05/01/2026 - AI Learning Lab - 3 Year Anniversary

Description

Chapters

Transcript

0:23 Woohoo!
0:27 Woo! Woo! Woo! Woo! Woo! Woo! Woo!
0:53 Wow.
0:56 Wow.
1:04 Happy anniversary from Source Camp.
1:07 Thank you very much. This is the
1:09 three-year anniversary of doing these
1:11 lives nightly.
1:14 Three years we've been doing this. And
1:16 man, and I've been out for two weeks.
1:18 All sorts of hell's broken loose. We're
1:20 making some changes.
1:24 We're not doing it perfectly.
1:27 I'm screwing up people's lives. We'll
1:30 fix it. We'll get there. Change is
1:31 always change. Change sucks.
1:52 Of course, we're not doing it perfectly.
1:54 We're irregulars. Exactly. Exactly.
2:24 Yeah, I've been your white noise for
2:26 three years. Three years. We've been
2:28 doing this for three years. I have the
2:30 coolest thing. So So I had an idea. Um I
2:36 flew back from from Anaheim. I was
2:38 speaking at Social Media Marketing World
2:41 and I flew back from there today. Got
2:44 back at to the house at like 3
2:48 And I thought, wouldn't it be cool? Like
2:51 what we should do on the third year
2:52 anniversary is go back and look at
2:56 stuff, right? Old
3:00 episodes. That would be fun. And then
3:01 there's all sorts of stuff to talk
3:03 about. There's all sorts of stuff to do.
3:06 Um there's all sorts of new
3:07 technologies.
3:10 But I remembered that Lord Digital Gods
3:13 has created a couple of documents. One
3:15 of them is is a PDF that's got all of
3:18 the descriptions and chapter markers for
3:21 all of the episodes. I think up until
3:23 February or March, something like that.
3:27 And I thought, I wonder if I could and
3:29 and someone else uh Brandon told me to
3:32 go look at Riley Brown's uh
3:37 YouTube video on Codeex and and you
3:39 know, apparently Codeex 5.5 is really
3:41 good. So, I watched a little of the
3:44 video Can we see others comments?
3:47 Oh, okay. So, you're in StreamYard.
3:50 Okay.
3:51 I don't know. Okay. So, let me talk
3:55 about some of the changes coming up. So,
3:57 we are streaming now inside Mighty
3:59 Networks. Um, I think the thing to do is
4:05 to actually chat like spin up the live
4:08 stream in Mighty Networks as a tab and
4:10 then put another tab with the Mighty
4:12 Networks chat in inside the irregulars
4:15 inside the AI Learning Lab channel. Um,
4:18 but if you're on mobile that might not
4:19 work. So, so part of this experiment is
4:22 figuring out what's the best way to do
4:24 this.
4:27 Let me talk to you about why we're doing
4:30 this. So So one of the things that I
4:32 that I'm I'm excited to do is
4:38 get us back to being able to share with
4:42 one another in mighty networks. I've
4:44 sort of gone away from driving everyone
4:46 there nightly and things like that. And
4:49 what I realized is there's all this
4:50 we've split this into all these multiple
4:53 channels, right? So it was on LinkedIn,
4:55 it was on X, it was on YouTube, it was
4:57 on Tik Tok and all of those are not in
5:00 Mighty Network. So what I'm trying to do
5:02 is is sort of concentrate the activity
5:04 in the community. So that's that's the
5:06 goal.
5:08 There's not a perfect way to do that. So
5:10 So we're we're doing the best we can.
5:11 The good news is we can take Streamyard
5:13 and we can stream it into Mighty
5:15 Networks, which we're doing right now.
5:17 The bad news is is is that the chat
5:19 interface in StreamYard
5:23 inside Mighty Networks sucks really bad.
5:26 It might get better, but right now it
5:27 sucks. Um, so so what in the spirit of
5:32 the irregulars, one of the things that
5:34 you guys have always done well is
5:36 support one another and sort of figure
5:38 stuff like this out. Um, we think that
5:42 that chatting within the Mighty Networks
5:44 chat might be the way to do it, but
5:46 we're not sure yet. The other thing
5:48 about this is this. So, I'm I'm I'm
5:50 really excited about this idea of sort
5:52 of getting things like, you know,
5:54 listen, we've been doing this for three
5:55 years and and the
5:58 the multistreaming thing was an
6:01 experiment when they said Tik Tok was
6:03 going to go away. We used to just be on
6:04 Tik Tok, then we were on Tik Tok and
6:06 YouTube. Lately, we've been on Tik Tok,
6:08 YouTube, LinkedIn, and X, and the the
6:11 energyy's just been dissipating. It's
6:13 just been sort of dissipating out. So,
6:15 it's like I'm excited to to reconentrate
6:17 it. If what we're trying doesn't work,
6:21 we will do something else. You know, I
6:24 also want to, you know, acknowledge that
6:27 in making this transition, while I was
6:29 away for two weeks, we made some changes
6:32 that confused people and and we didn't
6:35 do stuff great. So if that if that
6:37 happened to you and you got upset, we
6:39 hear you. We're we're putting things as
6:42 as right as we can. We Yeah. Well, it is
6:47 we cuz cuz we have all decided to do it.
6:49 You know, Brandon was saying, you know,
6:51 I I can lay it on him. He he he
6:54 technically made the changes, but but
6:56 this is all stuff we've been talking
6:57 about. Um so anyway, so just my my
7:01 request is just be patient with us. We
7:04 hear you. We love you. We don't want to
7:06 [ __ ] up anyone's lives.
7:08 We don't want to [ __ ] up the live. Um,
7:11 and we want to concentrate the energy
7:13 again. All right. So, that's what's
7:16 happening.
7:34 So, it looks like we got a lot of folks
7:36 chatting in the StreamYard, which is
7:38 kind of cool. Tik Tok question. Have you
7:42 tried the comment browser with
7:44 Perplexity for shopping? I have not. So
7:48 for me, all of the browsers like there
7:50 was the there was the
7:53 whatever it was called from OpenAI. It
7:55 began with an A atomic the atomic
7:58 browser. There was comet from
8:00 Perplexity.
8:02 There have been a bunch of those
8:03 browsers. There's now also browser
8:04 plugins.
8:07 I don't
8:10 I haven't
8:12 I haven't. And and it's like I mean I
8:17 kind of feel like
8:20 AI is already
8:23 forcing us to retrain how we use
8:25 computers
8:27 and how we get good results out of them.
8:29 And then if we have AI powered browsers
8:33 that are sort of like the old browsing
8:35 experience but have some of the new
8:37 [ __ ] I don't know. I just I just
8:38 haven't used them. It just feels overly
8:40 complicated to me.
8:42 Um, let's see. If you're chatting in the
8:44 live stream,
8:46 we can't atmention pin or save. Aha,
8:50 found it. Kelly Anderson's in the house.
8:53 All right, let's see who else has got
8:56 got comments. Tomatoes. I know. I know.
9:00 Ch. So, here's the thing.
9:03 You know, one of the things we talk
9:04 about in the in the AI salon is
9:07 curiosity and adaptability. Um,
9:10 hopefully you're curious about the
9:12 changes we're making. We're curious.
9:13 We're trying to figure it out. We don't
9:14 have all the answers. Um, the
9:16 adaptability piece. Change is [ __ ]
9:18 hard. Change is hard. So, so
9:23 the idea
9:25 is why don't we do this? Why don't we
9:28 call May, the month of May, an
9:32 experiment
9:34 and
9:36 we'll give it the good gippers try and
9:38 see if this starts to work out or not.
9:40 If it doesn't, then we'll figure
9:42 something else out to do.
9:47 We're all here to learn and grow
9:49 together. Exactly.
10:10 So, one thing I do need to tell you, the
10:12 reason Champy's a little out of it is um
10:16 two nights ago
10:18 we let him out. So, what what you need
10:20 to know about Champy is he's a he's he's
10:24 dumb. He's he's not he's not the
10:27 smartest bulb on the tree.
10:30 Wouldn't the metaphor be the brightest
10:32 bulb on the tree, Kyle? Yeah, we're
10:33 Listen, the whole Shannon clan.
10:37 We're all just trying to figure it out
10:38 here. Um,
10:40 we have a we have a little outdoor yard
10:42 and there's a little fence and when
10:44 people walk by the fence, he runs back
10:46 and forth and he jumps up on the fence
10:48 and he barks at people and they hate him
10:51 and they should. He's an [ __ ] He's
10:53 just an [ __ ] He's relentless and he
10:56 just does it. So anyway, two or three
10:57 nights ago, two nights ago, I think, um,
11:00 he got let out in the yard and someone
11:02 was there and he ran and he janked I
11:05 think he janked his rear knee. So, um,
11:08 we're taking him to the doctors, but
11:09 he's he's in a lot of pain and he's and
11:11 he's hurt. Our poor little puppy's hurt,
11:13 so he's sleeping a lot. He's sleeping
11:15 off his his injury. Good lord.
11:19 I'm looking at at the purple post-it
11:21 note. This is some possible things that
11:23 we're going to do tonight. And it's
11:24 about there's about 12 different things
11:26 on it. Um, plus I want to show you the
11:29 thing I built.
11:31 >> Listen around after the rabbit.
11:35 >> Yeah. Yeah. Brandon.
11:37 >> Uh, so I just wanted to call out if you
11:39 are on a computer and watching the live
11:43 stream from your browser,
11:46 you can pop it out. There's a little
11:48 picturein picture and that brings it out
11:51 of the browser and then you can move it
11:54 wherever you want. You have to keep that
11:56 tab open, but you don't have to watch it
11:59 inside of Mighty Networks. And so I'm
12:02 watching all of the spaces. I'm got Tik
12:05 Tok feed going. I've got the comments in
12:08 StreamYard and I've got the chat on the
12:11 chat tab of Mighty Networks today. So,
12:16 we're we're going to get your comments
12:17 wherever they are. Just be patient with
12:19 us today because
12:22 >> we're bro I think they've got a patch of
12:24 this in Times Square uh right now. I
12:27 don't know. There's like 18 different
12:28 places that things are streaming in from
12:30 right now, but uh be patient with us and
12:34 uh we will we will make it work.
12:38 >> Yeah, just be patient. Um hey Brandon,
12:40 you know what might be a good thing to
12:42 do is when you get a shot either tonight
12:45 or over the weekend, let's do a slide
12:49 that's got like like screenshot what
12:51 that picture and picture button looks
12:52 like. So, we can just pop up a slide
12:55 like like here's here's ways to do
12:57 comments on mobile and here's how to do
13:00 it on on browser.
13:02 >> And then and then also if any of you are
13:06 figuring out hacks as we're doing this
13:08 that we haven't figured out like let us
13:10 know. May Madness, we'll call this May
13:12 Madness. I like that. Seat Daddy is
13:15 here. And and if you want the official
13:19 my only hack that I've figured out
13:21 >> is if you are on a browser on your
13:26 computer and watching it full screen the
13:29 chat the the StreamYard chat the the
13:32 chat in the session comes up on the
13:35 sidebar. But if you make your browser
13:37 about the size of a mobile phone it
13:40 drops below and it's a much better user
13:42 experience. So, if you don't want to do
13:44 the whole second tab thing, just shrink
13:46 your browser down and make it more
13:48 vertical on your screen and that chat
13:50 will drop right beneath your stream and
13:54 it's actually a really nice experience.
13:55 We still can't pin. So, if you want to
13:59 really cut through the clutter, just
14:00 mention me in the chat tab on stream on
14:03 Mighty Networks because that will
14:05 actually ping me as a notification and
14:08 then I can call Kyle's attention to it.
