you can trade files. So it's kind of like that. So on the
114. 17:08surface, you look at it and you're like, cool. So here's what we're going to do. I am going to I like our banner. If you like this
115. 17:20content, you'll love what else is happening inside the AI salon. Join our vibrant community of AI adventurers for
116. 17:26free. introduce yourself and join the conversation. That's beautiful. Brandon, did you write that? I like it. I hated
117. 17:34Slack. I know. I listen listen. Slack ain't Slack ain't great, but it's fine. Like Slack is is serviceable, but
118. 17:44there Slack has some major major issu issues. The primary of which is that your account and all of your data belongs to them.
119. 18:03So if Slack for whatever reason decides that you have violated their community terms or done something bad or you
120. 18:11haven't paid your bill and they shut down your account, it's all gone. Your history, all the context, all the all
121. 18:18the accounts, it's all gone. So that's that's a major issue with Slack. The other thing about Slack is that it's for people.
122. 18:27You're like, "Yeah." Well, we live in a world now, not in the future. We live in a world now where we've got these things called agents
123. 18:38that kind of have people like qualities. You can talk to them, they can respond, and you can ask them to do and they will go do it.
124. 18:50So, what is buzz? Okay, I can't find the chat. The chat is you click underneath the AI learning lab logo. Boom. Wait,
125. 18:59what do you mean you can't find the chat? You just said I can't find the chat. [laughter] That's the chat.
126. 19:07Where where you just typed the chat is the chat. [laughter] Okay, I'm gonna share my screen.
127. 19:24Hide that. Move this. I'm going to go here. Okay.
128. 19:39I want to be able to see all of you and whether or not I'm still sharing. So, I've got you over to the left and then I've got this.
129. 19:55Okay. So, what you're looking at right here is an infographic I put together um today.
130. 20:12I'm going to do this in builds and stages. And again, I'm going to this like the the end result of this what I'm going to
131. 20:21show you is essentially the architecture of of what Buzz is like how it works. I'm going to try to talk about By the
132. 20:30way, we're seeing mostly a black screen. That's correct because I'm going to do this as a build. Um, so let me give you let me give you a
133. 20:39little bit of a little bit of context about who created this. process. On a side note for anyone who's watching
134. 20:47it, Altered Carbon on Netflix, the future of how agents will communicate with you with each other on our behalf.
135. 20:54It's not that crazy. Buzz.com. It's uh buzz.xyz if you want to if you So, so here's here's a thing that is is
136. 21:02probably worth you doing is go to buzz.xyz, download the app. I'm going to be explaining tonight different pieces of
137. 21:10it. It's it is not a simple platform. It's very early software. Oh, it's Mac only right now. Um, it's a little complicated,
138. 21:23but we're going to talk about why all these pieces are important. Okay. Buzz.xyz. Okay. So, the fir first thing about Oh,
139. 21:34let me go let me go back. Jack Dorsey. Jack Dorsey is the guy that invent in invented Twitter, founded Twitter, um,
140. 21:44now X, and he's got a company called Block. And back in April, Block famously laid off 4,000 of its 10,000 employees
141. 21:55while being profitable and growing. And Jack Dorsey's comment on that was we're profitable, we're growing, and
142. 22:08our people amplified by AI are changing how we do work, how we do startups and how we do work. And so he laid off these
143. 22:194,000 people and about two weeks later he wrote a post, which you should look up, called from um from hierarchy to intelligence.
144. 22:29And what that post is is a history lesson of 2,000 years of organizational design starting with Roman soldiers. [laughter]
145. 22:41And and he and he sort of talks about as you scale organizations where they all fall apart. No matter what
146. 22:49structures you use, no matter what innovations you use in structural design, organization design, they all
147. 22:56fall apart because as the organization gets bigger, you need to add layers and layers of management, middle management.
148. 23:05And humans are slow and inaccurate. They don't communicate well up and down the chain. So you have leadership saying
149. 23:14where we're going. You have the workers down here working. And then middle management everything up. Not
150. 23:19because they're bad people, but because it's just a game of telephone and it's very inefficient. And and for 2,000
151. 23:25years, there's been no way to make that more efficient. So what Jack Dorsey said is, I'm going to fire 4,000 people that are largely in
152. 23:34those middle management roles. I'm gonna replace middle management with AI because AI can communicate quickly and
153. 23:41can figure out where it should route things and he's going to dramatically simplify the roles of the organiz organization, things like that.
154. 23:51That's a ballsy move, right? Like whether you agree with it or not, whether you like it or not, whether or
155. 23:57not he's going to be successful at that. What he recognized was that every large organization and a 10,000 person
156. 24:05organization is large. I've I've managed uh I created and managed a 2,200 2,200 person organization. It's a
157. 24:15nightmare. Oops, I just lost my screen share. Oh, you just you're just bringing me back up. Okay, cool. Got it. Um
158. 24:24managing a large organization is a mess. So, for him to recognize that that AI potentially solves this 2,000-year-old
159. 24:34problem, and he's going to make a bet that this is how organizations are going to be built and managed in the future is
160. 24:44similar to making a bet like you could create a service where people could only make what was it 128 character post,
161. 24:52right? The original Twitter was it 144 characters. What was the original Twitter? Can anyone look that up? Um, okay, cool. It's back. Okay,
162. 25:07so this is a guy in Jack Dorsey that can not only see the future, but can anticipate how we're going to interact.
163. 25:15And so he's the one that 128 128 letters in the original Twitter. When I first heard that, I thought it was the dumbest
164. 25:24thing I'd ever heard. It was the dumbest thing I'd ever heard. MySpace is also apparently coming back. That's pretty funny. Um,
165. 25:40so what Dorsey recognized with Buzz, what he why he's created this is there's a there's a couple of things
166. 25:50at the at the core of it is what's on screen right now. Not a joke. Yeah, I know. I can't download Buzz. It's
167. 25:57probably because you're on a PC, Mr. It. It's Mac only. um one platform shared context, humans and AI agents as equal teammates.
