Ep. 76: How to Build AI Workflows That Never Start From Zero

0:00:00What is next for Fast Hours?

Play from 0:00:01 8:88:88 0:00:01 What's up everybody? Fast Hours Podcast, episode seventy-six. Rory Flynn, Drew Brucker. What's going on? We're back. Back. We've been consistent, guys. This is this is all very important to us. But we've had I will say I've been after listening to to Finn's podcast last week, our guest guest spot. If you weren't here, I would definitely recommend go checking that one out for anyone who wants some dirty in the weeds, nitty gritty graphic design and just like straight up AI nerd out session.

Play from 0:00:34 8:88:88 0:00:34 That was a good one. So you know, we've been we've been pumping, it's been a it's been a strategy. on our part to get more guests on to like give people different perspectives other than ours. But I think that one was one of, you know, hit between him, Salma, uh Tyler and J-Bugs. Like good run. Good run of guests. Good run. Good run. We're not done. Yeah. the next two weeks after today, we're scheduled to have guests on.

Play from 0:01:00 8:88:88 0:01:00 So just to guys give you a little bit of preview and kind of like prepare for that. Those conversations should be awesome. we've got two really great people that are in this space that I think are gonna provide totally different perspectives, but equally valuable stuff. First being Blacklight. We've Rory especially has mentioned Blacklight, you know, many times in this podcast. Big presence on X. Um, also, I think similar to Finn, he's got very strong perspectives, built in experience.

Play from 0:01:32 8:88:88 0:01:32 And that that extends beyond just the creative portion of AI, which I appreciate, right? Because like even Finn last, we were talking, you know, he's talking about the out of distribution. It's like, I think we've we've been thinking about that internally, but we've never really like talked about that as a group and named it, you know? And so like it was good to like get into that. And I I thought that was like super dope and a and a great part of that podcast. But I think Blacklight's gonna be able to share um that experience as well.

Play from 0:01:59 8:88:88 0:01:59 But he he's got some really great perspectives, big presence on X. Giving him a follow ahead of next week's episode. Also, we've got Rourke Heath on the show. Um we're we're planning to have him the week after Blacklight comes on. So I think this will be mid-August that we're aiming for here. Um if you're not familiar with him, I think. Instagram is his his main channel, right, Rory? Yeah. got one of the biggest AI communities on the planet, if not the biggest one.

Play from 0:02:27 8:88:88 0:02:27 So Rourke's awesome, great dude, really strong mission, but just like super technical, super knowledgeable, not just like if you've if you've seen any influencer peddling, you know, AI stuff on Instagram, they've probably copied his video. Let's put it that way. He's normally the first one out, and then everyone just models their videos after his. So Yeah. you there there is a lot of value in being first or very, very early to report these things.

Play from 0:02:55 8:88:88 0:02:55 And then uh like this reminds me of somebody else I know that does this or or was doing this very heavily. He just stressed he like he he told me how stressful it was to be in that sort of group because you're trying to be very quick and then couple that right with these high level visuals. that are going into this. I mean, you can just see that these are well done, but there's also a lot that goes into these edits.

Play from 0:03:22 8:88:88 0:03:22 Like these are not crappy edits at all, right? Like these are these are things that, you know, take planning, uh, take time to edit and tweak. Um, they're layered, right? So they're, you know, like going into Premiere Pro and probably doing a lot of stuff in there. Um, so like I think that's a heavy game. But Rourke's been around, yeah, he's been around for years now. I mean, he was he was definitely one of I don't know if one of the first, but like felt to me like he was one of the first.

Play from 0:03:50 8:88:88 0:03:50 he's been doing this for a long time. So I think he's gonna have some really good observations both from the creator side, the community that he's leading, um, things that he's seeing in the space. He also has an interesting interesting content creation background. And it's I'll I'll leave the story for when he comes on. I don't know if he knows I'm gonna ask him. During COVID and like what got him on like the content creator path. And it's awesome.