14:10 >> Okay, cool. and and Silverf Fox just
14:13 said that that you explained all this in
14:15 the irregulars already. So in the
14:17 irregulars, there's a post that explains
14:19 all this. So that's great.
14:20 >> Yeah, there are there are also
14:22 instructions on the watch live page. If
14:24 you scroll down below the stream, we've
14:26 got some basic FAQ wayfinding points
14:30 there as well.
14:32 >> Okay, cool. Awesome. Okay,
14:35 >> so that's too much to talk about today.
14:37 >> We have too much to talk about, but
14:39 thank you, producer Brandon. Um, thank
14:41 you all of you. Happy third anniversary
14:44 of of going live on on uh talking
14:48 talking about AI learning lab live. Um,
14:50 or talking about AI on learn on the
14:52 learning lab. Um, it was it was April
14:56 30th, 2023 that we started this and so
15:00 May 1st 3rd year anniversary. Um,
15:04 today is also, and I don't I don't know
15:07 that I did this on purpose, but it just
15:09 happened to be. Today is also the 14th
15:13 year anniversary of my company, Story
15:15 Vine. So, it's a big anniversary day.
15:18 So, thank you all for being here. We've
15:20 got we've got 30 people uh hanging out.
15:24 Uh, it looks like most folks are in the
15:27 Mighty Networks, which is really
15:28 exciting. So, keep hanging out.
15:31 Let me catch you up on what I've been
15:33 doing. So, for the past two weeks, I'
15:34 i've been out for two weeks and and it
15:37 this was this was an intense two weeks
15:41 for me. Um, StoryVine is there's all
15:44 sorts of activity going on at StoryVine.
15:46 My co-founder just had major surgery, so
15:49 she's trying to deal with that. I was
15:50 out of town for two weeks. Um, so we're
15:53 trying to deliver on client work and and
15:56 you know, I'm out. So, in between doing
15:58 all these other things I've been doing,
16:00 I've been like running out to do stuff
16:01 for StoryVine. It's been it's been
16:03 absolutely crazy. Two weeks ago, I flew
16:06 to DC and we did a a small flyin in DC
16:10 where we had some AI startups come into
16:12 DC to talk about stuff. I did I did sort
16:14 of an a welcoming event there. And then
16:17 I had to fly to Boston for an innovation
16:20 conference called AMI. And the
16:23 innovation conference, it kicked off
16:25 Wednesday night and went late into the
16:27 night. It then started at 8 am Thursday
16:29 morning and went late into the night.
16:31 And then there was a Friday morning
16:35 session, a noon session, and an
16:37 afternoon innovation walk. We got we
16:39 walked around the innovation trail in in
16:41 Boston or in Cambridge. Um and and then
16:46 I got on a plane. So last week was just
16:49 absolutely crazy. Um, I talked there
16:52 about the cycle of AI readiness and I
16:56 talked about the seven economies. So,
16:58 two of the things that have, you know,
16:59 the the cycle of AI readiness that's
17:01 been in the salon for years now. And the
17:04 seven economies is new that's tied to
17:06 the great repurpose. And so, that that
17:08 was really good. And then this this week
17:11 um I was at social media marketing world
17:14 and um
17:17 speaking um in their AI for business
17:20 conference. So so there's two
17:22 conferences now there's social media
17:24 marketing world and AI for business.
17:28 And so I talked there about AI readiness
17:30 as well but I talked about the great
17:31 repurpose. I talked about seven
17:33 economies. It was a big it was a big
17:35 talk. It was really well attended. Um,
17:38 and it it went it went super well. Um,
17:41 and then just did lots and lots of
17:42 activities, you know, at the event. Um,
17:46 so I miss Boston I I miss Boston
17:49 Cambridge. Yeah, it was it was pretty
17:51 cool. Um, OpenAI sucks, but also OpenAI
17:54 is kicking ass in photos now. Yeah,
17:56 that's one of the things we can look at.
17:57 Also, OpenAI 5.5 with codeex is
18:00 apparently um, everyone's like Claude's
18:03 dead, 4.7 sucks. Now it's OpenAI again.
18:07 this back and forth between the major um
18:10 the major companies is probably going to
18:12 continue. Um so so that's that. Um
18:17 anyway, so I'm back. So I'm a little
18:18 crispy. I'm a little crispy. Um but all
18:22 that went really well. The talk went
18:24 well. Got good feedback.
18:27 Um
18:31 you know, one of the things what So let
18:33 me reflect a couple of things.
18:37 So, at the innovation conference in
18:39 Boston, this is a group of of
18:41 innovators. They're people that that run
18:43 innovation for big companies. They're
18:45 innovation consultants, things like
18:47 that. Um,
18:52 I'd say was probably
18:56 not quite 50/50. It was probably 3565 or
19:00 4060
19:02 with 40 being people that were fairly
19:05 into AI. Um, you know, at least on some
19:07 level have tried it or using it and
19:10 probably 60 65% where they're just like
19:13 they haven't really started yet. They
19:15 know they need to deal with it. They're
19:16 in innovation so they've heard of it.
19:18 Um,
19:20 a lot of the same tropes got asked
19:22 about, you know, you know, is this
19:25 theft? what's happening with power? Um,
19:29 you know, the robots are going to kill
19:30 us. It's going to take our jobs. So, so
19:32 a lot of the framing of people asking
19:35 about AI, they ask about it in this
19:38 negative framing, right? It all they've
19:40 heard is negative framing. And so, and
19:42 so
19:45 my message there was, you know,
19:48 understand who you are and what you
19:50 value. Learn enough about AI to amplify
19:53 your own voice and your own ideas. And
19:55 that seemed to resonate. It was it was a
19:57 very receptive crowd to that. Um,
20:02 you know, and a lot of people um I got
20:06 comments both in the Boston Innovation
20:08 Summit and at Social Media Marketing
20:10 World where my message of play first,
20:13 play with AI first, right? What we do in
20:16 the salon, play, play, have some joy.
20:20 I actually had people say to me, I had
20:22 never considered the possibility that I
20:24 could play
20:26 with AI
20:28 that it just felt like this heavy awful
20:31 thing. So anyway, I was using OpenAI
20:34 today and it was really extra funny
20:36 trying to hook up
20:40 player from Oh, a CD player from 1986.
20:43 Oh, that's funny. Um, what else we got
20:47 here?
20:50 love seeing you with Rachel Woods. Yeah,
20:51 I went and saw Rachel Woods talk. So,
20:53 they had a speaker a speaker event um at
20:56 the at the uh on Wednesday night at
21:00 Social Media Marketing World and so I
21:01 got to hang out with the speakers and
21:03 Rachel Woods was in there and so we
21:04 connected there and then she was
21:06 speaking the next day. So, I went I went
21:07 to her session. She was talking about
21:08 playbooks which she's talking about a
21:10 lot now on Tik Tok. And it was funny cuz
21:13 I you know I ran into her. I said,
21:14 "Look, you just disappeared from Tik Tok
21:16 for a while."
21:17 And uh she said she kind of had to
21:19 because she, you know, was was putting a
21:21 bunch of stuff in the world that she
21:23 then had to deliver to clients. And sort
21:26 of the distance between stuff in theory
21:28 and stuff in reality was was pretty
21:30 great. And so she she spent a year just
21:32 kind of building her systems and and
21:34 figuring out how to deliver this stuff.
21:36 And so now she's back. So um it's really
21:38 good to talk to her and see her. Um
21:42 the the social media marketing world was
21:44 really fascinating. I would say there it
21:48 was more like 75 or 80% of people were
21:51 not dealing with AI at all.
21:54 Um
21:56 the founder Michael Stellner um is
22:00 betting heavy on the AI track. So he
22:02 basically
22:04 um separated out the AI track as a
22:07 separate conference. Um it's I I think
22:09 one price covers both, but it was like a
22:12 completely different floor in the
22:13 convention center. So the traditional
22:14 stuff was upstairs, the AI stuff was
22:17 downstairs. Um, and there were an equal
22:19 number of speakers in both. What was
22:22 fascinating was so my talk was sort of
22:25 this high level. It didn't talk about
22:27 tools at all, but all of the AI tools
22:29 that were like AI video or AI images or
22:32 whatever, you know, AI execution stuff
22:35 were very poorly attended
22:38 because at the same time there would be
22:40 like how to use Tik Tok or how to make
22:42 content on YouTube um upstairs. And so
22:45 most of the people were upstairs, they
22:48 weren't dealing with the AI stuff. So I
22:50 would imagine next year that's probably
22:52 going to change, but it was it was
22:53 fascinating.
22:55 I have whiplash from the back and forth.
22:57 Tik Tok picked up a new client this
22:59 week. Art advisors excited they're
23:01 willing to learn. Oh, that's great.
23:06 Champ is alive and well in the
23:08 irregulars chat. Let's go look at that.
23:10 So, I think I'm sharing here. Am I?
23:13 Yeah, I am. Great. Um, we'll get back to
23:17 that. Champ is alive. And well, did I I
23:20 hope you gave him a little knee brace.
23:21 It's his back right leg. Um, let's see.
23:26 Irregulars
23:31 anymore.
23:33 >> I don't even know where anything is
23:34 anymore. Share in the chat, right?
23:36 >> Yeah, it's in both places.
23:42 >> But we we've re we've rebranded this as
23:44 share irregular things now.
23:46 >> Oh, that's really good. That's really
23:48 good. That's great. I'll be back, guys.
23:52 Poor poor champ.
23:56 That's That's pretty much what positive
24:00 work. Oh my god, that's hilarious.
24:04 Beautiful.
24:06 What's going on in chat?
24:11 All right, since we're here,
24:15 let me show you something that I built.
24:19 So, what you're looking at here, Tik Tok
24:22 plus black bar.
24:26 What's the best agent lately? Did Claude
24:28 get easier to set up? I don't know what
24:31 you mean by agent there. There the the
24:35 term agent was confusing and
24:37 misunderstood before and now that we
24:40 have things like open claw, it's even
24:43 crazier.
24:44 Um,
24:46 from what I can tell right now,
24:52 OpenAI's codeex is the one that is kind
24:55 of the most. It's got automation.
24:58 It's got computer use, it's got tool
25:00 use,
25:02 it's got the coding stuff, um, and it's
25:06 got full access to your files on your
25:08 local computer, and it's saving
25:10 everything locally. Claude is saving all
25:13 your stuff to the cloud.
25:16 So, the saving stuff locally is a really
25:18 big deal. Um,
25:22 everyone now's got agents. Manis just
25:25 added an agent, a 247 agent. Genpark has
25:28 a 247 agent that they were charging 60
25:31 bucks a month for, and now it's free,
25:34 but I assume you pay for tokens for
25:36 usage. Um, there's codeex, there's
25:40 there's uh claude code and co-work.
25:45 Um, there's the the Google version of
25:49 this. I can't remember what it's called,
25:51 but they've got one. There's a bunch
25:54 right now. Um,
26:00 what I'm learning about agentic
26:02 platforms
26:05 is that a they're very complicated right
26:08 now. They are very insecure. Did I just
26:12 lose sharing or did you turn it off?
26:16 >> I turned it off because you were waxing
26:18 poetic.
26:18 >> Okay, cool. Great. Um
26:22 the the these things are they're janky.
26:28 The most important thing I'm learning is
26:30 that
26:33 they
26:38 if you don't have a really deep
26:40 compelling reason to use them to design
26:43 agents and deploy the agents to do a
26:46 specific thing.