168. 26:09So at the core of this thing is not only do you have your human partners, but you have agents that are are represented as equal teammates in
169. 26:22this Slack-like conversation. And so just like I could tag, you know, Cindy in a conversation and say,
170. 26:30"Hey, Cindy, I have a question for you." I can tag an agent and say, "I have an agent for you." Couple other things that he realized is
171. 26:39that SAS products, software as a service, rob you of your data. And
172. 26:54up until 2022, up until Chat GPT, we didn't really need our data. It wasn't valuable to us, right? Like, we
173. 27:06freely gave our data to Facebook and to Google because it's like, well, what are we going to do with all of our search
174. 27:13history? We don't we're not data scientists. We don't know how to deal with that Well, now that AI comes along, all of a
175. 27:20sudden AI can parse all of our data and it can find value in all of our data. And so, one of the things that that
176. 27:27Dorsey recognized is that data is super valuable and it shouldn't be owned by Slack or by Facebook or by Google or by
177. 27:36right. It should be owned by you. If it's your community, it should be owned by you. Digital god's coming on screen. So now
178. 27:46AI will be the middle middle management Yeah, they will. And you you know what's what's the the thing about them being middle management is
179. 27:56what you don't get with AI digital gods is you don't get all the politics and and you know middle managers
180. 28:02that are like I'm going to keep this guy down because if he shines too bright you know I'm not going to get promoted like
181. 28:08all the human dynamics in the middle of those corporations. It's one is communication inefficiency, but the
182. 28:15other one is just human beings are messy, right? So, it's just it's just taking the mess out of it. Now, is AI
183. 28:21going to be perfect in that middle management layer? No. It's going to be there's going to be all sorts of
184. 28:25disasters, but it's you know, it's at least an attempt to do something. So, our data is really important to us. So is our identity, right?
185. 28:38If you've used chat GPT for more than a week or two, you recognize then that it's got memory and every chat that you
186. 28:45have, it gets better and better at knowing who you are. But right now, most of that chat history belongs to OpenAI.
187. 28:53Now, sure you can export your data and you can download your data, but who does that first of all? No one. Nobody changes the defaults,
188. 29:01right? And if chat GPT decides to ban your account before you get all your data, it's gone. So all your history is gone.
189. 29:12All that context is gone. So that's a really important piece of this identity, your data, data sovereignty.
190. 29:20Um, and then the other thing is this. There's all sorts of tools that we use, right? We go to Claude and we go to
191. 29:31chatbt and we go to um perplexity and we're using codeex and we're using claude code and co-work right we we've
192. 29:39got all these different harnesses we took the time to spin up an openclaw like me and create Adam the world's worst employee
193. 29:48you know we set up Hermes so we've got that we set up local compute on our fancy ass computer that we spent too much money for
194. 29:59why do we have to do all that at different places? What if you could do that in the same place? All those
195. 30:05different things in the same place. So, there's a bunch of things he's solving for. Okay, so couple of the key features of Buzz
196. 30:16portable identity. When you create an account, when you download Buzz and you create an account, what it gives you is a private key.
197. 30:24So, this is just like if you ever played with NFTTS, you know, you've got you've got your private keys.
198. 30:34You manage your keys. You manage your money. If you lose your keys, we've heard these stories of people that had
199. 30:4071,000 Bitcoin on a USB drive and they forgot their key. They forgot their [laughter] their recovery passwords and
200. 30:48their key is gone and it's gone. But the advantage of having that that key pair that cryptographic key pair key
201. 30:55pair is that your identity is portable. So this is built on what's called the Nosteret uh Noster protocol NOSR
202. 31:06and it's it's a ledger, right? It's a cryptographically signed ledger and then identities that are cryptographic pairs
203. 31:14where I keep my private key and I can share my public key. Um, it's AI native. So, so Buzz doesn't treat Cindy
204. 31:23different from Fizz, the programming dude, the programming agent. I can go in and interact with them interchangeably
205. 31:29and actually the three of us could have a conversation. Um, there's shared context. So, all of the context that I'm
206. 31:37doing in all these different places is now shared and cryptographically signed into a single stream, which I'll talk about in a second.
207. 31:47um that I own. If I'm if I'm the community owner, I own it. Um it's signed and attributable. Everything is signed. Every action
208. 31:57has provenence. So what that means is let's say let's say that you do a revenue sharing uh a revenue sharing
209. 32:08model that's based on contribution in the community. Well, if you were in Slack right now and you say, "We're
210. 32:14going to we're going to base how we pay you based on, you know, who contributes more in the Slack community." Well,
211. 32:19who's going to police that? How do you know what's in there? With this, every single thing is tracked in in a signed
212. 32:27linear just timeline. Here's all these events that have happened. Here's everyone that did everything. And you
213. 32:33can use AI to analyze all that data and figure out who contributed what where and things like that. Um, and then the
214. 32:39final thing is is that this is open and decentralized. There's the there are Buzz servers, which I'll talk about, but
215. 32:47it's not like there's there's a server farm sitting with all the Buzz accounts on it where if they shut that down, you lose everything.
216. 32:57There are relays that they've got where they're hosting them for you for convenience, but you own all the data
217. 33:02that's on those and you could even spin up your own servers and do it without them. So, they're giving you this as
218. 33:08convenience, but you can do this on your own. Okay, any questions? That make sense? This sounds a lot like the contribution
219. 33:17eco economy system I just wrote into the first book of my novel series, Rewrite. Well, Charles, this Yeah. Is that like
220. 33:24blockchain? Yes, it's very much like blockchain. It's it's it's the Nostra protocol which is you know cryptographic
221. 33:31um you know signed cryptographically signed events that go into a ledger that is portable. So so I can be hosting
222. 33:44a community on this server and then I decide I don't want to host that community anymore. I can just transfer
223. 33:53this community to someone else and like the entire history goes with it. It's it's completely portable. All of the all
224. 33:59of the context, all of the everything is completely portable. The relationships to the agents, what the agents do, all
225. 34:05that sort of stuff is in this is in this record. Um as is my identity, which is this unique thing. Okay.