Play from 0:04:16 8:88:88 0:04:16 Like it's hilarious. It's unique. It's very original. And I was like, dude, this is no wonder you're here. Like he's it it's it's a good story. Vrock's a good dude. Um, you know, hopefully we get to we get to get in depth in a little stuff 'cause he's doing so much. So yeah, it's been like I said, we've been we've been pumping on the guests. We're trying, guys. We're trying to give you more than just Drew and I divide.

Play from 0:04:42 8:88:88 0:04:42 I think like and we and we appreciate like all the like I think a couple of weeks ago we were asking for, hey, what what now? What kind of things would be good to get into? What would be, you know, who would be some guests to get, you know, get on the show. So we've got some additional backlogging and planning we're doing for guests there, but also some topics. I think what we're gonna do too is I know a couple of people have been asking for sort of like deep dives into the node-based workflows, right? So I think we're we're looking at a a deep dive episode on that.

Play from 0:05:09 8:88:88 0:05:09 As well coming up in the future. So I'm I'm excited, man. I mean, this this might be the most planned and organized we've been since we started this thing. Actually, I I'm not even gonna say probably. And I still think we're not really. Yeah, there's so much opportunity to like get even more um organized and polished with it. But like we're having a blast, man. I think like this is probably a period of time where me personally, like I'm I'm probably having just as much fun as I've ever had been doing this.

Play from 0:05:36 8:88:88 0:05:36 And we're seventy-six episodes in. So let's roll. this made it this made AI feel not as laborus as it has been. I know we said that last week and there was a con there was a comment on here that was like, Why are you using AI if if you know, you're using AI and you're complaining about AI? It's like, Well, yeah, dude, anything's what what do you not complain about in your job? Do you love every aspect of your job? You know what I mean? Like, give me a break. You know? Oh y you know what I mean?

0:05:39Why criticize AI while using it?

Play from 0:06:03 8:88:88 0:06:03 shilling AI. I mean, I mean, like if how how irresponsible and echo chambery would it be if we're just pro AI on everything? I mean, we like we've been talking about, there is nothing that is absolutely truthful on either side of the equation. We are not fully AI. Do we use AI? Does AI pay the bills for us? Yes. Are there downsized AI? Yes. So we're gonna talk about them. We're gonna be truthful. I mean, of course.

Play from 0:06:30 8:88:88 0:06:30 And those you know, that that comment specifically, I'm like, yeah, like let's just let's just all look at our current jobs and tell me you have not complained about one thing that you use in that job. Whether that be graphic design, if you're graphic design and you haven't complained about Photoshop, you are a liar. Like you are what a hundred percent a liar. You've complained about it. You've complained about every Adobe product, you've complained about Gmail, you've complained about using Outlook, you've complained about whatever.

Play from 0:06:55 8:88:88 0:06:55 Like we can complain about it because we all nothing is perfect. So like That's one and then I saw someone else comment. Like some of these comments, it's like obviously it's just, you know, they saw Finn on the podcast and was like, uh, you know, can't believe Finn's chilling for AI. I'm like, Did you see any of the podcasts where he was talking about the stuff he doesn't do with it? Like the seventy five percent of the episode? That's what I mean, right? Yeah.

Play from 0:07:21 8:88:88 0:07:21 and they don't even watch, like they're not the regular users, because you can tell who the regular users are. And I I also love that um keep the comments coming, guys. Like uh give us your opinions, give us, you know, ideas, right? Like this is a open community. We do this for you guys just as much as we do this cathartically for ourselves. And this is all what it's about.

Play from 0:07:46 8:88:88 0:07:46 We're not we don't want to bring up things that we just think are cool. We're bringing up things that we hope you think are cool, right? And that we think should be talked about on a larger stage. So that's that's sort of the thought process there. So keep it coming. And you know, that's uh that's a great that's a great sort of reminder. Just gotta do one thing. I feel like everyone's been getting everyone's been getting it. But I mean like it's it's funny to what I like to see is that I never know if anyone's listening at the end and I see the comments, you know, when we're like you just gotta do one thing.