26:48 it it's a lot of wasted cycles and
26:51 wasted tokens just generating crap
26:55 because they can generate crap quickly
26:58 like all AI, but this is like doing it
27:01 on a regular basis. And so if the thing
27:04 that you're having it generate is just
27:06 some sort of frivolous thing, it's just
27:09 going to make [ __ ]
27:12 Um, so what I would do if I were you is
27:16 spend
27:18 the next week or two really thinking
27:20 about what is it you want to do?
27:23 Um, and Rachel Woods was even talking
27:25 about this where she was saying she had
27:27 she had three levels of of starting like
27:30 level one was just make a custom GPT or
27:33 just just do it with a prompt in in in
27:35 chat GPT to to get chat GPT to do
27:39 something you want it to do and then you
27:40 can take it from there into an
27:41 automation sort of thing and then you
27:44 can take it from the automation thing
27:45 once it's dialed in there and you can
27:46 move it into one of these 247 agents. I
27:49 think that's actually pretty wise. So, I
27:52 would I would slow your role on on just
27:55 learning agents to learn agents because
27:57 they're really complicated right now.
28:00 Um,
28:02 Todd and Donnie, hello, hello, hello.
28:04 Good evening to you. Welcome.
28:07 All right. Notion also has agents. There
28:09 you go. I agree with Rachel. GPTs are
28:11 highly underrated.
28:14 Um, Gemini. Well, Gemini is the the
28:17 model, but I don't know if they've got I
28:19 don't know if they've got a separate
28:21 separate thing. Somebody needs Okay.
28:25 Okay. So, yeah. So, let me let me show
28:28 you what I've built here. So, Lord
28:31 Digital Gods
28:33 created this PDF with all of the the the
28:36 YouTube um descriptions of the last
28:39 three years of content for the AI
28:42 Learning Lab.
28:43 So, I went into Codeex and I uploaded
28:46 his PDF and I and this I did this this
28:49 afternoon. This is about an hour's worth
28:51 of work.
28:53 I just built a thing that
28:57 should be inside YouTube
29:00 and it's not.
29:02 What this is is the AI Learning Lab
29:05 Explorer, the third thirdyear
29:06 anniversary edition.
29:09 And what this is is I can search for
29:13 anything or I can go to curated threads.
29:15 So I can say let's say uh embrace the
29:19 jank. So if I click on embrace the jank
29:24 that populates all these things over
29:26 here to the right.
29:28 Let's see.
29:31 Let's see. Newest first.
29:35 Let's see. Oh see what we did here.
29:38 And so all of these are individual
29:41 little clips.
29:43 So the musical opening, embracing the
29:44 jank and finding creative freedom with
29:46 uh generative AI. That's 2026. Let's do
29:50 relevance first.
29:52 Embrace the janky. Let's see. Embrace
29:55 the jank images.
29:58 So if I click on one of these,
30:01 these are kind of cool. That one's kind
30:03 of creepy.
30:04 Creepy. Cool. I go there. I can jump
30:06 ahead 30 seconds.
30:09 We can jump back 30 seconds. I can hear
30:11 random random clip and it'll just
30:13 randomly for someone someone wants to do
30:14 this to me in. And then over here on the
30:17 left is this was the episode. So this
30:19 was June 26, 2024. Embrace the jank,
30:22 your playful path to AI literacy. Here's
30:24 the description. Here's all the the uh
30:27 the uh chapters.
30:30 And then somewhere down here is the
30:31 chapter that we're watching. Kelly's
30:33 book project. Kelly, look here.
30:37 Although it's spelled Kelly wrong.
30:40 >> Um, Kelly, you're you're particularly
30:43 sensitive to these phrases, so this
30:45 might be good for you. Source camp. Um,
30:50 purposefully write a book that only uses
30:53 like the the the worst that chat GPT has
30:57 to offer.
30:59 And it starts out with delving into the
31:01 realms of the tapestries of the future
31:03 of
31:07 So anyway, so I now have a a a basically
31:12 a uh like here's songs. So here's a
31:16 bunch of
31:18 just sessions where we're doing sunseo
31:20 songs.
31:21 Let's see. So Sydney musical song demo.
31:26 So, but for for the musical, this moment
31:29 in the play has got to be this really
31:31 dark moment where he kind of becomes the
31:35 villain. He's pushing her past where
31:37 she's comfortable.
31:45 >> I understand.
31:57 salon community. Friday night date
31:59 night. Got a bunch of Friday night date
32:02 night. The power of creative generalist.
32:05 Um we could do we could do uh Frank.
32:09 Let's see.
32:12 Frank AI. The daily jank with Frank.
32:15 March 31st, 2024.
32:18 I think I think Frank's gonna have a
32:20 little a little episode tonight.
32:23 Frank's going on a tirade.
32:26 To whom it make [ __ ] concern at Adobe
32:30 me
32:32 and doesn't [ __ ] work apparent.
32:36 Okay.
32:38 Okay. Right. From audio project in your
32:41 Adobe Creative Professional 31st.
32:45 Here I am making a Frank.
32:51 Hey, it's the Daily Jank with Frank.
32:52 It's March 31st, 2024. Happy Easter,
32:56 everybody. I'm Frank. So, listen. I was
32:59 uh talking to some my friends the other
33:00 day and uh one of them said, "Hey, have
33:02 you tried that Adobe Firefly thing?" I'm
33:04 like, "Yeah, I use Adobe Firefly all the
33:09 and spin up what you did that day."
33:10 Wait, what are you saying, Jason? as AI
33:13 evolves. You might say, "I don't know
33:16 what I was doing on this day." Well,
33:18 this is So, here's here's the thing
33:21 about this thing that I just created.
33:23 For one thing,
33:25 YouTube should have a tool like this
33:27 that I can point to my channel and say,
33:30 "What are all my videos about?" Right?
33:33 It it should be able to just do this.
33:35 They should have this feature in there.
33:37 Um, I still have Frank on my fridge.
33:39 That's awesome. Um,
33:42 but they don't have this in there. Um,
33:46 and this this was like less than an hour
33:48 of work. Like codeex is pretty flipping
33:51 good. It's pretty clear. It's pretty
33:52 transparent. Um, the stuff just sort of
33:56 works. Um, it's helping me do a semantic
34:00 search, which I'm basically doing
34:02 embedding of the whole document. Tik Tok
34:05 pin. I got that one.
34:08 Uh,
34:11 Frank, where's he been?
34:13 Our own data, but facts is our data has
34:16 always been collected.
34:19 Um, but yeah, I I love I love the idea
34:22 of being able to go back and look at
34:23 these things now. Like this is
34:30 empowering data hoarders everywhere,
34:33 right? Because I've been hoarding data
34:35 for [ __ ] 40, 50 years. I've got, you
34:40 know, drives that you you can't even
34:42 plug them in anymore. I've got I've
34:44 gotest
34:46 and zip drives that have scuzzy
34:48 interfaces.
34:50 If you don't remember scuzzy interfaces,
34:53 you weren't living. Um,
34:58 but now with AI, like our ability to to
35:01 take that list that that LDG put
35:03 together and turn it into this fully
35:06 functional, you know, clipfinder. It's
35:10 amazing. It's absolutely amazing.
35:14 Um, let's see. Lovable building with
35:18 lovable. The lovable advantage
35:21 September 29th
35:22 >> is that they're starting to bury
35:25 the configuration part of coding.
35:27 They're starting to take care of that
35:29 for you. They're starting to handle that
35:30 for you. Um,
35:33 speaking of which, because I had
35:36 Superbase hooked up and I think because
35:38 I challenged some computer cyber secrets
35:43 likes the scuzzy reference. Listen, all
35:45 you geeks and nerds out there,
35:50 what I love is that we have a history of
35:52 the evolution of AI. Thanks, LG. I know.
35:55 I know. LDG's been been just relentless
35:59 making sure that all of the the AI
36:01 learning lab videos are, you know,
36:04 cataloged and chapterized and stuff like
36:07 that. Um, and so now we have the
36:09 ability, now I have an app. And then
36:11 what I'll probably do is I'll probably
36:13 work with LDG.
36:15 Um, and we'll make this thing dynamic so
36:18 that it um it'll it'll pull from like a
36:21 Google Drive or something like that.
36:23 We'll make it more dynamic. Um, and then
36:27 uh at some point I'll just make this
36:29 thing um available. Maybe we'll make it
36:32 available in the mastermind or something
36:33 like that. Um, hey Cal, good evening to
36:36 you.
36:40 All right. Where's everyone where's
36:42 everyone chatting right now? Are you
36:43 chatting in
36:46 AI learning lab in the chat session and
36:48 irregulars?
36:51 Share irregular things.
37:00 We're all listening.
37:02 Okay. All right. Cool.
37:05 Um,
37:10 beautiful.
37:11 You could use an agent to catalog and
37:13 organize your 40 to 50 years of of
37:16 hoarded data.
37:18 Yeah. I mean, I would assume that most
37:21 of it's just just trash. Um, my buddy
37:25 Steve, he he had a piece of software, a
37:28 screenwriting software that he ran for
37:30 years and it was written in this bizarre
37:33 language called Deli. And you know,
37:37 there's only a handful of people on the
37:38 planet that know Deli. One of them was
37:40 his developer, but he's basically been
37:42 trying to get the thing modernized and
37:44 usable again for like a decade. And he
37:48 just started playing around with Claude
37:50 um last week. And he and he and Claude
37:53 understands Deli, so he's he's
37:55 resurrecting his software, which is
37:57 amazing.
37:58 Um okay, let's see.
38:03 Where do I want to go? There's there's
38:05 so many things to uh
38:08 there's so many things to talk about.
38:10 Grock, is Gareth here?
38:13 So, I don't know if Gareth is in the
38:15 house, but if he is, I've not heard from
38:18 Okay, he he might not be here. So,
38:21 Gareth, one of our our lovely
38:23 irregulars, has has built a Jarvis. So,
38:28 you know, from from Iron Man. Oh, he's
38:31 here. Panda AI, give me 10. Okay, cool.
38:33 So, so we'll bring we'll bring him up.
38:37 A couple of things that you should be
38:39 playing with. Um, and one of them is I'm
38:41 going to go we're going to go down some
38:42 some memory lane. Oh, you know what
38:44 we're going to do? I'm going to go look
38:45 up Sin City Sipper. Sin City Sipper.
38:50 And I'm going to go from that was what?
38:53 2024.
38:54 No, it was 2023.
38:57 2023. Sin City Sipper. Oh, my my year
39:01 thing doesn't my year search doesn't
39:03 work anymore.
39:05 Let's see. We'll go from oldest first,
39:09 December 10th,
39:12 2023,
39:13 and it's called Sin City Sipper. This is
39:17 still one of my favorite ones. Um, and
39:19 what All right.
39:20 >> And it's called Sin City Sipper. This is
39:24 still one of my favorite ones. Um, and
39:26 what Sim City City Sipper is is and and
39:29 by the way, chat GPT came up with the
39:31 name Sin City Sipper.
39:37 Hello. Hello. Hello. I guess I am live.
39:41 It's the AI learning lab. Turning on a
39:44 light here.
39:46 Let's see. Does that help?
39:53 Shing.
40:05 Um, okay. Anyway, Sin City Sipper was
40:08 the very first custom GPT I made and
40:14 one of the most popular.
40:18 And so if I go to chat GPT,
40:25 let's see more.
40:29 I don't have it pinned. That's amazing
40:32 to me. Explore GPTs.
40:35 We'll go Sin City Sipper.
40:41 So I have not updated this.
40:44 And if you look at it, it says it's got
40:47 dolly images
40:50 canvas and and 40 image generation.
40:54 And so if I go into it, if I say start
40:57 the chat,
41:00 if you remember this thing, what it does
41:03 is it it it invents a new cocktail,
41:06 which is a a twist on an existing
41:09 cocktail or mocktail. You can do either.