226. 34:14So the big idea is humans and agents inter interact and then those agents can sit on top of different platforms. So I
227. 34:24could have a programming agent that sits on top of Claude Fable or maybe I have a planning agent that sits on top of
228. 34:32Claude Fable and I have a coding agent that sits on top of uh Chat GPT 5.6 Soul Pro, right? Which is the really big
229. 34:41badass coding thing. And then Goose is another network. And then you can also run your own local models. Like I could
230. 34:48run my um my Quen 3.6 model on my MacBook Pro and make that available to this community. So you can create agents
231. 34:58that run on different platforms and then they're in here as entities that you can interact with. Chef Kelly question. Yes.
232. 35:06But what happens when you share your data with others in the community? I'll get to that. This is where this is where
233. 35:11this gets quite complicated, but it's not dissimilar to Slack. So, in in Slack right now, I've got a channel, a direct
234. 35:19channel with my co-founder and I, and she and I can share things in there that, you know, we don't want anyone
235. 35:27else to see. Um, there's an area for finance and they can talk about finance stuff and any anyone who has access to
236. 35:34the finance stuff can have access to that channel, but other people can't. So you can restrict things on a channel
237. 35:41basis, but you can also restrict things on an agent basis. So I could create an agent that uses my local inference over
238. 35:49here and I could just say I only want me to be able to use that agent. And so if I put that agent into a space,
239. 35:57no one else can use it. Only I can use it. Linda Sherman uh reminder that Jack Dorsey created Blue Sky with similar
240. 36:05concept of individual pro prominent prominence. Um of course that was before AI was a factor. Yep. Exactly. Um the
241. 36:13reason why I'm more bullish on Buzz than I was on Blue Sky is that Blue Sky was a a more portable version of Twitter, but
242. 36:26Twitter wasn't going away. and like unseating that incumbent was was was essentially an impossibility.
243. 36:37The thing about this is this is going to allow allow us to work with one another and to work with these agents that are
244. 36:45going to go do work for us and I think going to go do work for us and make money and then bring that back to the
245. 36:53community. Okay. Are they encrypted, Kyle? Um, I think so. Yes, I think so, but I don't know. I assume they have to be one cryptographic
246. 37:07identity layer. So, the keys that you own, you've got a portable identity. Um, it's got permissions in it, it's got
247. 37:13reputation in it, and it's got an audit trail in it. And then all of that, all of that stuff sits on what's called a
248. 37:22relay server. So the buzz relay server and they call it a single event stream. So the single event stream means that in
249. 37:30your community. So, so when you create a Slack um channel, right? You you you know your Slack server for Story Vine. That's you know
250. 37:44all of the conversations with my company and then they're com you can subscribe to multiple communities, right?
251. 37:51Same thing here. And so each one of those communities lives on a server called a relay. Now, um I I'll explain this in a second,
252. 38:01but Buzz has a thing called um build buildbot.xyz that is that is a relay hosting service that that Buzz
253. 38:16is providing, but you can also host them on your own. That doesn't matter. Um, it it's just just important to know because
254. 38:24that's a very confusing part of this is when you create a community, you're actually creating a relay server on
255. 38:30their their relay server farm that they're basically giving you three communities for free for now. Um but
256. 38:39then what this means single event stream is all of the things that you can do in this community chat threads writing code um
257. 38:51branches patches reviews um command line um workflows canvases files audit logs all of that stuff is every single event. So every comment
258. 39:05you make, every piece of code that's written, all that stuff is signed and added to that to that ledger,
259. 39:14it feels like blue sky and Slack and Obsidian. And yeah, it's it's a lot of things. It's a lot of things. Um, the other cool
260. 39:24thing here is you can use multiple harnesses, right? So you can use all your different harnesses. Um,
261. 39:32so this is the thing where I was talking Buzz, the hosted option um is it's at builderlab.xyz. So when you say I want to create a
262. 39:44community, you download the buzz app and you say I want to create my first community. Pay attention to your URL,
263. 39:51you're going to go from buzz.xyz to builderlab.xyz. Steo, okay? Not to be a party pooper, but if AI is already replacing thousands
264. 40:02of workers because it's cheaper, why should we believe the future vision where AI creates abundance, everything becomes cheaper, free, and
265. 40:10ordin ordinary people no longer need to work? If companies are already using AI to cut jobs and boost pro boost profits,
266. 40:17doesn't it make sense that they'll keep the financial benefits rather than give the money away? Doesn't Jack Dorsey's
267. 40:22decision highlight that contradiction? I think multiple things can be true. Steo,
268. 40:38the dystopian, cynical view is that every single company is going to fire every single employee and keep all the profits for themselves.
269. 40:50I don't think that's a true statement. I think many companies will do what they can to retain their employees and rather
270. 40:58than reduce costs and keep their productivity the same. Maybe they keep the same amount of employees and 10x
271. 41:04their productivity. Right? There's there's two models here. One is we can we can make the existing amount of money
272. 41:1110 times cheaper or we can keep the existing cost base and make 10 times more revenue. So there's there's two
273. 41:19philosophically different models. That said, I think the idea of a hierarchical organization
274. 41:33really starts to fall apart when you can have a group of 10 or 20 people that can do the work of a 200 or 500 or
275. 41:47thousand person company. it probably starts to make sense for big organizations to start to break into smaller economic units
276. 41:58where where you take people that are good at working together. Maybe you've got a group of people that are really
277. 42:02passionate about I don't know whatever financial services and they just love numbers. Put them in a put them in a
278. 42:09cohort and have them build stuff, right? And then you've got other people that are into product and other people that
279. 42:13are into customer service. So I I could see that large organizations become individual pods. The other side to your point, there are
280. 42:26going to be a lot of layoffs. There's going to be a lot of layoffs and some sectors are going to be decimated.