Play from 0:08:16 8:88:88 0:08:16 And we get people that are Yeah, tell your quantum physicist, right? Like that's a perfect. Like it makes me it makes us happy that people listen to the end. They hear our dumb stuff at the at the end of the podcast when we're brain's about to pour out of the ears and we've just really just uh dumped our consciousness into this podcast for an hour and a half. But it's that's that's awesome. So keep it coming. That was our housekeeping, you know, king of king of transitions here and king of, you know, s shifting the the segments.

Play from 0:08:43 8:88:88 0:08:43 I did want to sh you know considering that Finn was talking about t shirt design last week, right? I wanted to I wanted to show this on the podcast that I had. So this was This was if I can find it right now, wherever I put it. Here is we were talking about when I did t-shirt design for Shark Week, right? this was this was one of the t-shirts that sold millions of units.

0:08:56How can AI recover old creative work?

Play from 0:09:13 8:88:88 0:09:13 And I had this was a two-second design that I that I drew up in Illustrator. It was not, it was not hard, it was not crazy. Um, but I wanted to sort of like I had first off, I had Claude go search Wayback Machine to find it because I didn't have the file, right? So it was like I was like, go go back and find the, you know, from about like 2018, use Wayback Machine, you know, the Discovery Store, right? And then I had it isolate the design so I could like look at it again.

Play from 0:09:41 8:88:88 0:09:41 which I thought was kind of interesting, right? Just to be like to look like something that I had created with Illustrator back then, you know, hand drew some stuff. I used vectors for other stuff. But This was just fun like he he brought me back to that place where I wanted to go back. And I was like, yeah, I used to draw and then I'm like looking at the little details and I'm like, hmm, I remember like making that decision. I remember making that decision. And I was just playing with it, right? I'm like, what happens if I would you know, bring a little bit of a different style to it?

Play from 0:10:07 8:88:88 0:10:07 Now I'm playing with my designs from ten years ago, right? Where I'm like the And this this goes back to the episode prior to that, when we were talking about the the nineties stuff, right? And like how things feel a little bit more like you kinda want that weirdness, you want that like not that n anti polish, if that makes sense. And this was I think this was probably a high selling design back then because like the the nineties stuff wasn't necessarily back in style then like it is now.

Play from 0:10:38 8:88:88 0:10:38 You know what I mean? Like in two thousand eighteen, two thousand seventeen, you weren't seeing as much like vintage nineties sort of apparel. Like that look, that ugly look, you know, that we used to love. So yeah. like that's when you kind of started like seeing those starter jackets from the nineties when we were kids start to pop back up and then they're like three X in the the price on them. You know, it's like those were like $40 jackets and now you're charging like $150.

0:10:46Why did 90s sports design feel better?

Play from 0:11:07 8:88:88 0:11:07 You started to see the the oversized tees come back, the acid wash, the you know, the snapback hats. There was that period where it all kind of like came back around and it was like, we like this again. You know, a little bit more vivid neon colors, like that two thousands sort of like palette started to come back where it wasn't just like sharp, clean, it was like chaotic and a little bit more energy, right? You know, weird hairstyles, bucket hats, like all that stuff came back.

Play from 0:11:34 8:88:88 0:11:34 But that was I remember making that design and they were like Why'd you go with this choice? I'm like, I don't know. It reminds me of like a nineties sort of like surf t shirts that I used to have from, you know, Quiksilver or like Ocean Pacific, stuff like that. Rip Curl. you know, all those brands. So they were like the hot looked like street chalk design. Um, like neon colors. Yeah. And uh yeah, like that kind of stuff. I mean I love this jacket. I had that.