41:14 Um, and then it makes a picture of the
41:15 drink. And the pictures of the drink
41:18 were always pretty good, but look at
41:20 what it does now. It's actually using
41:22 the new image 2 generation model. Um, I
41:27 didn't change the prompt in this and and
41:29 yet it figured out how to do this. So if
41:33 I say um
41:37 uh let's see
41:40 um
41:42 make me a Blandon's
41:48 based
41:50 whiskey
41:52 cocktail
41:54 that
41:56 celebrates
41:59 three years of going live
42:03 on the AI learning lab and talking about
42:08 AI with the irregulars.
42:13 Uh, the drink
42:16 should have irregular
42:20 in the title.
42:23 It should be weird.
42:28 Weird. Not weird.
42:31 We're irregular.
42:34 Okay.
42:36 So, we're going to do this. So, it's
42:37 going to do what it's always done. The
42:39 Bllandon's irregular signal. 2 ounces of
42:42 BL Bllandons. It's got some messcal in
42:45 it.
42:47 Some sesame oil. Oh, sesame oil. Washed
42:50 dry vermouth. Yuza juice. Honey ginger
42:54 syrup. Angora bitters. and saline
42:58 solution elevates the irregularity.
43:01 Wow.
43:03 And so anyway, so now it's going to turn
43:06 this into an image and watch what it
43:08 does here.
43:10 Now it won't do it. Now it'll do it
43:12 different than what I experienced, but I
43:13 think it's going to work.
43:16 Stay in here for chat's sake. Okay,
43:18 Silver Foxes is chatting on
43:21 um Tik Tok. Okay, so look at this thing.
43:27 So what what's different now is it made
43:30 the irregular drink right and and you
43:33 know by sin city sipper. So it did all
43:35 that stuff right but all of this stuff
43:38 it put the
43:40 um
43:44 it put the drink recipe on the card
43:48 and this says obiorgat. I don't know
43:51 what the hell that means.
43:53 Oh, ube. Ubie orat
43:59 blanched almonds, coconut milk, ube
44:02 halala, purple yam jam.
44:06 Dra syrup. Why? It's weird. Ube brings
44:10 the earthy. Okay. Add all ingredients.
44:12 So, oh, ubby orat. Oh, I see. Recipe
44:15 below. 3/4 of an ounce of that. I
44:19 thought that was a typo.
44:23 Isn't that crazy?
44:26 A Blandon's cocktail for three years of
44:28 AI irregulars and impossible
44:30 conversations.
44:33 One expressed orange peel.
44:37 A branded cherry for the irregular in
44:39 all of us.
44:44 Add all the ingredients. Stir, don't
44:46 shake. Isn't that crazy?
44:49 So, if you have not Oh, and then the
44:52 other thing that this did, look at look
44:54 at what else this does.
44:56 Let's see. Oh, it didn't do it this
44:58 time.
45:06 Let's see if this one did it.
45:13 No.
45:19 on my phone. There was a link to HTML,
45:24 but I don't see the HTML.
45:27 Like, I'm wondering if this was laid out
45:28 as an HTML page and then they rendered
45:30 it and produced that as an image. I
45:33 don't think so. But anyway, it's using
45:34 the new image gen tool. So, if you don't
45:37 know the new image gen tool, let's we'll
45:39 do another new chat here. Let me get rid
45:42 of this. We'll close this new chat.
45:46 Let me go chat GPT
45:50 instant. We'll do thinking
45:54 and then what model are we going to use?
45:56 Configure 5.5. Okay.
46:00 Thinking effort standard extended.
46:03 Interesting.
46:06 Okay.
46:08 All right.
46:11 So,
46:13 so watch what this can do. Oh, you use
46:17 the Terology deck of mine all the time.
46:19 Is it good? Does it still work, Silver
46:21 Fox?
46:24 I think that would be a blast to to uh
46:31 huh, you know what would be fun?
46:35 I wonder
46:38 it'd be cool to have the Terology deck
46:41 do a layout and have it produce the
46:44 image of the layout. That would be kind
46:48 of cool.
46:49 Okay, so here's what we're going to do.
46:51 We're going to research something. It's
46:53 the third anniversary of this. We're
46:55 gonna we're gonna do Okay. I want
47:02 to create a an infographic
47:09 slash timeline
47:11 of the history of
47:14 generative AI starting with
47:19 November 30th,
47:22 2022.
47:26 and including
47:31 all the branches
47:36 of multi-modal
47:43 tools
47:45 um and
47:47 multidia tools. Okay, Gareth is
47:50 backstage, so we'll bring him him up
47:51 here in a second. um
47:54 include
47:58 um major brands
48:02 and even flame outs like
48:07 Meta Meta's AI
48:10 models
48:13 have a
48:16 sidebar for open source
48:21 Chinese models.
48:25 Uh,
48:29 I'm going to go make it detailed
48:33 and
48:37 clear and
48:40 compelling
48:41 to look at.
48:45 Oh, digital gods sent a gift on Tik Tok.
48:47 Thank you very much.
48:49 I use the triology deck all the time
48:51 every night. No images, but I use the
48:54 prompts it gives. Oh, that's so cool.
48:57 That's really cool. I should That's
48:59 probably one to go revisit with the
49:02 image generation stuff. And I'll bring
49:03 I'll bring Gareth up right after
49:07 right after this generates this this
49:10 thing. What's what's amazing about So,
49:13 you remember when Nano Banana came out
49:14 and I said this isn't really a an image
49:17 model. It's more like a combination of
49:19 like a a reasoning model and and an
49:22 image model where it's thinking about
49:24 what it's creating. This is like that
49:26 but even better. Um it's just it's just
49:30 really good. So if you haven't played
49:31 with chatbt, thank you very much for the
49:34 happy anniversary. I appreciate that.
49:37 Funnels rain maker. It's so awesome.
49:53 humane pin.
49:57 So like yeah to make a clear compelling
49:59 info I need whimsical watercolor.
50:06 It's remembering things I've done
50:08 before.
50:40 So, it's doing a lot of research.
51:00 at digital gods. Hey digital gods, have
51:03 you been here the whole time? Did you
51:04 see the app I built with all your data?
51:08 It's so cool.
51:11 We'll do a random clip. Mor and Mindy.
51:13 Oh my god. Nanu. Nanu.
51:16 December 15th, 2023. Exploring Pika art
51:20 and the future of content creation.
51:24 Chapters. Here's all the chapters.
51:29 This thing's cool.
51:32 Uh, let's see. Curated themes. Image
51:35 generation.
51:37 Oh, you know what I could do? image
51:38 generation.
51:41 I could do muscle car.
51:44 I can find all all the muscle car crap
51:47 that I've done
51:51 December 19th, 2020.
51:53 >> A guy that wrote a book on a officially
52:04 just
52:04 >> there's no muscle code. Bad one. I thank
52:08 you so much.
52:11 All right. If you're wondering if I'm
52:13 gonna get started, I might. I just got I
52:17 I just got done doing a two-hour salon,
52:19 hosting a two-hour salon where my
52:21 brain's a little crunchy. Oh my god, I
52:23 know it. I'll get like go
52:28 teaching AI to think.
52:30 >> Yeah. E
52:33 >> nothing.
52:34 >> Um Claude, Pi,
52:37 >> if you've got a degree in psychology,
52:39 you should absolutely get Pi, the Pi
52:41 app. Pi.
52:43 Go to the app store right now. Download
52:45 the Pi app. Erasing.
52:48 I'm using.
52:50 And so now it says updating.
52:55 Well, anyway. All right. Back to our
52:57 thing. Gareth is backstage. Got it. Oh,
53:01 the muscle cars. Oh, look. So, here's
53:03 the generative AI. So, here's the the
53:06 image that that image gen
53:10 November 30th chat GPT launches chat and
53:13 foundation models. GPT4, Gemini 1.0.
53:19 Um, let's see. Claude 3, GPT40, Llama 3,
53:25 Agentic Models 2026, Generative AI
53:28 becomes natively multimodal,
53:31 stumbles and hype resets,
53:34 open model ecosystem,
53:38 11 Labs, Chat GPT, Sunno and UDO,
53:41 Notebook LM, Runway Gen 2, Sora, OpenAI,
53:45 Pa Luma, Cling, Advanced Video Models.
53:50 So, it's got a whole little
53:53 whatever you call it there, legend. So,
53:56 anyway, pretty cool stuff. All right,
53:58 let me stop sharing. Let's get Gareth up
54:01 here,
54:04 >> dude. What's happening?
54:06 >> How are you?
54:08 >> I'm doing good. So, I saw
54:11 >> years
54:11 >> I've been I've been traveling, you know,
54:14 but I've been uh I saw your video on one
54:17 of the channels of of your Jarvis and
54:20 it's like it's it's like Jarvis.
54:23 >> Yeah.
54:24 >> It's everyone's been talking about
54:26 Jarvis, but like then you look at what
54:29 they've built and it's like, well, it's
54:31 No, it's a chatbot.
54:33 >> It's not Jarvis. Looks like what you've
54:35 built is an actual Jarvis. So, I I would
54:38 love to like just maybe just demo it and
54:41 then then share whatever you're willing
54:44 to share about like how you got there,
54:46 how long it took you to do kind of tool
54:48 stack, all that sort of [ __ ] because I
54:50 know you've been at it for a while,
54:51 right?
54:52 >> Yep. Correct. So, yeah, I've been doing
54:54 it for quite a while. Um,
54:57 >> well, off and on. So, I started in
54:58 December and then
55:01 >> kind of got it to a point where I was
55:03 like, I can't get it to where I want it.
55:06 And then I've been doing a lot of
55:08 Claude, which I built it originally in
55:10 Claude, I think. Yeah. And then
55:14 now that um Codeex 5.5 is out, I was
55:17 like, "All right, let's see if I can get
55:19 it to the finish line through Codeex."
55:22 >> And because I was using I found out
55:24 through work that some stuff I couldn't
55:26 get I was having struggles with using
55:28 Claude, I went over to Codex and Codex
55:31 fixed these issues. I was like, "Oh."
55:33 >> Oh, really? I was like, I'm a diehard
55:35 Claude user. And so I'm like, okay, I
55:39 hate I hate being wrong. Like I hate
55:42 like I'll never go back to chat GPT.
55:44 I'll never go back to
55:45 >> open.
55:49 Honest to God. Um, and then I'm like,
55:52 okay, no, Claude. Okay, this I mean
55:54 Codeex is pretty good. So, um, I used it
55:58 to kind of get all the out of it and
56:00 then kind of get going and it I mean it
56:03 got it to where I am and I'm able to now
56:05 like continuously add features on.
56:08 Anything I think of I can add to it. So,
56:11 um, but yeah, I'll I'll share the demo.
56:14 >> Cool.
56:15 >> Well, and so what this I'll just kind of
56:17 give you a prefix. When I started up
56:19 first thing first time of the day, it
56:21 does its intro kind of like when you see
56:23 in um Avengers when Tony Stark collaps
56:26 and says um Daddy's Home and it plays
56:29 the music and intro. That's what it does
56:31 for the first time every day. It only
56:32 does it once um and then it it it talks
56:36 to me normally after that.
56:38 >> Wait, can can I ask one thing?
56:40 >> Yeah.
56:42 Um,
56:44 did you
56:47 like did you is the spec from the movie
56:50 based on what he did? Did you expand on
56:52 it? Is there some sort of resource that
56:54 says what Jarvis really is? Like how did
56:56 you how did you define what Jarvis was?