281. 42:34So, like customer service, for example, is probably going to go to a place where most customer service is going to be
282. 42:41handled by agents and then there's going to be a couple of really high-end customer service people that do the the
283. 42:46sort of white glove treatment. So, where do all those people go? What I'm talking about is 10 of those people
284. 42:54could get together and put together one of these things. the the the thing that makes this not quite what you're talking
285. 43:01about is the fact that all of this stuff is portable. Buzz doesn't own our data. Buzz doesn't own our identities. Buzz doesn't own the
286. 43:11context that we put into our community. So, if I create an AI salon Buzz community and I bring in 10 or 20 people,
287. 43:20all of that context, all of the smarts and the genius and whatever we come up with in there, that's ours. and we get
288. 43:27to take that with us, right? It's tied to our identity and and it's portable. So, I think that some of what you're talking about is
289. 43:38going to happen. I think there's other dynamics at play where it won't just be the dystopian version of this. I think
290. 43:46there's going to be all sorts of new opportunities and and I the thing I'm excited about here is that a group of
291. 43:53motivated people that know and trust one another could use a platform like this to dramatically amplify their impact and
292. 44:02their economic potential. So that's that's where I'm looking at. um can't function on 10x the administrators
293. 44:16um per doctor forever. Yeah, exactly. I think the elite overp production issue was already about to collapse with
294. 44:24modern world. Um buffering the middle management was better efficiency always needed to occur. Yeah. Yeah. It's a the
295. 44:33it's it's AI trickle down. Yeah. Yeah. I mean, it's not though. Like I mean, here's the thing. Trickle
296. 44:40down assumes that they're going to give us What I'm talking about here is we can reclaim our agency and go build a really powerful company
297. 44:48in in a in a cooperative fashion that's got all sorts of ways to track everything we do. Um, time for a great
298. 44:56repurpose commercial. Okay. So, um, especially if you're new here, um, welcome. If you haven't made an account,
299. 45:08an AI salon account, please go make an AI salon account. Please introduce yourself. You should also check out,
300. 45:13you'll see in the in the at the very top there there's an area where it's like, you know, check out our values and the
301. 45:19the welcome message and and you'll see there's the mastermind. [snorts] Check out the mastermind. So, the
302. 45:25mastermind is a subscription area of the salon. It's it's 49 bucks a month and within it we're doing things like um the
303. 45:34AI salon mastermind practice lab where we're creating daily practices around how we use AI really understanding
304. 45:41what's possible really understanding who you are what your value is and how you use AI in an intentional way
305. 45:51where you're not just like chasing the next model and chasing the next model and chasing the next model even though
306. 45:57I'm chasing the next I think this this software is very very significant. So even though it's new um it's not
307. 46:04hypocritical. I I'm claiming it's not hypocritical because um because I think this actually solves a bunch of things
308. 46:14that I've been I've been thinking about for a long long time. The other thing that's within the mastermind is a whole area called the
309. 46:22great repurpose. And the the idea of the great repurpose is that every single one of us, whether we get laid off because
310. 46:29of AI or whether we keep our jobs and AI starts to change our jobs, right? So the canary in the coal mine for what's about
311. 46:36to happen to every single job is programmers. So programmers until the third quarter of 2025 were pretty
312. 46:47much like, yeah, programmers, we still programmers still going to program, right? because the AI kind of sucks and
313. 46:53it isn't really good. And then in December of 2025, Opus 4.6 comes out and a lot of the engineers are like, "Oh
314. 47:02It's here. It's good enough." And then about two months later, you started to see a lot of people on X going, you know, I loved programming,
315. 47:11but that's not my job anymore. I'm just babysitting AI agents. I don't like my job anymore. And so the whole idea
316. 47:17behind the great repurpose is that to the extent that we've got our identity coupled to the way we used to work, as
317. 47:24that work changes, we're sort of left with this gaping hole of wait a minute, what's my value here? It used to be
318. 47:33programming and now I'm like kind of a babysitter. Is that my is that what I really want to do with my life? So a lot
319. 47:40of people are going to be facing that. The great repurpose is about really understanding that about yourself and then using this idea
320. 47:48within the AI salon we call AI readiness to get yourself up to speed on what AI makes possible and then start to
321. 47:57understand what is the digital you. How do you create agents that represent who you are and and what you value and what you want to
322. 48:09accomplish in the world. Buzz, this thing that I'm talking about becomes a platform to be able to do that. So, if
323. 48:16you haven't checked out um the mastermind, please go do that. Um it's the the people in there are quite remarkable. The people in the salon
324. 48:23overall are quite remarkable. Anyway, okay. So in the upper right hand corner here, Builder Lab, which is basically this
325. 48:33hosting service by Buzz, they will give you three servers, three relay servers um for three different communities.
326. 48:45And then after that, you can you can host your own on your own platforms. Right now, they're going they're giving
327. 48:50you three. So if you want to get you know, basically a company up and running, you can do three of them, three
328. 48:55communities. Um and then um and then if you host them yourself, the advantage of hosting it yourself is that
329. 49:06if Buzz goes away, your company still sits there. Like if Slack went out of business tomorrow, do you know how many
330. 49:14businesses would be screwed if you do this with Buzz? That wouldn't be the case. Okay, so what else we're going to
331. 49:20talk about here? Okay, another thing that Buzz does, and this is something that I've I've been thinking about for a
332. 49:26while. So, Buzz Buzz also supports this protocol or this this app or whatever it is, this this it's probably a protocol
333. 49:36called Mesh LLM service maybe. And what Mesh LLM does is it allows multiple people to contribute their compute to a
334. 49:48central mesh. So, I could take my 64 gig MacBook Pro and someone else could take their 128 gig Nvidia Spark and someone
335. 49:58else could take their 512 gig Mac Studio Pro and we can basically um share them on this same relay server and we can set
336. 50:09up agents that use that shared compute. You can even pull it. I think it slows it down fairly dramatically, but you can
337. 50:17pull that compute. And here's the here's the the the the the important part of being able to pull the compute on my 64 gig machine, I can't
338. 50:30run GLM 5.2, which is this big massive model. Um, but if I had a bunch of different computers that were all contributing
339. 50:41their compute and we pull that together, now all of a sudden all of us can run that that really expensive or that
340. 50:48really uh powerful model locally um where we own the data, we own the compute and the cost of it is really
341. 50:57just the cost of electricity of each of the individual things plus the hardware they invested in to make it available.