Play from 0:12:00 8:88:88 0:12:00 I had that. What happened to it? but see those colors too. Like those col like that color palette was, you know, that teal and purple, that was like a nineties color palette. And went away. Yeah. Those nineties those nineties NBA jerseys are like epic. You know what the the magic with the the black magic jerseys with this with the pinstripes. Those are those are epic. Those nineties Nuggets jerseys are great.

Play from 0:12:29 8:88:88 0:12:29 Yeah, like with Shaq in it. Yeah, those are just uh awesome. Yeah. think Penny and Shaq as soon as I see that. A hundred percent. Then then into Tracy McGrady. But like, you know, the them the Raptors from the nineties had the sick uniforms, uh, and the sick logos. Grizzlies also. Same thing. They had the the cool like Raptor with the sick font, like not like this digital everything is, you know, uniform.

Play from 0:12:59 8:88:88 0:12:59 Like look at that font. It's great. I mean, it was so much fun back then. Like a you look at all the stuff. The Sonic's jerseys were cool as hell. yeah, these grizzlies, right? Yeah. With the bear. Bryant Big Country Reeves. I think of who else? Bike Bibby. Mike No, no, that's right. Mike Bibby. Yep, for sure.

Play from 0:13:25 8:88:88 0:13:25 For sure. There it is, Reeves. US, if you want some really awesome inspiration, nineties sports jerseys from the NFL, the NBA, the uh NHL, they're all cool. They're cool. like two decades there. I thought like eighties and nineties were the best logos. The s the suns were also sick back then. The purple ones with the uh yeah.

Play from 0:13:51 8:88:88 0:13:51 Yeah, like those are it's so it's so cool, right? I mean like it's so different. You would never see that today. Never see that. There's all the acronyms we now. I mean, ugh. I mean it's just, you know, it could be yeah, right? Like it could be and then you see right there, right? Bottom right that orange that that's what they are now. It's just just flat just flat text, nothing fun, no everything is like so optimized for like it looks like web.

Play from 0:14:25 8:88:88 0:14:25 You know, like it that looks like it's optimized for your UX/UI spacing's perfect, like nothing is wrong. the I think the Hawks had a really good I can't remember if it was the nineties or the eighties though. Let me see. They did have some great jerseys back then. Oh, those are silly. I love the the hawks spread across. They're look look at those. I mean, they're awesome. And everything is thinking I think I'm thinking 80s with yeah like Dominique and Spud Webb.

Play from 0:14:52 8:88:88 0:14:52 Hmm. The I'm trying to think of who the other one was. I I always like the I always like the i even though they're kind of like boring, they're always like the pacers. I always like the the nineties pistons with the horse with the horse with the flames on the on the main. I like those. I always thought that one was their best one. Or let me, this is eighties. Hold on one second. Nineties.

Play from 0:15:17 8:88:88 0:15:17 Boom. Yeah. people don't you know, you don't like now. I like the white I always like the white ones. Those were my that was my jam. Yeah, Grand Hill. It's like we you got the pistons on the on the font blowing the flames, like it's just it's cool. I think even the heat. Remember just the ball going through the there was no net? well that one well that one was hard too.

Play from 0:15:43 8:88:88 0:15:43 The just the little flame on the T. Yeah. And then did you you can go through almost any you can go through almost any team of the nineties and find something that was cool. You know, the Hornets jerseys were o obviously awesome. The who else was the the you Bulls jer all the bull the Bulls jerseys were great. Naturally. We're for we're forgetting someone here that's gonna be like, You forgot about this p you know, okay, great, sorry.

Play from 0:16:12 8:88:88 0:16:12 Like they're all good. We just we just went through like sixty even the jazz. Like it was just like everything was so it was like it was like how how would you even describe it? You know, it was like um The nineties Trailblazers were also sick. The that I loved those, you know, like the the Clyde Drexler look back then. I I love these 'cause these were simple, but I just love the Yeah.