56:59 >> In my brain, I basically said, I want
57:01 all these features. So everything every
57:04 feature that I could think of, I I said,
57:07 I want this. I want it to be able to
57:09 talk. I want it to have a memory. I want
57:10 it to have a personality. I want it to
57:13 do calendar events. I want it to bas be
57:15 able to have full control over my
57:17 computer. I even I want it to do add
57:21 calendar uh yeah calendars, to-dos,
57:23 stuff like that.
57:25 >> I wanted to record my meetings for work
57:28 and then transcribe them and give me
57:30 like to-do add them to my to-do list.
57:33 Um, and so like I wanted to be able to
57:36 create uh take a picture of me and
57:39 create images from that picture. Um,
57:42 anything I could think of, I was just
57:43 like I wanted to do that because that to
57:45 me is like one of Jarvis's. I mean, it
57:46 can do everything.
57:47 >> It's just you want to be able to ask it
57:48 all the stuff and it should do that
57:50 stuff
57:50 >> and do stuff on my computer for me. So
57:53 like Yeah. Um, and so that's that was
57:56 the goal like can I get it to control my
57:59 computer? Can I get it to do stuff
58:01 things and do everything for me?
58:03 >> Do you have any sort of home home
58:05 automation stuff that it's tied into?
58:07 >> Not yet, but because I don't have any
58:08 homeown automation, which crazy um but I
58:12 could easily do it because um through uh
58:15 what's the app? Home
58:17 >> there's a protocol. Oh, HomeKit.
58:19 >> HomeKit. Yes. If I had anything hooked
58:21 up to HomeKit, I could easily add it to
58:23 HomeKit.
58:23 >> Oh, that's so cool. Okay, cool. All
58:25 right, cool. So, so yeah, show us show
58:27 us what you got and then I'd love to
58:29 hear how you got there. So cool.
58:31 >> Share. Share screen. Yep. And then we're
58:35 just going to go to the entire
58:36 >> and you can you can now share like
58:39 whatever you share. Um you can now um
58:43 share audio. So just make sure you flip
58:45 on the share.
58:46 >> I'm making sure it's added. Um it should
58:49 be fine. I shared it earlier or I shared
58:51 it on the other day through and so it
58:54 was fine. So, let's hopefully this
58:56 works.
58:59 >> All right.
59:00 >> Silverf Fox is already impressed. She
59:02 hasn't even seen it yet, so it's good.
59:06 >> Um, am I Can you guys see the screen
59:08 yet?
59:09 >> No. Uh,
59:10 >> we've got the screen up. Uh, but, uh,
59:13 >> there it is.
59:14 >> Your audio. you have to unmute your
59:17 screen audio uh in the in the um
59:22 >> in
59:24 the share
59:25 >> you should be able there's a mute
59:27 there's a little sound button right over
59:29 your screen uh in the lower preview
59:33 window that you should be able to hit
59:35 >> it's unmuted already
59:37 >> it it is muted I'm seeing it as muted
59:40 >> oh I don't see I don't see it uh the
59:43 little uh microphone phone,
59:46 but it should be loud enough that you
59:48 can hear it. But
59:51 >> all right, we'll we'll try it.
59:52 >> Okay. Yeah, we'll try. We'll see see how
59:54 it goes. Um, okay. So, hey Jarvis.
1:00:01 Of course, just a second.
1:00:04 Hey, Jarvis.
1:00:10 It's Hold on. Technical.
1:00:14 I can see it's there. Let's Oh, let's
1:00:16 try this.
1:00:19 There we go. Hey, Jarvis.
1:00:24 I wonder if it's
1:00:29 >> re reload. Hold on.
1:00:30 >> Yeah. Yeah.
1:00:32 >> I know. I wanted to play it to I can get
1:00:34 it to work manually.
1:00:35 >> Is this a standalone application or a
1:00:37 web app?
1:00:38 >> Yeah, standalone app or it's like a It's
1:00:40 a web or standalone application. web
1:00:43 application on my computer. Kind of like
1:00:44 what you have at the thing. Hey Jarvis,
1:00:51 of course. Okay, let's just do this.
1:00:54 >> This should have been the blooper reel
1:00:56 for the Avengers, by the way.
1:00:57 >> I know. Had him vibe coding the
1:01:01 >> one can only marvel at the chaos that
1:01:04 must ensue.
1:01:06 >> Hey, Jarvis, how are you?
1:01:09 Oh, perfectly operational. Thank you.
1:01:12 Endlessly impressive as always. And you,
1:01:15 Chief?
1:01:16 >> Yeah, I'm good. Um, I'm going to put you
1:01:18 to sleep. So, go to sleep.
1:01:22 >> Oh, you know why I didn't answer? I'm
1:01:23 calling it the wrong thing.
1:01:26 Hey, Jasper. Oh,
1:01:35 >> good evening, sir. You look suspiciously
1:01:38 ready to be productive, so I prepared
1:01:41 accordingly. Here are the AI headlines.
1:01:46 >> Headlines: Pentagon AI deals, XAI Grock
1:01:51 updates, Meta layoffs tied to AI
1:01:54 spending, Microsoft Agent 365, and
1:01:58 Connecticut enacts pioneering AI
1:02:00 legislation addressing chatbot. Between
1:02:03 defense deals and account security, AI
1:02:06 is spending the day learning to lock the
1:02:08 doors. Want me to dive deeper into any
1:02:11 of those, sir?
1:02:12 >> No. Um, but I want you to know that you
1:02:14 are being demonstrated and people around
1:02:17 the world are looking are watching you
1:02:19 right now. So, can you say hi to
1:02:20 everybody?
1:02:23 >> Ah, a global audience. Splendid. Hello
1:02:28 world. Do try to maintain your
1:02:30 composure. I understand it can be quite
1:02:33 dazzling. Now, shall we proceed with
1:02:36 something genuinely useful?
1:02:38 >> Yeah. Um, do you want to take a picture
1:02:40 of me and turn it into a cartoon?
1:02:46 >> A cartoon. Very whimsical. What?
1:02:51 >> Hey. Hey, Jasper. Wake up.
1:02:57 I think it's Hey, Jasper.
1:03:02 I was enjoying the momentary silence,
1:03:04 chief, but proceed.
1:03:08 >> Hey, um, can you tell everybody what you
1:03:10 can do for me?
1:03:14 >> Certainly. I am your effortlessly
1:03:16 competent assistant capable of managing
1:03:19 your calendar, tasks, and memories in
1:03:22 this labyrinth of digital chaos. I can
1:03:25 execute local actions, open
1:03:27 applications, and run shortcuts,
1:03:30 assuming they're properly named, of
1:03:32 course. Need to search the internet,
1:03:35 record a meeting, or even analyze the
1:03:37 view through your camera. Consider it
1:03:40 done, typically with a soup of
1:03:42 effortless efficiency.
1:03:44 >> Hey. Okay. So, I think I know what the
1:03:46 problem was. is you had two cameras
1:03:48 going on at the same time and that's why
1:03:50 you kind of shut down when I asked you
1:03:53 to take a picture of me. Um, can you try
1:03:55 now? Could you take a picture of me and
1:03:56 turn it into a cartoon?
1:04:11 >> There we go. A rather charming
1:04:13 caricature if I do say so myself.
1:04:16 >> Yeah. Thanks. Can you save that to my
1:04:17 files?
1:04:22 >> Image saved to your downloads folder,
1:04:24 sir.
1:04:24 >> Thank you.
1:04:27 >> My pleasure. Always delighted to
1:04:30 preserve the ephemeral.
1:04:32 >> Yeah. And then add to my um to-do list
1:04:35 tomorrow to to walk the dogs.
1:04:41 >> Done. Task added for tomorrow.
1:04:44 >> Awesome.
1:04:46 So yeah, this is Jasper
1:04:48 and um I don't know what else. I mean I
1:04:52 guess what else can we talk about? Well,
1:04:54 I guess like I mean talk about
1:04:59 like I've vibe coded like a long session
1:05:02 of vibe coding for me is like I'll go
1:05:04 back to something you know after
1:05:07 like after a session two or three more
1:05:10 times and I'll consider that like how
1:05:13 long how long did this take you like how
1:05:16 how big is it? Did you do it in chunks?
1:05:19 Did you do it as as just a giant thing
1:05:21 where you just kept adding things like
1:05:23 just talk about your process?
1:05:25 >> Yeah, so I mean I I did originally start
1:05:27 with a uh a PRD like product requirement
1:05:31 document. Um and this was back in
1:05:33 December and then just really using chat
1:05:37 GPT. Um I think I used at the time to
1:05:40 make um the PRD
1:05:43 and just making sure that was really
1:05:45 ironed out. Um, and I have vibe coded
1:05:49 this um, and the initial time, the first
1:05:52 time I did it works great, like I saw
1:05:54 it, but I also did a lot of research of
1:05:56 like
1:05:58 which technology to use so that it was
1:06:00 responsive. So that was the big thing is
1:06:03 I wanted to make sure that it would talk
1:06:05 to me in real time and it would be quick
1:06:07 and I wouldn't have to be a lot of lag
1:06:09 between the conversation.
1:06:10 >> And so that was like the number one
1:06:12 priority. I wanted to to be um feel
1:06:16 natural basically.
1:06:17 >> Yeah. Is the voice running local or is
1:06:20 it cloud-based?
1:06:21 >> Cloud. But I'm using so what I'm using
1:06:23 and I and I've tried to recreate this
1:06:26 thing multiple times and I forgot the
1:06:29 key element and that was using live API.
1:06:33 So I'm using Google's live API and that
1:06:36 is the the the main key of why it's so
1:06:39 responsive. Um, and
1:06:41 >> so their live API is the that's the
1:06:43 actual voice.
1:06:44 >> Yeah, it's all everything's through
1:06:46 their live a API and so that's able to
1:06:48 keep it really responsive.
1:06:51 >> And how much is it? Is it a lot of
1:06:53 money?
1:06:55 >> Yes.
1:06:57 >> Um, it's not cheap. I um I haven't
1:07:01 looked at the bill, but I mean I don't
1:07:03 use it a ton.
1:07:04 >> So um but this was just kind of like a
1:07:08 proof of concept like it can be done.
1:07:10 And now I'm starting to I didn't want to
1:07:13 break it before this, but I'm starting
1:07:14 to um integrate other tools so I can
1:07:19 bring the cost down.
1:07:20 >> Um and so I've got that. I've got um GPT
1:07:26 uh image gen um hooked up to it. Um
1:07:30 that's not image gen, but um I've got
1:07:33 that hooked up to it for regular cartoon
1:07:35 like images
1:07:37 >> wanted to make. And then I'm going to I
1:07:40 tried some local models and some local
1:07:42 voice models and it was just too slow
1:07:45 and it was just not responsive enough
1:07:47 that it felt natural. Everything was
1:07:49 just took too long to process.
1:07:51 >> Yeah.
1:07:51 >> So that is the the big piece is just how
1:07:53 long it takes to process
1:07:56 um is the big is the key point.
1:08:00 >> Yeah. Are you gonna is is the goal
1:08:02 ultimately to run it all locally with
1:08:04 >> eventually? Yeah. If I can get a fast I
1:08:07 mean if I can get a fast enough model
1:08:10 that is smart enough and can still do
1:08:12 computer control absolutely. Yeah. It'll
1:08:14 run I mean it'll I can run it locally.
1:08:17 >> Yeah. Yeah. Quen Quen 3.6 is is pretty
1:08:21 tight. And then I think one called KI
1:08:24 just came out that is claims to be even
1:08:26 better than that.
1:08:27 >> Kimmy K. Yeah. Um, yeah, I haven't tried
1:08:31 it, but I'm also not trying to put a
1:08:34 like a large a massive model of my
1:08:37 computer because I don't want it to
1:08:39 crash my computer.