342. 51:03Um, so that's a big important piece of this. And then the final piece is just your identity is portable, your history
343. 51:09is portable, your communities are portable, you can take them anywhere, they can live in all these different
344. 51:14things. Um, it's built on Noster NSTR. So you can go look up what the Nostra protocol is. And if you want to learn
345. 51:22about cryptographic pairs and all that sort of stuff and and how signing of these events, this single event stream
346. 51:29that where all the events are are signed, you can go learn about that. You can go geek out. You don't have to. Okay, let me come back.
347. 51:37Kelly Camp Rut row. I'm getting timed out error when trying to sign up for buzz.xyz. I don't know what that's
348. 51:43about. They might just be my ISP is blocking the buzz site. Total bummer. That's weird. Maybe they're It's possible that they are um they're
349. 51:54experiencing badness, you know. Badness. [laughter] Okay. So, here's here here it is. This is what you're looking at right now is
350. 52:08Buzz. And so, you'll notice here I um this is the AI salon um community. And [clears throat] if I go to um
351. 52:24community settings, I added the, you know, salon logo. So, you can create your own community. You could create a community for a
352. 52:34business. Like, think just think of it like a server. You you could create a buzz server for your business.
353. 52:41You could do one for your family. You could do one with your geeky friends. Um, you could decide you want to, I
354. 52:48don't know, make a magazine and you're going to you're going to use agents and and other human beings to collaborate on a magazine. Um,
355. 52:59you can create new channels. Down the lefth hand side here, you can create new agents. Um, Buzz ships with three
356. 53:06agents. And if you go look at they are they are Honey and Bumble and Fizz. [laughter] And and so when when you first launch
357. 53:18Buzz, uh, it it I guess it's doing the compute on the Buzz servers. They give you some amount of compute on the Buzz servers. Um,
358. 53:27and then they all introduce themselves. So, Fizz says, "Hey, Kyle, Shannon, I'm Fizz. Welcome to Buzz. This is your
359. 53:32private home base, and we're here to help you get oriented or work through something you're building." Honey and
360. 53:39Bumble introduce themselves in a sentence or two. Um, introduce yourselves in a sentence or two. Share
361. 53:45what you're good at and what you and and when when to bring you in. Don't start any work yet. And then Bumble says,
362. 53:54"Hey, I'm Bumble. Bring me in when you want research, option comparisons, and careful fact check of clear sources. I
363. 54:01enjoy turning fuzzy questions into a useful answer." Honey. Hey, Kyle. Shannon, I'm Honey. Bring me in when you
364. 54:08want to turn a rough uh thought into clear writing, organiz organized ideas, um brainstorm options, blah blah blah.
365. 54:16Fizz, want some help? Uh what can we help you build? bring us something you're working on and give us a quick
366. 54:22challenge to see how we can work together. So, these three agents are the defaults. You can make your own. I can also go to
367. 54:31the agents tab here. And so, here's the three agents. I I made an agent called codeex 5.6 terra um which which I'll
368. 54:40explain in a second. But if I go into say fizz and I say edit here are the agent instructions. So this
369. 54:51is think back when we were making custom GPTs. Remember we were making the custom GPTs. This is your system prompt. Okay. So you
370. 55:00can give it a name, you can upload a picture and then you give it system prompts. You are fizz an energetic maker
371. 55:06who turns ideas into action. Be upbeat, practical and decisive. Help users plan, create, solve problems, and finish work.
372. 55:14Add occasional beword play or or icons or emojis. Keep it charming, never distracting, right? So, creating an agent here is as simple as
373. 55:27creating, you know, a custom GPT or just a system prompt in chat GPT. Now, here's where the thing gets cool.
374. 55:35the AI configuration right here. You can use the default harnesses, the harness defaults, or you can customize it for
375. 55:43this agent. So, I could say, you know, I I think it defaults to the Buzz agent. Um, and then there's Buzz shared compute and
376. 55:54then here's the model. So, it's always automatic, but I can say I want to use codeex. um which which I so one of the things
377. 56:03you do when you start your community you bind it to your agentic platform. So you can bind it to claude code or you can
378. 56:12bind it to codeex and then anything you can do in codeex with your $20 a month or $200 a month subscription
379. 56:20now is available inside buzz to turn into an agent. Right? So I could turn fizz I could turn it into fizz. I want
380. 56:27him to be 5.6 soul. And then there's advanced. I don't know what these are. Oh yeah. Who can talk to this agent?
381. 56:35Only me by default. Anyone or an allow list, right? So I can I can allow specific people. It is free. Yes. Right now Buzz is free.
382. 56:45I don't think it's open sourced, but I'm not sure. But it's running on Noster, so your ability to build a custom agents
383. 56:55and things inside this are going to be quite quite simple. Um you can add um variable environment variables. So those
384. 57:04might be things like API keys. Um instance name pool. Oh, I guess this is just things that you can nectar, comet,
385. 57:14bramble, clover, pollen. I don't know what those are for, but I'm sure they're for something. Um how many conversations
386. 57:20each running instance handles at once? So 1 to 32. So you can turn an agent into a multi-threaded agent.
387. 57:29But very very easily you can spin up new agents. Right. So so one of the things that I was thinking of doing and maybe
388. 57:37we'll do this is my very first custom GPT was the Sin City Sipper and I'm I'm thinking that what I can probably do is
389. 57:46take one of the lower level chat GPT models, right? And then just take the system prompt from Sin City Sipper and
390. 57:54put it in here and probably do Sin City Sipper as a like the the local bartender for all these things. Um, all right. Let's see
391. 58:07what's going on over here. Honey sounds like an HR nightmare. I've been using Victor. It's not free, but it's very useful. Never distracted.