Play from 0:16:41 8:88:88 0:16:41 Like these were I just love that that color scheme and I just thought like the the wrap around it looked looked awesome. Yeah. It's it's just like it's simple but it's still got flavor. back too with like with that with the college, you know, basketball. Like like I think of like um Illinois, 'cause that's that's who I root for. like they started bringing back more of the script, right?

Play from 0:17:07 8:88:88 0:17:07 You know, this and this. And then um this was like kind of like a newer iteration on like an 80s uh thing that they had. You know, like the it's sort of like that. coming back versus like this, which was like one of my least favorite uniforms ever, you know? Um, it's like, what is that? You know, same with the, you know, even this one is just like, dude, no, no, no, no, no. You know, it's like this is just so sick. Like I I I give credit to a lot of the a lot of the older teams.

Play from 0:17:36 8:88:88 0:17:36 Like I think it I think the NHL is pretty good with keeping their logos, like the red wings. Like that doesn't seem like an optimized logo for right now, but like it's got it's got so much detail. you know, that like those kind of you know, like look at the with the the wheel with the wings, right? Like it's that's so much more extreme than any logo would be today. And I I love that. Blue's good. Blues key, you know, hockey's another great one from the nineties that has all the cool all the cool jerseys.

Play from 0:18:05 8:88:88 0:18:05 The sharks from the nineties, that's like the standard. Yeah. That one, you know. Mighty ducks. Yeah. I need to put But they weren't Winnipeg. Were they Winnipeg before too? I guess they were. Yeah, these ones. I always loved the um the Nordiques, the Montreal Nordiques. There dude there there's so many there's so many good ones.

Play from 0:18:32 8:88:88 0:18:32 That's a great one. It's got like a subtle like Inuit design there. end. Yeah. It's just like everything about it. It's it's cool as hell. I love you know if any of the if any of the sports leagues are listening. Yeah. Yeah. If any of the sports leagues are listening, go go back, be a little bit more creative. Right? Like just go back. Come on. Like like, you know, we had some we had some stuff that that stuck with us last week.

Play from 0:18:58 8:88:88 0:18:58 I think um what about what about this week, man? Where you where do you wanna start? I feel like there's a lot that we could touch on. I mean, um, you know, what's what's top of mind for you? We can start anywhere. have you been uh you know, struck with the LinkedIn slop button yet? Has anyone anyone reported you? baby, I'm sure. You know, like I I saw this pop up and I instantly was like, I I need to post about this because I have some thoughts.

0:19:07What is LinkedIn's AI slop button?

Play from 0:19:25 8:88:88 0:19:25 I I I have long been um a LinkedIn hater, but somebody that uses the platform. Um and i I think it's it's that way for a reason. I feel very strongly that LinkedIn does not Do a good job of the product. They don't make enough updates. When they do roll out updates, they just they're they fall flat. They're not thought out. So if you're not familiar, right, LinkedIn now, we I think there was a report, and I we might have brought this up on an ear episode two or two weeks ago, maybe.

Play from 0:19:57 8:88:88 0:19:57 It was like something like 40% of posts on LinkedIn now are considered AI slop. Now, what's the definition of AI slop? You know, is it just like you can tell that ChatGPT wrote it? Entirely, partly, right? Like the patterns, the the recognition patterns of it, whatever it is. But I just think like those detectors that detect that anyway, I mean, w when it's obvious, sure.

Play from 0:20:25 8:88:88 0:20:25 But those things aren't perfect by any means, anyway. And so what happened was LinkedIn rolled this out, where now on any given post, right, you can click this, and then uh assuming it's not your own post like it is mine. You're gonna get something like that shows up here where you can like click this and seems like AI slop. So I I made a nice little post about this. And I'll like here's my little diatribe, right? Um, they rolled this out.

Play from 0:20:50 8:88:88 0:20:50 I think the idea behind it is good because hey, we we do want to combat, you know, all this like poor quality or mediocre quality content, you know. And so, but the problem is there's no really foresight behind it, it just seems like they shipped it, and so. This is for posts, not comments. That was my number one issue because comments are probably worse than the posts. If you've posted on here, you know what I'm talking about.