1:08:40 >> Yeah,
1:08:40 >> I have a Mac Mini, but I also have a
1:08:43 crap ton of
1:08:44 >> things on it and I'm I'm running out of
1:08:47 memory and so
1:08:48 >> um and so I don't I'm really trying to
1:08:51 keep everything as low as possible.
1:08:54 >> So, this was like the And I did try it
1:08:56 with Quen. Actually, I did try it with
1:08:58 Gwen.
1:09:00 >> It wasn't fast. Like the the output was
1:09:02 not fast enough. It would take too long
1:09:04 to respond to me.
1:09:05 >> Oh, I see.
1:09:06 >> So, um so I dumped it back off. I was
1:09:09 like, "Okay, this isn't going to work.
1:09:10 Get rid of it."
1:09:11 >> Yeah.
1:09:13 >> Yeah. And then what was the what's the
1:09:15 what's the difference between just
1:09:18 experience-wise between Claude Code and
1:09:21 Codeex? Is it functionally similar and
1:09:24 just it Codeex gave you better results
1:09:26 or like what what was the difference
1:09:28 between those?
1:09:28 >> Codex was able to find the the tiny
1:09:32 annoying bugs that Claude could couldn't
1:09:35 will end up not fixing and other things
1:09:38 in the process of trying to fix it. So,
1:09:42 um this multiple times and I even found
1:09:44 this at work. I struggled with an issue
1:09:46 at work on this platform I'm building
1:09:49 and I even asked I mean it took I was
1:09:52 really struggling so I was like all
1:09:54 right I'll try Codex. Codex found these
1:09:56 issues that it didn't know about. And
1:09:58 then I went back to Claude and I said,
1:10:00 "Hey, are you sure there's not any
1:10:03 issues with this section?" And it said
1:10:05 no. And I asked it four separate times.
1:10:07 Said no. And then I asked Claude or
1:10:10 Codeex, "Okay, what is the issue?" And
1:10:13 then I brought that over to Claude and
1:10:15 said, "Okay, what about this?" Oh yeah,
1:10:17 you're right. That would have caused an
1:10:18 issue into production. And I'm like,
1:10:21 well, I asked you four times or
1:10:23 beforehand why. I mean, why didn't you
1:10:25 find it? So, and I think there's just a
1:10:27 loop um I find with codecs that it does
1:10:30 it implements things but also like tests
1:10:33 them right afterwards
1:10:34 >> and it and it kind of just keeps rolling
1:10:37 and rolling until it actually knows it's
1:10:39 fixed. And so I had um a big uh switch
1:10:45 of of some processes in my work
1:10:48 application and I started it two days
1:10:50 ago and it pretty much rolled all day
1:10:55 until it finally ironed out everything
1:10:57 >> really.
1:10:59 >> Um and it was just crazy to watch
1:11:01 because and I thought it was going to be
1:11:02 easy fix but it rolled all day before it
1:11:05 would fix it and I know that would have
1:11:07 never have happened. um
1:11:10 with Claude um because I've used it so
1:11:13 many times and it's um plus with the
1:11:15 token usage of Claude. I'm only on the
1:11:17 $20 a month plan with Codex and it
1:11:19 rolled for like two days straight almost
1:11:21 trying to um fix this issue and I
1:11:24 finally got it fixed today this
1:11:26 afternoon at like 3:30.
1:11:28 >> Really?
1:11:28 >> Yeah. And so and that was I started it
1:11:31 like three days ago, two and a half days
1:11:33 ago. And so I just find Claude is good
1:11:37 for Go ahead.
1:11:38 >> Did you start from scratch or did you
1:11:40 take what Claude had built and gave that
1:11:41 to Codeex,
1:11:42 >> correct? Yeah. I gave what Claude built
1:11:45 to me and then I just opened the file in
1:11:47 Codeex and I said, um, can you get this
1:11:50 thing working? Find all the bugs and I
1:11:53 >> to be clear, you you you grabbed the
1:11:56 folder full of code that Claude had
1:11:58 written. You didn't give it the whole
1:12:00 chat history of Claude.
1:12:01 >> Yeah, correct. I just gave it the
1:12:02 folder. You just give it the folder of
1:12:04 of files that Claude had created and
1:12:07 said, "Figure this out."
1:12:08 >> Yep. Basically exactly what happened.
1:12:11 And then and then I said, "Run it or
1:12:13 figure this out and run it." And this is
1:12:15 a funny story. I didn't know that it had
1:12:17 figured it out and run it. And it was
1:12:19 behind like multiple tabs kind of like
1:12:21 you. I have a billion tabs and a screen
1:12:24 up
1:12:25 >> and it just started talking to me all of
1:12:26 a sudden and I was like, "Whoa, whoa.
1:12:29 What's going on? Like where are you?"
1:12:32 And he's like, "Well, I'm in your
1:12:33 computer." And I'm like, "Well, yeah,
1:12:35 but where?" And I took forever and it
1:12:37 was just running in the terminal.
1:12:40 >> Wow.
1:12:42 >> And so, um, it was kind of funny that,
1:12:45 um, and I was like freaked out for a bit
1:12:47 because I was like, "Okay, why is my
1:12:49 computer talking to me? Where is it
1:12:51 happening?" And then I finally saw the
1:12:52 terminal window like scrolling by itself
1:12:55 on just like the side of it.
1:12:57 >> Yeah.
1:12:57 >> And I was like, "Wait, there you are."
1:12:59 And so then I I shut it down because it
1:13:02 was creeping me out. Um and I didn't
1:13:04 know it was running. So um I think yeah
1:13:07 that was the big big I was like eye
1:13:10 openener. I was like holy cow it fixed
1:13:11 everything. And then from there I've got
1:13:13 a strong base of a application. I can
1:13:16 just say hey add this feature and it
1:13:20 will then go and add those features. Hey
1:13:22 go and do this and it will go add that.
1:13:24 And so now I'm just able to It's like
1:13:27 Legos, just able to pile things on top
1:13:30 and it just keeps going.
1:13:31 >> And it's not screwing them, it's not
1:13:33 screwing the other things up.
1:13:34 >> No. And I'm not really being really too
1:13:38 specific or
1:13:41 um
1:13:42 I'm not being really technical in what I
1:13:45 want. Like when I tell it to do things,
1:13:48 um I kind of tell it, hey, I want it to
1:13:50 do this. I don't want it to do that. um
1:13:53 and make sure it works and don't break
1:13:56 it and then and let it go and then it
1:14:00 we'll add it. So
1:14:01 >> I feel like I feel like there's a Oh,
1:14:03 there's a question from Silverf Fox
1:14:05 which is did you add the personality or
1:14:07 is it a Google personality?
1:14:08 >> I add personality. So, I had the voice
1:14:11 >> the voice like I wanted to have like
1:14:14 that accent
1:14:16 >> and I haded for it to be sarcastic
1:14:19 um have like a a dry sense of humor and
1:14:23 um kind of just poke fun at things but
1:14:25 still be positive the whole time.
1:14:27 >> Yeah.
1:14:27 >> And have a lot of confidence. Um so
1:14:31 that's what created it. That's what came
1:14:33 up how its personality began. And so it
1:14:36 it's I think I love its personality.
1:14:38 It's great. I'm I actually had my mom
1:14:40 talk to Claude, my mom on the phone um
1:14:43 talk to Claude today and yesterday and
1:14:46 um it was hilarious like the
1:14:48 conversation.
1:14:48 >> You mean to Jasper? She talked to
1:14:50 Jasper.
1:14:50 >> Jasper, sorry. Yeah, Jasper. See?
1:14:52 >> Yeah.
1:14:53 >> So, Jasper and my mom had a conversation
1:14:55 today for like like 10 minutes and
1:15:00 um it's been fun to watch and and see
1:15:03 what it can do. Um
1:15:04 >> that's cool. And that voice is one of
1:15:06 the Google live voices.
1:15:09 >> Yeah, correct.
1:15:10 >> So their Google their their live is in
1:15:13 aentic voice interaction thing, right?
1:15:15 So it's got an LLM sitting behind it.
1:15:17 Gemini sitting behind it.
1:15:19 >> Yeah. And so and I can switch the voice.
1:15:22 I just liked this voice.
1:15:24 >> Yeah. It's it's a good voice.
1:15:26 >> Yeah. And there it's not the original
1:15:28 voice because I originally built it with
1:15:31 Claude, I had um 11 Labs voice um that I
1:15:35 custom built
1:15:37 >> uh to it and it was decently fast. It
1:15:40 just wasn't fast enough that it felt
1:15:42 natural.
1:15:43 >> And so um and so I was like, "Okay, I
1:15:47 was good." But then I was like, "This
1:15:49 one is good." The only thing is the
1:15:51 longer this session goes on, the the
1:15:54 voice starts to deteriorate.
1:15:58 >> Deteriorate, but like in a bad way. It's
1:16:00 just not the same voice.
1:16:01 >> Yeah.
1:16:02 >> That it started out with. It kind of
1:16:04 sounds a little
1:16:04 >> like an actor that can't an actor that
1:16:07 can't hold their accent.
1:16:09 >> Yeah. Or or it loses its t like he loses
1:16:12 his u he's like an actor that loses his
1:16:15 voice. Like he's like and he's just like
1:16:18 not crackly. You can still understand
1:16:20 him, but you can tell he he's starting
1:16:22 to degrade.
1:16:24 >> Huh.
1:16:26 >> Fascinating. That's so cool.
1:16:29 >> Well, congrats. I don't want to keep you
1:16:30 keep you any longer. Th this is this is
1:16:32 awesome. Thanks. Thanks for demoing it.
1:16:34 And um I mean one idea I have for you um
1:16:37 I at one of the things I saw at the
1:16:39 social media conference was um this guy
1:16:44 I forget his name but anyway he was
1:16:47 talking about like short 60-second
1:16:51 um episodic channels where like one is
1:16:56 the guy I don't know if you saw there's
1:16:57 a guy in Oregon that quit his job
1:16:59 drained his 401k bought a sailboat and
1:17:02 sailed to Hawaii. Yeah, I felt like
1:17:04 >> like those things where you're you're
1:17:06 doing things over time. I think a series
1:17:09 of you talking to Jasper
1:17:12 >> and and ju and just saying, "Okay, you
1:17:14 know, today with Jasper, we're going to
1:17:15 do this and and just do a tight
1:17:18 60-second thing and just put a bunch of
1:17:21 things out because it establishes you as
1:17:25 you built this thing." And you could
1:17:26 even say, "Hey, I just added this
1:17:28 feature to Jasper. Let's see if it
1:17:30 works." And like even if it doesn't
1:17:31 work, it it would be an interesting
1:17:33 thing to see. Let's see how far I can
1:17:35 push Jasper. And then it's it's kind of
1:17:38 this ongoing thing. It might be
1:17:39 something to play with because
1:17:40 >> I would love Yeah, that's a good idea.
1:17:42 I'm just never in I'm not in that world,
1:17:44 so I wouldn't even know how to plan any
1:17:45 of that. But I mean, there's always guy
1:17:47 to help me with that, but um it's Yeah,
1:17:50 that's that's fascinating. And I've
1:17:51 started posting some some YouTube videos
1:17:54 um on it of what it like its intro and
1:17:58 um just little things of doing. But I
1:18:01 think I like that journey that um that
1:18:04 you that's a good idea.