392. 58:17So, you could copy your GPT instructions into that area. Correct. Steo42 was speaking about a different
393. 58:24kind of buzz being costly. Back to my padded chat. [laughter] Um, let me pop in here to um settings. So, this is here's a kind of confusing
394. 58:37thing. There are community settings. So, this is just like the icon, the name of the community, um, where you're storing your
395. 58:48repos locally. Um, if there's an API token, I don't quite know what to do with that yet. Um, point the agents repos directory to an
396. 58:59existing folder so agents can work in your local checkouts. Leave blank to use a default location. Um, so community settings are just sort
397. 59:10of a thin letter set of settings. But settings is I guess settings is tied to your identity. Um no I guess it's for the whole for the
398. 59:23whole community. Um so I'm the owner of the of the community. So I guess this is this is where I set up me. Um there's all sorts
399. 59:34of different kind of appearance settings you can change. You can turn on and off notifications and change different sounds for the
400. 59:42notifications. Sin city where all the agents hand hang. Um, navigation shortcuts. Custom emojis. I don't know what you do with a custom
401. 59:55emoji. The local archive. Save copies of relay messages to a local SQL 8 database in your Buzz Nest. events are reverified
402. 1:00:04against the relay relay at archive time. So basically if there's anything that you're doing that is is ephemeral
403. 1:00:14disposable stuff you can actually capture that and back it up as well. So this is again this is kind of geeky
404. 1:00:20right now to set up [clears throat] because you're managing your own data now and like what you save, what you
405. 1:00:28don't, where it gets saved, all that stuff is actually really important. Um, Gwen, it was made to be in buzz. You
406. 1:00:37could set it up and go around the room every 45 minutes and maybe have closing time to save mental health.
407. 1:00:46Um, hosted communities. Okay, so here's this thing. This is probably the most confusing thing I've come on so far. So to see
408. 1:00:59the the hosted communities that you're hosting on Buzz, you have to log into builder lab. So builderlab.xyz, think of that as the servers.
409. 1:01:11And buzz.xyz is where you get the app. And so think of um the buzz app is kind of like Slack. And then the builder lab servers are
410. 1:01:26where your data is stored. And so if I sign into builder lab, excuse me. Um it here are the three communities I've
411. 1:01:40created. AI salon um communities.buzz.xyz Kyle Shannon and Storyvine. So I've got three of them. You can only have three.
412. 1:01:49But I can if I just have a a separate server or I could probably even use my MacBook Pro as the relay server. Get get
413. 1:01:57a big external hard drive for it. Something like that. Um I can I can have as many of these as I want. So you can
414. 1:02:03create as many as you want, but they will host three of them for you. And again, right now I don't I don't think it costs anything.
415. 1:02:10Um templates are uh when you create new spaces, you can have a template that will like automatically set up spaces
416. 1:02:18the way you want them. Um invites, you can you can invite people to the community. Um agents, so agent runtimes, agents
417. 1:02:29control how agents behave in conversations and run on this machine. Keep addressed agents active. Let's see.
418. 1:02:36I don't know what that means. Keep them awake while they're active. So, there's all sorts of settings here that I don't
419. 1:02:40I don't quite know what they are, but look at all these. There's the Buzz Agent. So, that's that's the Buzz
420. 1:02:46servers. So, so they've got AI available for us to use. Um there's goose, which is command line. Um AI cla codeex.
421. 1:02:56So, you can connect both of your cloud code and your codeex if you have both. If you have openclaw, if you have
422. 1:03:02Hermes, I can also add runtimes for local AI inference. So the cool thing about this is in this single interface I can have agents that
423. 1:03:14are like like one of the things that we've talked about in here is that people are using Claude Fable for
424. 1:03:22planning and GPT 5.6 for coding, right? Because that tends to be better at building Fable's really good at planning Well, within this
425. 1:03:32environment, you could create a planning agent that uses Claude, and you could do a building agent that uses GPT 5.6 Soul.
426. 1:03:40And then you could put those two agents in a space where you're talking about planning and building something, and you
427. 1:03:45just say, "Hey, Claude agent, go plan this for me." It'll do the planning. And then codeex, go build it, and it'll go
428. 1:03:52do the building. Um, and you can add different things in there, different harnesses. Um, let's see. They support
429. 1:04:01Buzz Agent, Claude Code, Codeex, Open Claw, Goose, AMP, Cursor, Grock, Build, Hermes agent, Kei Code, Oh My Pi, and
430. 1:04:11Open Code, and and probably more um advanced. Let's see what this is. Environment variables. So, that's going
431. 1:04:18to be APIs and things like that. So, what what that looks like is if you're using something like uh Midjourney
432. 1:04:27doesn't have an API, but Runway does. If you're using the Runway API and you need to put your API key in there, you can do
433. 1:04:35that here. Here's the compute. This is where I can run local compute on this computer or I can run it. I can run Buzz
434. 1:04:42on other computers and make those available to this computer. Um, experiments, these are just features. I've got four of them turned on.
435. 1:04:50Workflows, projects, pulse and forum channels. Um, there is a mobile app that you can connect and then you can check
436. 1:04:59and see if there are updates. Let me check again. I'm I'm up to the latest version. Um, so, so that's all the settings. Um,
437. 1:05:10so let's do this. Let's create a new channel. And we're going to create a new channel called the speak easy.
438. 1:05:23Create no channels meet your search. Create channel. The speak easy. Um, this is a third space. Never heard that term before until this
439. 1:05:36year. a third space where we can just hang and blow off some steam and order a cocktail
440. 1:05:53from the sin city sipper barkeep. [laughter] Okay. Um the channel type is ongoing. The visibility is public. So, anyone in
441. 1:06:07my community can see the speak easy. So, I'm going to create the channel. So, here's the speak easy. Now, I'm
442. 1:06:15going to add people. Let's see who have I got in here. I don't have many people in here yet. If I say Cindy I've
443. 1:06:22invited. She's here. So, I'll add her. Um, and that's it. Right now, it's just going to be her and I drinking in here.
444. 1:06:27And now I can create an agent. So, what we're going to do is we're going to go over to chat jpai and we're going to go find sin city sipper
445. 1:06:38GPTs. Get rid of that. Make this smaller. Why is that weird like that? Okay. So, let me do sin
446. 1:06:53city sipper. There's Sin City Sipper. And so, let's see. How do I go here?
447. 1:07:08I can see clearly now. The rain is gone. I can see all obstacles in my way. How do I get to the How do I get to the bottom? [laughter]
448. 1:07:24Um, what is going on? >> [whistles]
449. 1:07:39[whistles] >> Okay. So, I'm going to say edit GPT because I made this one. So, let's go to configure. Here's our instructions. Let's see.