Play from 0:21:16 8:88:88 0:21:16 I think there were even some on this post, ironically enough. So you've got that. Anyone, here's. perspective, Drew. How do you see this unfolding in the future? Here's the here's the here's the real, real big issue. Maybe the biggest of them all. Anybody can use this button. And how is this button used? You know, like internally for LinkedIn. Um, because I don't like somebody, could I just go through and hit AI slop?

Play from 0:21:44 8:88:88 0:21:44 If I'm a business and I have a competitor, could I just go through and tag these things? Like what are we doing? Like I feel like there's a weaponization that could come into this. Who's gonna review this? Not a human, so AI reviewing AI and deeming slop. How's that working? I based on other products you've put together, I'm guessing not well. And then no definition, I mentioned that earlier. And then, of course, the best part, which is just the irony of it all, LinkedIn, if you remember, was pushing so hard around these AI features.

Play from 0:22:17 8:88:88 0:22:17 You had to be a premium user, or remember when they had those like thoughts. leader things like badges you could get if you answered questions and people would just use AI for that. And now this is the exact thing that they're trying to stop. And I just I laughed my ass off and I just had to make a post about it. And uh yeah, I'll get off my soapbox and let you comment. I'll hang up and listen. this one. LinkedIn has been there's something wrong with LinkedIn.

0:22:37Why did LinkedIn reward lower effort?

Play from 0:22:44 8:88:88 0:22:44 And it's d getting worse. And not because it's getting worse, because the you know, just the platform's getting worse. It's like a strategy and direction. Like that's getting worse. I don't know who's running it, I don't know what they're doing because it's wrong. Right. Like the two years ago, it was the best professional place to to be and present your work because number one, it was verified accounts where everyone was a real person. And two, there was like you weren't gonna go and embarrass yourself because you're in front of the entire professional network there, right?

Play from 0:23:14 8:88:88 0:23:14 So it was like you brought higher quality content because also like you couldn't do the slop stuff because people just disregard you. But then they started rewarding that somehow. And I don't understand because a lot of the times, like I've said this before, where they were they went through like a whole period of all right, we're gonna de-incentive, you we're gonna stop giving you so much reach because we wanna focus on finding your core audience, right? And that has that has negative effects on two sides, right?

Play from 0:23:41 8:88:88 0:23:41 Negative effect number one, this is Rory's this is Rory's lecture here, right? Negative effect number one. People like myself, who have spent years growing a following, now understand that if I put two days into a sixty page PDF, which is what I used to do to get insane amount of impressions and business from, I'm not gonna do that anymore. So that quality of sixty page free PDF, use it however you'd like it, take it.

Play from 0:24:07 8:88:88 0:24:07 I will never do that again because I'm not going to get rewarded for it. So, like, okay, now I have to play this game of what does the algorithm want and what's going to take me the least amount of time to get there, right? Because that's sort of what they when you start to restrict impressions, like I can't even get 20,000 impressions now with 150,000 followers. It's like I can't even get like you know the any substantial and d don, no one feel bad for me.

Play from 0:24:33 8:88:88 0:24:33 I'm just saying like this is what probably real. Creators on here who've been putting out content for years are feeling, right? And then it's like, okay, fine. So I go to the lowest common denominator of what's the least amount of work I can do to get the highest return. Because why, why is it worth five days of effort to get nothing? Right. So there's that. The other side of it is the creators that are just starting, right? So the people that, you know, when we used to get incentivized when you started, right? Like it it got you built up on the platform.

Play from 0:24:59 8:88:88 0:24:59 The same thing that TikTok does. You start, gives you a little bit of a boost, right? And you you like, oh, like dopamine hit, I'll post more. Right? Like that's where it goes. So now when you don't get that either, it's like you have no one new creating good content, and you have the people that at the top that were, you know, like really putting in a ton of effort are dropping down. And then you also now have AI that writes all this stuff. We're we're right where we are. And that's they've created all this whole problem themselves and they do nothing to govern it.