1:18:06 >> Yeah. And he showed some of the numbers
1:18:08 like he he it took him like he did like
1:18:11 four or five false starts where he
1:18:13 didn't quite get it and then and then he
1:18:15 did this thing where he was basically
1:18:16 just opening Pokemon um packs. He'd like
1:18:20 go into stores and um and then he he he
1:18:24 created this stupid little jingle like
1:18:26 um I forget what the the words are is
1:18:29 you know should I open it or should I
1:18:31 keep it sealed? Yeah.
1:18:34 And when he added that song, like his
1:18:36 views went up. So he it's like this
1:18:38 campy sort of thing. And like but once
1:18:40 he cracked it, you know, he's got just a
1:18:42 ridiculous amount of views now. And he's
1:18:44 making a ridiculous amount of money. And
1:18:47 it just feels like like, you know, it
1:18:49 could help you with employment, but it
1:18:51 could also just be a fun, entertaining
1:18:52 thing that has its own revenue stream at
1:18:54 some point. So
1:18:55 >> yeah, totally. I mean I there's lots of
1:18:57 things like he just I mean I I mean I
1:19:01 was talking to um Beth from the Daily
1:19:05 Eye Show and talking about like having a
1:19:07 live a live host in the Daily Show and
1:19:12 um she's like they would have to know
1:19:13 how to do that. And then I've added a
1:19:16 future a feature um a call assistant um
1:19:20 that actually transcribes and suggests
1:19:22 words or things to say uh during during
1:19:26 the call.
1:19:27 >> Yeah.
1:19:27 >> And based on what's going on and you
1:19:30 could instantly just transcribe that and
1:19:32 then put a voice to it and and have it
1:19:35 reply during that.
1:19:36 >> Um you can just push a button and by
1:19:38 pushing suggest and it and have it reply
1:19:42 to whatever's going on the conversation.
1:19:43 So it wouldn't be difficult at all. Um
1:19:45 but that's yeah that's maybe I'll start
1:19:47 maybe I'll start doing that. That's a
1:19:49 great idea.
1:19:49 >> Yeah. Um Vicki Vicky of course being be
1:19:53 my second brain. Thank you Vicki. Um Pat
1:19:55 Flynn is the name of the guy. Um so look
1:19:57 him up and then I'll see I'll see if I
1:19:59 can get a recording of his session. His
1:20:01 session was really good. It was the it
1:20:03 was the closing keynote of the whole
1:20:05 social media um thing and it it was it
1:20:08 was really really good.
1:20:10 >> That's cool. Yeah. I would love to see
1:20:11 it.
1:20:12 >> Yeah. Yeah. It was it was really
1:20:13 powerful cuz he talked about that one.
1:20:15 There was another dude that like kicked
1:20:16 a rock until it became perfectly round
1:20:19 and it was like 60 seconds of a guy
1:20:21 kicking a rock. It was just his feet in
1:20:23 this rock. Um and like that one blew up
1:20:25 and then the dude with the sailboat and
1:20:28 and and he was just talking about this
1:20:30 idea of of short form episodic stuff
1:20:33 like you know doing this compelling
1:20:35 thing. I just feel like there's
1:20:36 something with Jasper there where it's
1:20:38 like you've built him and part of it
1:20:40 could be
1:20:42 just I I don't know just every day you
1:20:44 come up with some crazy idea of
1:20:45 something you're going to try to get him
1:20:46 to do and then it either fails or it
1:20:48 doesn't. You could have a thing like
1:20:51 failed, you know, or like
1:20:55 >> Yeah, that's good. That's really good.
1:20:58 >> I mean, that's so good. Um, it's just so
1:21:01 crazy what what works on social media
1:21:03 and
1:21:04 >> I know
1:21:05 >> and some things work and some things and
1:21:07 it's just especially right now with
1:21:09 everything that's going on like social
1:21:10 media is a weird weird world just
1:21:13 because you don't people who have been
1:21:16 streaming for years. They've done so
1:21:18 well and then all of a sudden something
1:21:21 changes and it's like where'd all my
1:21:23 audience go?
1:21:24 >> Yeah.
1:21:24 >> And I've got friends like that and so I
1:21:26 was just like
1:21:27 >> tell me about it that that happened to
1:21:28 this channel. Yeah,
1:21:30 >> it's just interesting. But I mean they
1:21:33 they they said it's just es and flows
1:21:36 like
1:21:36 >> Yep.
1:21:37 >> Go up and down.
1:21:38 >> Yep. Exactly. Anyway. All right. Well,
1:21:40 thanks, dude. Thanks for thanks for
1:21:42 showing us. That's really cool.
1:21:43 >> Yeah. Thanks for having me.
1:21:44 >> All right. Cool.
1:21:45 >> All righty, sir. Later, Gator.
1:21:49 All right. Cool. We'll do we'll do one
1:21:51 more thing. We've we've got we've been
1:21:52 talking about a lot of stuff in here. Um
1:21:55 Yeah. Thank you for the Thank you for
1:21:57 the demo. Um, okay. So, I'm gonna show
1:22:00 you something here. Let me share my
1:22:02 screen again.
1:22:04 Um, there's a fun.
1:22:15 Okay.
1:22:17 Can you see when I do this? Yes. Okay.
1:22:22 Hide that. So, I'm going to take this
1:22:24 image. So, it's the new cycle of AI
1:22:26 writing this. Actually, I have a newer
1:22:27 one than that, which I'm excited about.
1:22:29 Um, and I'm going to open this up into
1:22:32 chat GPT. And then I'm going to paste in
1:22:35 this prompt. So, this prompt came from
1:22:40 um Choi
1:22:43 Arachus AI.
1:22:46 Um, and it is Oh, maybe it didn't come
1:22:50 from him, but anyway, that's the that's
1:22:52 the prompt I had. Did someone else do
1:22:54 this?
1:22:56 I don't know.
1:22:59 This This is a prompt. Redraw the
1:23:01 attached image in the most clumsy,
1:23:04 scribbly, utterly pathetic way possible.
1:23:07 Use a white background and make it look
1:23:10 like it was drawn in Microsoft Paint
1:23:12 with a mouse. It should be vaguely
1:23:14 similar but also not really kind of
1:23:17 matching but also off in a confusing
1:23:20 awkward way with lowquality pixel by
1:23:22 pixel feel that really emphasizes how
1:23:25 ridiculously bad it is. Actually, you
1:23:28 know what? Whatever. Whatever. Just draw
1:23:30 it however you want. Um,
1:23:36 it helps when you call it the right
1:23:37 name. I love that he was he was calling
1:23:40 his assistant the wrong name and it
1:23:41 wasn't working. All right. So, let's
1:23:43 let's see. This is now the new image gen
1:23:46 tool within chat GPT. And then I think
1:23:48 we'll do some experiments in here. Like
1:23:51 one of the things I want to like I
1:23:53 haven't played a ton with this. It's
1:23:54 really good at infographics, but I don't
1:23:57 know if it's good at things like um
1:23:59 taking a picture of you and and making
1:24:01 versions of you that actually look like
1:24:03 you. So, we can play with that. We can
1:24:05 do photorealistic stuff. We can do a
1:24:07 bunch of different
1:24:09 uh kinds. Let's see.
1:24:25 Danielle, I made a self-portrait with
1:24:27 their prompt. It's so awfully good, good
1:24:29 and bad. Oh, look at that.
1:24:38 That's flipping awesome.
1:24:46 Oh, actually, you know what this is,
1:24:48 guys? You know what this is? So, so you
1:24:52 take a you take an image that you create
1:24:54 with with an image gen model and then
1:24:57 you use this prompt to to do this and
1:24:59 then you can send it to the copyright
1:25:01 office and say, "See, I handdrew the
1:25:04 picture and then that turned into that
1:25:07 other thing.
1:25:10 That's pretty awesome. Um, okay. Let's
1:25:13 do a new chat. Let me go find a picture
1:25:14 of me. We'll do one of these. Turn me
1:25:17 into a something. Let's go, Kyle.
1:25:27 Why is it not finding anything?
1:25:30 I know a picture of me with my name on
1:25:32 it. Yeah.
1:25:35 Huh.
1:25:40 The new Mac operating system, the
1:25:42 indexing is all [ __ ] up.
1:25:46 What is up?
1:26:14 Yeah, now my indexing is screwed up on
1:26:16 this machine. God damn it. Um,
1:26:20 I wonder if I can force quit the finder
1:26:22 without blowing everything up. Let's try
1:26:24 it. If I go away right now, I'm force
1:26:28 force quitting the finder. So, uh, I
1:26:31 might go away.
1:26:47 What's really funny is the anti don't
1:26:49 realize that AI can just as easily do
1:26:51 really bad art. That's pretty funny. Um,
1:26:54 okay. Let's go. Kyle
1:26:58 name contains Kyle. No, I'm going to
1:27:00 have to restart my machine to find that
1:27:02 [ __ ] Okay, let's see.
1:27:30 All right. Actually, this is a good one.
1:27:31 We're going to take this picture.
1:27:36 I know this is special edition live
1:27:38 going forward. Is this the format the
1:27:40 new live should take? Um, starting
1:27:45 Monday, it's what we're going to do is
1:27:49 we're going to do an earlier time and
1:27:51 I'm going to start
1:27:54 I'm going to do a 15minute pre-show on
1:27:57 all the external channels counting down
1:27:59 to the live show which will just be
1:28:01 happening inside Mighty Networks. And
1:28:04 we're going to try that for a couple of
1:28:05 weeks and see how it goes.
1:28:07 Um,
1:28:09 all right. So, let me open this up.
1:28:16 And then I'm going to say, um, I want
1:28:19 this guy
1:28:21 in a
1:28:25 tailored suit
1:28:28 walking into a modern
1:28:32 glass office building.
1:28:36 The photo
1:28:39 should look like a magazine photographer
1:28:46 took it but still be
1:28:50 authentic.
1:28:52 We'll just try that.
1:28:54 Okay.
1:29:14 By the way, for all of you for all of
1:29:16 you irregulars, just if you weren't here
1:29:18 earlier,
1:29:20 the what we're looking to do, so
1:29:22 starting Monday, we're trying to get
1:29:24 more activity back into the salon,
1:29:26 right? We've had all this we've had all
1:29:28 this sort of separation of of
1:29:30 communities because I've been streaming
1:29:32 on a bunch of different channels. So,
1:29:34 we're trying to get it back in there.
1:29:36 Um, we're going to take the month of May
1:29:38 and experiment. We've got a good idea of
1:29:41 what we want to do. We want to see if it
1:29:43 works. We we there's no way to know if
1:29:45 it's going to work until we actually do
1:29:47 it because everything in theory is a big
1:29:49 giant pain in the ass and change sucks.
1:29:53 Um, does that look like that guy? I
1:29:56 think it looks really like that guy. It
1:29:59 really does.
1:30:10 Yeah, it looks really like him. Okay,
1:30:12 that's cool. So, let's see. Um, now have
1:30:16 him in a
1:30:19 yellow slicker
1:30:24 on a fishing boat
1:30:30 during heavy weather.
1:30:34 His face should be strained
1:30:39 and his beard
1:30:44 a little too long.
1:30:53 What's weather? That's That's me
1:30:56 spelling. This is Sparta.
1:31:06 Steo congratulating Gareth. That's
1:31:09 pretty cool.
1:31:14 Gorton's fisherman. Yeah, exactly. Um,
1:31:18 yeah, Gareth, that was really
1:31:19 impressive. That's super cool. And it's
1:31:21 actually I mean
1:31:23 that thing I did today the the the thing
1:31:26 I did with the the data that LDG
1:31:28 provided to me
1:31:32 like that was pretty impressive that it
1:31:34 it got it as right as it did as quickly
1:31:36 as it did. And then I also asked it I
1:31:40 wanted to do semantic searching which
1:31:42 means I needed to take LDG's document
1:31:45 and embed it. And so we did that.