450. 1:07:51Select all. Copy. All right. There's all my There's all my instructions. And so now I'm going to go to buzz and
451. 1:08:01I'm going to say create an agent. You're all still seeing this, right? Any questions? The AI learning bar.
452. 1:08:09Ross O'Shea. I like that. Meaning the agent purpose with a model that works best for certain tasks. Is that not something you
453. 1:08:18can do on agent systems like NAND? Now you can but but N8N you know NADN and Zapier and things like that are are more about like logic
454. 1:08:32treebased linear automations. Um whereas this is a system that that says these agents may go off and
455. 1:08:43automate something and they may even use an NAND automation but you're going to interact with the agent as if they're
456. 1:08:50another person. So you have to go into NADN and you have to configure things and things like that. And I haven't
457. 1:08:55played with NN in a while. So I would assume they've got a lot of AI helpers, right, that help you build the
458. 1:09:03automations. But this is more like think of these agents as employees or co-workers, right? Colleagues um that have
459. 1:09:14special intelligence. They can have multiple intelligences um and then you can, you know, send them off to go do stuff. It's it's not
460. 1:09:21dissimilar, but it's it's a think of this is more freewheeling. Um and NADN and Zapier are more like deterministic.
461. 1:09:30when this event happens, then I'm going to trigger this automation that always happens the same way. And this would be
462. 1:09:36more like, hey, go out and think up a, you know, go do some research, think up a new business, and come back and
463. 1:09:43propose five new businesses to me that we can start this week. Um, okay. So, I'm going to create an agent. We're
464. 1:09:51going to create a new agent. I'm going to paste in there's the Sin City Sipper stuff. I'm going to say customize for
465. 1:09:59this agent. So, I'm going to use codeex. I'm going to choose a model. I'm going to choose 5.6 Luna because that's the smallest 5.6
466. 1:10:09model. Then I'm go I'm going to go advanced. Who can talk to this agent? This is anyone. So, we're going to let anyone
467. 1:10:18into the speak easy and anyone can talk to Sin City Sipper. We'll just do one at a time. Name instance pool. I wish I knew what that meant.
468. 1:10:31Instance name pool. Hey, someone want to go do research for me while I'm building this thing? Go go go ask probably Grock
469. 1:10:41um in the the new Buzz app. What is the purpose of instance name pool? Because I'd like to know that. Why can I not create this? What did I
470. 1:10:54not do? Oh, I didn't name it. Sin City sipper.
471. 1:11:10Oh, and then let me go back over here. I can copy that. Can I paste? I can paste a URL. Okay. Wait.
472. 1:11:31Uh, can I not copy this image?
473. 1:11:57Sin City. [whistles] All right. I don't know where that is. Uh,
474. 1:12:11here's what I'll do. So, let me go Sin City. Go back. Let me go. Um,
475. 1:12:25make me make me a fancy sin city neon
476. 1:12:41highball. Not a highball or a Yeah, let's see. Cocktail on the rocks. All right, we're going to get a picture. We'll get a picture.
477. 1:12:58I'm not saying I'm professional wine taster. Sin City Sipper is is my most used sidebar. [laughter] I can train this agent.
478. 1:13:11Something went wrong. Please try again. All right, it's thinking. I haven't used a custom GPT in probably
479. 1:13:20more than a year, so I don't even know if this is going to work. All right, let's go back here. Did I make that
480. 1:13:25agent or did I cancel out of it? I think I canceled out of it. Oh, damn it. [whistles]
481. 1:13:49>> [whistles] [whistles] >> Uh, it doesn't like what I'm doing.
482. 1:14:10Extended thinking effort. Standard. Can I stop this thing? Retry. Stop. Retry
483. 1:14:37designing cocktail artwork. Where's my Where's my um Where's my recipe? These things don't work anymore.
484. 1:14:52[laughter] I guess I I guess I'll just stop this.
485. 1:15:09All right, let me stop it. Stop. Can I stop it? How do you stop it? Reload.
486. 1:15:26Okay, let's go. Edit, configure, select all, copy. All right, let's go back over here. In the Buzz ecosystem, the term the
487. 1:15:41terms instance and pool usually refer to how resources and AI agents are managed across the workspace. I don't know what
488. 1:15:47that means, Kyle. The art doesn't work in your GPT. Hasn't for two years, actually. That's weird. I I played with it maybe a
489. 1:16:00year ago, and it actually did really well, but I don't know. It It It's likely broken. All right. Sin City Sipper
490. 1:16:10um custom agent codeex model 5.6 six Luna advanced anyone. And then we're going to create the agent.
491. 1:16:27There's a private key which you shouldn't be seeing. [laughter] And then I'm going to say add an agent here. So there's Sin City
492. 1:16:41Sipper. It's already in here. Okay. So if I say at sin city sipper. So there's my agent. Let's see if this works. Um, make me
493. 1:16:58a bourbon based dead heat of the summer 100
494. 1:17:14degree day. Refreshing cocktail. All right. So the look says it's do it's I see you. Then the this says I'm doing
495. 1:17:29something. And down here Sin City sipper is doing something. So I'm crafting a Heatwave bourbon cooler with mint citrus
496. 1:17:38and light sparkling finish. And I'll place the recipe in a fresh canvas. A fresh canvas. So, we're going to have
497. 1:17:47to probably change our uh our system prompt because I don't think Codeex has it. Permission requested. Approved. Allow once.
498. 1:17:57So, our agent is off making us a drink. I think the recipe is published. One reply. Here's the reply.
499. 1:18:12Bourbon. So, it's not very well um formatted, but it did make us a drink recipe.
500. 1:18:27Ran view activity. The recipe is published in the fresh channel canvases set. I'm generating a photo photo a photorealistic final drink
501. 1:18:40shot now with the drink name, the coaster, and the lettering on the glass. All right, so it's off doing its thing.
502. 1:18:51Did Kyle get his definition? No, I did not. [whistles]
503. 1:19:11How we doing? 36 people still here. That's more people than I thought would be here because I know this is really boring
504. 1:19:18because we're trying to sort of figure this out on the fly together.
505. 1:19:33The first render completed without returning a preview payload. All right. So, it's trying to figure out if it can do
506. 1:19:41if it can do images. And I'm sure it can. I'm sure we've got to uh whatever uh configure stuff to get it to do that.