Play from 0:25:29 8:88:88 0:25:29 And now this is just like a a stupid attempt. Because look, I wanna go on there and hit that button for a million things. Like I I'm feeling the urge to do it for a million posts, but I'm not gonna do it because I don't wanna take someone else down who's just trying to do what I'm trying to do, which is just like, you know, like Do we don't even know? Like we you know, are are you if you get enough of those, are you sh are do you get limited reach? Is there a s back end score that LinkedIn has that like you know you get negative net effects from?

Play from 0:25:59 8:88:88 0:25:59 Like what what is it? And I think that's part of the problem is there's like no clarity. There was no to my knowledge, no article put out on this. There was no product, you know, feature release thing that I saw. Maybe I'd maybe I'd need to look harder, but it's just like that should have accompanied this. This is can only hate on AI slop, right? You can't hate on just bad content. Which is to me is like, you know, there's there's other types of content that I should I don't want to report, but I want to be like, this is trash. Why is this on LinkedIn?

Play from 0:26:25 8:88:88 0:26:25 You know, like there's the who am I to be the governor of that? But I'm saying, like, you're only siphoning out AI stuff as slop and not like these stupid little Instagram videos that have made their way onto here where it's like, you know. Like the videos of the the guys in like the jungle building like the pools and stuff and like hyperlapse. It's like why is that on here with your message? Like I understand that's just like a clickbait tactic. Like I understand I'm gonna watch it because like that's the psychology of it, but like that doesn't that doesn't provide any value.

Play from 0:26:54 8:88:88 0:26:54 We're all just trying to game the system. So it to me it's just dumb. And it's just stupid. And there's nowhere to complain about it. And uh, you know, we're gonna complain about it here. So, you know, you guys could think we're 'cause 'Cause they're like elitist LinkedIn creators. It's like, no, we're just putting a lot of time and effort on this thing to watch it go to trash. Yeah. think that's the other thing, right? Is like you, you know, we we both put a lot of time into the platform.

Play from 0:27:22 8:88:88 0:27:22 And, you know, there should be rewards for that. Um, meaning like I I I genuinely think, right, like we've we it feels like we've moved away from social media in the organic sense. Now we're just we've we've completely overextended on algorithms and testing and experimenting with different things and putting a bunch of like putting putting you down like a path versus just like, can I get the original Facebook timeline where I'm just seeing people that I followed purposefully's content in a linear fashion, right?

Play from 0:28:04 8:88:88 0:28:04 And I'm not like seeing stuff from like three weeks ago and you know, like just real time stuff from the people that I want to see it from. And I have no can we have no control over that. And that shift away from that part of social media has been like the biggest turn off for me about it in general, is just that itself. And then I I think the other problem is because we've grown audiences too, you would want, you would expect the reach to reach more of your followers.

Play from 0:28:33 8:88:88 0:28:33 Yeah. is no, there is no benefit really now because somebody could start tomorrow and they're gonna get just about the same amount of reach as you are, if not more. You know, and it seems so in that sense, it's almost like, damn, we came at a time where we had to like build and build and build. And, you know, we did get the benefits for a little while. But now it's like if you're just starting, shit, now's a great time to start, I guess. You know, I mean, you're not getting this broad push from an impression standpoint, but there's nothing that's stopping you from getting.

Play from 0:29:03 8:88:88 0:29:03 the same, if not better, engagement than me or you, you know? And so I I don't know. It's it's it's really weird, but I saw I saw a unique growth hack strategy that I loved from Gary V the other day. Cause I was I don't follow Gary V, but like when his clips come up, I listen because the guy is normally typically ahead of everything. Yeah, and he's he's just he's always thinking about gaming the system and stuff. So I'm like, all right, you whatever it is.