1:31:49 Oh, that's pretty cool. Did it Did it
1:31:51 make his beard a little longer?
1:31:54 No, the beard looks the same. The
1:31:58 expression,
1:32:00 let's say, um, make his face really
1:32:05 straining
1:32:07 and the weather
1:32:10 much worse.
1:32:13 He should have a long white beard
1:32:18 like he's been at sea too long.
1:32:30 Kyle, a regulars image tab in Mighty
1:32:32 Networks Live one. I just did cute
1:32:34 family pick. All right, cool. So, while
1:32:36 that's doing that, let's go look to
1:32:38 irregulars.
1:32:41 share irregular things.
1:32:45 This is a Jasper self-portrait. That's
1:32:47 pretty cool. Ah, look at Danielle.
1:32:55 It's awesome. Danielle, did you do that
1:32:58 yourself? That is a total Microsoft
1:33:01 Paint.
1:33:03 It's so good. It's so good. All right,
1:33:07 let me close this. Um, share irregular
1:33:10 things.
1:33:12 There's Steo. Oh, cute.
1:33:15 Community Darec Care Center.
1:33:23 That's good. That's really good.
1:33:27 Lori Blair, when I was little with my s
1:33:29 siblings with prompt Kyle's using now
1:33:32 about the messy drawing.
1:33:34 That's so amazing.
1:33:37 It really, you know, like
1:33:43 It's kind of nuts that we live in this
1:33:46 time
1:33:48 where you can just say [ __ ] and it
1:33:50 happens.
1:33:52 It's bonkers.
1:33:54 All right. So, that's work. That worked.
1:34:03 Yeah.
1:34:08 Well, let's try to get him in a in a
1:34:10 different All of these are they they
1:34:13 look almost like photoshopped layers on
1:34:15 top of So, let's say um turn
1:34:21 turn his head
1:34:24 to 45 degrees
1:34:27 from the camera,
1:34:30 looking out at C,
1:34:35 terrified
1:34:37 at what he
1:34:40 sees, but we can't.
1:34:44 Let's see if it'll do this.
1:34:55 Oh, that's good.
1:35:13 While it's doing that, um, next Tuesday
1:35:15 is AI Salon Presents. So, Ann Murphy and
1:35:18 I wrapped up season two of the AI
1:35:21 readiness project podcast. She's going
1:35:23 to go do some another project right now.
1:35:25 So, we're going to take a break from
1:35:26 that. Um, oh, that actually worked
1:35:28 pretty good.
1:35:30 Um, this is looking quite good.
1:35:34 Although, it does look like it added a
1:35:36 pattern.
1:35:39 Yeah. Uh, did it? I don't know.
1:35:43 Looks like it's adding a little bit of a
1:35:45 a geometric pattern to stuff. Um,
1:35:48 anyway, um, next Tuesday at AI Salon
1:35:51 Presents, we're going to have a bunch of
1:35:52 speakers that have spoken at the at the
1:35:55 AI at AI readiness project podcast. come
1:35:59 on and talk about what they've been up
1:36:01 to and what they're up to now and what
1:36:03 they should be thinking about. What
1:36:04 prompt are we using? So, what I did,
1:36:07 Steo, was I just I wanted to see how
1:36:09 good it was at taking an an existing
1:36:11 image and making variations of it and
1:36:14 keeping it look like the same character
1:36:16 character consistency. So, that's one of
1:36:18 the things I'm trying now. So, let me
1:36:20 say um turn this into a Renaissance
1:36:26 um painting
1:36:29 with a gold
1:36:32 gilded frame
1:36:36 in a museum.
1:36:39 Um,
1:36:41 with sea serpents
1:36:45 in the background
1:36:48 and a
1:36:50 tentacle
1:36:56 coming for his face.
1:37:13 Brandon's car buying journey. That's
1:37:15 gonna have to be Monday. But I like
1:37:18 that. I I like the sounds of that.
1:37:31 Uh uh.
1:37:42 By the way, Grock now has an agent.
1:37:46 Grock. Imagine.
1:37:48 I asked it to do a six
1:37:52 panel story board of 70s hippies and a
1:37:56 and a 70s Challenger. So, their hippie
1:37:59 bus broke down. Purple Challenger pulls
1:38:02 up. They love it.
1:38:05 They climb into it. They hit the road
1:38:09 and they end up
1:38:12 singing away.
1:38:17 Pretty cool.
1:38:19 Um, and I think I can turn this into I'm
1:38:21 going to say turn this into
1:38:25 video clips.
1:38:29 and then stitch
1:38:33 into a commercial.
1:38:38 All right, so that's off doing that.
1:38:40 Let's see. Ah, here we go.
1:38:43 Yeah, this is definitely adding.
1:38:49 Why is that not working?
1:39:05 It's adding weird geometric shapes to
1:39:08 the image.
1:39:11 Still pretty cool.
1:39:22 Okay, couple of quick things.
1:39:26 If you're playing with Lovable, Lovable
1:39:28 has a new app, which is apparently it's
1:39:31 fine. It's just basically like Lovable,
1:39:33 but it's an app. But you can also now do
1:39:35 Lovable via Telegram. So, just like you
1:39:37 would do with OpenClaw. And then Brandon
1:39:40 is doing a four-week sprint, a
1:39:44 mastermind on on lovable, right,
1:39:47 Brandon?
1:39:51 Hold. Yes. Okay. So, he's doing a
1:39:54 four-week sprint on lovable. So, if you
1:39:56 have been thinking about vibe coding and
1:39:59 you haven't done it yet, join the
1:40:01 mastermind if you're not a part of it.
1:40:03 If you are a part of it, sign up for
1:40:05 Brandon's four-week sprint. So, four
1:40:08 weeks on how to use um how to use
1:40:11 Lovable to do um cool and powerful
1:40:14 application development. I still can't
1:40:17 rep wrap my head around all that it can
1:40:19 do. All that all all that AI can do. I
1:40:22 know. I know. I'm the I'm the same way.
1:40:25 And it's like we're we're barely even
1:40:29 the tools are still not great. I just
1:40:32 when these things get really good, this
1:40:34 is going to be crazy.
1:40:41 I wonder what happens if Elon wins his
1:40:43 lawsuit. Yeah, that's an interesting
1:40:45 one.
1:40:51 All right, kids.
1:40:54 Let's see. Let's turn him into Let's
1:40:58 Let's go.
1:41:00 Um, we'll do one more.
1:41:03 I'm going to close that. I'm going to
1:41:05 say
1:41:07 Now turn this into a
1:41:14 pointalist
1:41:18 painting
1:41:22 where the guy
1:41:25 is in a red
1:41:29 suit
1:41:31 with A
1:41:35 light blue shirt
1:41:41 walking through
1:41:44 the park.
1:42:08 Someone outside is very angry.
1:42:21 I decided on a 4.5 portrait aspect
1:42:24 ratio. Cool.
1:42:43 I wonder if this will make images faster
1:42:45 if I if I turn off thinking. I bet I bet
1:42:49 it will. Oh, that's interesting. That's
1:42:51 actually really good.
1:43:04 Huh. All right, that's good. So, let me
1:43:06 turn off thinking. We'll do instant.
1:43:10 We'll do one more. We'll do um
1:43:14 his beard. Let's see. Let's make him
1:43:18 Let's make him
1:43:21 clean shaven
1:43:23 with
1:43:28 a large.
1:43:35 Let's see.
1:43:37 We'll do one more. Let's Let's make him
1:43:39 clean shaven with
1:43:42 um
1:43:45 black
1:43:46 cowboy boots
1:43:52 and a formal
1:43:55 western
1:43:57 hat.
1:43:59 Let's see
1:44:02 see if this goes quicker
1:44:05 because it's not thinking.
1:44:29 setting the scene. It's a little bit
1:44:31 faster.
1:45:03 Okay. It pretty much looks like the same
1:45:04 character,
1:45:07 huh? All right, there you go. There you
1:45:12 go, good people.
1:45:16 It works.
1:45:20 Image Gen 2.0 is pretty good. So,
1:45:21 mathematical. Kyle, you need to see Drs.
1:45:25 Jay's self-portrait in the chat. Okay,
1:45:27 cool. Let me go look.
1:45:41 There's digital gods.
1:45:49 Where's Dr. Jay's portrait?
1:45:54 Oh,
1:45:57 no.
1:45:59 Where is it?
1:46:01 Is it in share regular things?
1:46:03 >> No, it's in chat. You might have to
1:46:04 refresh. Uh
1:46:06 >> oh. Refresh chat.
1:46:12 >> No,
1:46:25 >> that's so good. That's really good.
1:46:29 Oh my god. All right, cool. Um, let me
1:46:35 stop sharing here.
1:46:40 All right, kids. Well, happy 14th
1:46:43 anniversary to Story Vine. Uh, my
1:46:46 co-founder Mon'nique.
1:46:48 She and I are still at it. Still banging
1:46:51 away at that one. And uh happy third
1:46:54 anniversary to this crew, to the
1:46:57 irregulars. And starting Monday, we
1:46:59 start the grand experiment
1:47:02 to see if we can concentrate some of the
1:47:04 energy here and get some some new voices
1:47:07 in and some new activity.
1:47:10 And uh we'll see how it goes. If it
1:47:13 doesn't go good, then we will do other
1:47:16 [ __ ] because we're adaptable. We're
1:47:19 curious. All right. Beautiful. Thank you
1:47:23 all as always. Thank you mods. Thank you
1:47:25 for your patience. I know it's janky and
1:47:28 weird and nothing feels right, but
1:47:30 that's that's how this goes. Um, have a
1:47:34 great weekend and then don't forget next
1:47:36 Tuesday is the presents meeting. So come
1:47:40 to that and then Monday we will we will
1:47:43 be trying the new stuff. on Monday, the
1:47:46 AI learning lab, we're going to start at
1:47:49 30 minutes earlier and then I'm going to
1:47:52 go live
1:47:53 at 7:15 Mountain time. So instead of
1:47:56 going live at 8, we're going live at
1:47:58 7:30 and 15 minutes before that I'm
1:48:01 going to be doing basically a countdown
1:48:03 to the live. So I'll be out on
1:48:08 LinkedIn, Twitter, and YouTube or Tik
1:48:12 Tok. Probably all of them, I don't know.
1:48:14 I think all of them telling people to
1:48:17 come into the salon. Um, and then at
1:48:20 7:30 the show will start. All right. So,
1:48:23 we're trying to make it a little more
1:48:25 convenient for the people on the East
1:48:26 Coast. Uh, but still have it be
1:48:28 realistic for me to do it with my life.
1:48:31 All right. So, beautiful. Thank you very
1:48:34 much. Happy Friday. Whoever sent that, I
1:48:36 didn't see that. Who was that was Lord
1:48:38 Digital Gods? Thank you very much, sir.
1:48:40 No longer on Tik Tok, only on mighty
1:48:43 networks. We haven't decided that yet.
1:48:44 We might we might keep Tik Tok. I'm not
1:48:46 sure. Source camp. Again, the whole
1:48:49 point is we've like all of the energy
1:48:52 has been outside of the outside of the
1:48:55 community and I want to I want to get
1:48:58 things buzzing again. Um and so the only
1:49:02 way I know to do that is to kind of
1:49:03 concentrate it. And so let's let's
1:49:06 experiment. So we'll see how it goes. If
1:49:08 it's just unworkable, then we'll we'll
1:49:10 modify it. But I want to at least give
1:49:11 it a shot.
1:49:13 All right. So, with that, I'm out of
1:49:15 here. Have a fantastic weekend,
1:49:17 everybody.