507. 1:19:49Can Can you confirm we have to download it from GitHub? Download what? Buzz? No, you download Buzz from buzz.xyz. Here, let me show you.
508. 1:20:06Um, if I go here and go buzz buzz.xyz in the upper right hand corner is get the app. And if I click on that, it's it
509. 1:20:20just downloaded it. So here's the DMG on Windows. It takes you to GitHub. Yeah. So I assume since this is a Noster based thing,
510. 1:20:34but there is a an .exe file for Windows. Well, we we can find out here. We can I wonder if we can do that. Let's go to
511. 1:20:43Buzz and find out. Okay, so let's go back to Welcome. And who was the research dude? The research dude is
512. 1:20:54Bumble, right? So, we're gonna say at Bumble. At Bumble, we're going to grab Bumble there. We're going to say, "Can you go
513. 1:21:04research and tell me if someone on Oh, look. It looks like something came back from the speak easy. Um,
514. 1:21:16if you can go research and tell me if someone on a PC can install Buzz app on Windows from the
515. 1:21:32GitHub repo. All right. And then Bumble should respond to that. But I don't see
516. 1:21:50view activity log. Huh? Why didn't he Oh, maybe Bumble's not set up. Maybe. Let me go to agents. And Bumble seems
517. 1:22:05okay. Let me edit him. Harness codeex. Let's let's use buzz agent. We use buzz shared compute model
518. 1:22:21automatic. Save changes. So let's go back to welcome. We'll copy that. We'll go back in here. We'll try it again at Bumble.
519. 1:22:33Like I said, early software is going to be shitty like this. Okay. Now, now he heard so he he was malconfigured.
520. 1:22:47So now he's we can go we can go see what he's up to. View activity. I'll check the project's release and build guidance
521. 1:22:58and verify whether the repository has a path. Let's go see at the speak easy what what we got back. Two replies.
522. 1:23:07The bourbon mirage cooler is rendered where is rendered above. It's not.
523. 1:23:23I don't say All right. Whatever. All right. Let's see what he says. One reply. Oh boy. Look at this. Kyle. Yes, someone on
524. 1:23:36Windows can install Buzz directly from the official GitHub repo. Go to there, download the Windows .exe from the
525. 1:23:43latest release and install it normally. The official read readme explicitly lists Windows.exe build says the app
526. 1:23:51defaults to localhost 3000. So they'll need to choose configure a shared relay in the app or a buzz relay URL. Right?
527. 1:24:00So the builder the builder lab.xyz XYZ to join the right workspace. Building from source is also supported but
528. 1:24:09requires Docker support. Git for Windows Git Bash is needed for agent shell features. All right. Yes, you can. They
529. 1:24:17just haven't created an installer yet. Okay. So, let me come back to you all. Hi. My hair's a mess. Okay. It's 8:56.
530. 1:24:33Um, so tomorrow is Thursday. So if you're in the mastermind, tomorrow is uh practice lab at 10:00 a.m. Mountain time. So noon
531. 1:24:46east coast, 9:00 a.m. Pacific time, headed up by Sid Hardrow and people in there doing their daily practice trying
532. 1:24:55to figure out how to make sense of all this AI stuff. Um, so if you haven't been to one of those meetings and if you
533. 1:25:03haven't created a daily practice for yourself, I'd strongly encourage it. Um, Friday, we've got office hours at 11:00
534. 1:25:12a.m. Mountain time. I have to get tomorrow. All right, cool. I can get Buzz on my phone, but not on my laptop. What?
535. 1:25:23Why not on your laptop? Are you on a PC, Kelly? You can go get it from GitHub. AI first because it's iPhone only. Lol.
536. 1:25:34All right, let's see. I love watching the mastermind. Yeah, the mastermind stuff's really good. Um, all right, kids.
537. 1:25:46Got all this here. What else do we have? If you're on an iPhone, okay, I hate that you can't zoom in on this.
538. 1:25:57It's a big understatement. So, AI first, the the recordings of these are available in the AI salon. Um, where are they, Brandon?
539. 1:26:11>> The recordings. >> The recordings of this show. >> Yeah, they're in they're in this page right here >> to the right next to the
540. 1:26:19>> taps over. Yeah, >> they're replays, but only for the past two weeks. After that, they go up to YouTube.
541. 1:26:24>> Right. Exactly. But if you want to watch this one again and zoom in on stuff, the the replay is available. I can't even
542. 1:26:30get it on my phone. Not sure how to get Buzz to work. Okay. So, we'll keep playing with this. This this is something, like I said, getting
543. 1:26:43an agentic platform like Codeex or Claude Co-work up to speed is complicated. getting something like Buzzup is even more complicated because
544. 1:26:51you're you're now pulling in multiple harnesses and it it's it's just a more complicated thing. I really do think though that the
545. 1:27:02ability to collaborate with multiple agents and multiple people in the same space and and create tools for one another and use each other
546. 1:27:14one another's expertise [cough] in a private secure um ownable portable environment feels like the future to me.
547. 1:27:24All right, beautiful. I've tried my Mac and my iPhone. That makes no sense. It looks awesome. I'd love to collaborate
548. 1:27:32on something like this. Beautiful. All right. Cool. Beautiful. So, what I would encourage you to do, start to find
549. 1:27:38people in the community you might want to geek out with. This is all solvable, right? It's it's it's a little
550. 1:27:44complicated, but it's all solvable. There's videos out there. A lot of people are talking about this right now.
551. 1:27:50Okay. Um, have yourselves a beautiful evening. If you're in the mastermind, I will see you tomorrow at 10:00 a.m.
552. 1:27:57Mountain time, my time. And uh it's uh it's going to be a good day. It's going to be a good day. I hope this was
553. 1:28:04helpful. I know this was this was a lot. [cough] And this conceptually is out there, but things are heading in this direction
554. 1:28:19or they're not. Peace. Thanks. Good night, Ross. Good night. No, Emmy Silver Fox Keel Camp. Peace out everybody.
555. 1:28:26Philacific. All right. Peace. Have a good night. Bye.