0:29:10How are creators gaming account reach?

Play from 0:29:28 8:88:88 0:29:28 But he he was saying that he does this is an interesting one. Because because of that new sort of user benefit that they give on like TikTok and Instagram, basically what their f your first post is doing is like seeding it out to the algorithm to see if your account's gonna be interesting moving forward basically. So you get a lot more reach. So what he's doing is taking like previously viral videos that he's created. And now all he's doing is taking that and then mass creating new accounts, viral videos on the new accounts.

Play from 0:30:00 8:88:88 0:30:00 And whichever one just takes off and just keeps running that one. And then the other ones just scrap them and delete them. So it's just like we're He's like, I've I've gained four million followers on this one account to my fifty million on my regular account with one post. And I'm like Interesting, can't do that on LinkedIn, but like Yes. Yes. not that doesn't feel like w the way that things are supposed to be. I mean, in any way.

Play from 0:30:26 8:88:88 0:30:26 gonna play now, right? Because he's now saying that. And then you're gonna start to see this everywhere. Everyone's gonna have a hundred accounts because the bots can run it. And then it's just gonna be like the same sort of situation that we're in now. how many years down the line? Hopefully, we get back to a place where they just be like, we just like way overthought this and we're just gonna let you see everybody that you follow. Cool. We figured it out, you know?

Play from 0:30:52 8:88:88 0:30:52 It's like, dude, come on. did that though. X X did that, and I started seeing people in my feeds again. And it was like, this is cool. Like I'm still getting like I feel like that's been the best balance of anything so far. Instagram actually hasn't been that bad at it as of recent, but like seeing your friends, people that you follow, and then also like other stuff baked in. And I'm like, this is a nice balance. Like I don't hate this. like I was maybe like four months ago. I'm like, I haven't seen anyone's stuff that I follow.

Play from 0:31:18 8:88:88 0:31:18 Like I wouldn't even see your stuff. And I'm like, Guy you're like the only person I interact with like every day on X. So it's it's weird, right? But they need to do something about it. Even like the naming of that tool seems like AI slop. It's like either, you know, go all in or don't go all in, right? Like seems like it. It's like no, this is AI slop or this is not AI slop, right? You know, don't pussy foot around the conversation. I hate that.

Play from 0:31:44 8:88:88 0:31:44 and I don't give LinkedIn this much credit, would be if this were just there as a placebo effect to make you feel good that you are tagging things. And then they're not doing anything reactively with it. They're just ingesting and getting uh they're running a basically an experiment that doesn't punish anybody, but they're just seeing how people act.

Play from 0:32:11 8:88:88 0:32:11 That would be the the genius move. But I don't I don't I don't think I don't think they're that smart. I don't think they're that smart, but anyway. done anything. You know th all they've done is not update the platform in any way, shape, or form to make it better for content creation. Multimedia uploading. for like two decades now. Can we get a video and a picture in one post? Like is it that hard?

Play from 0:32:38 8:88:88 0:32:38 Every other platform on earth can do it. I can vibe code a platform right now that could do it. I don't understand. Like thing we were talking about before, like before we hopped on? It's just like the these bigger companies now, they're the maybe the biggest problem is just like you might have an idea and then to execute on that idea and do it. Like we were using just the example of like an MCP. It's like you probably have been thinking about doing this and planning to do it and putting that project together for months, and then you finally get to completion.

0:32:43Why do large platforms ship so slowly?

Play from 0:33:06 8:88:88 0:33:06 It's like you could literally hire somebody to do this the same week, right? Like maybe even the day of. These enterprises have to move faster. Like they have to. I just it's like anybody could like I could create a Chrome extension for any platform I wanted to right now in a day. And why can't that expedite a process? Sure, you could like still QA it and like go through stuff, but it's like from an MVP perspective, day of.

Play from 0:33:37 8:88:88 0:33:37 What are we doing? There's yeah, LinkedIn if you're listening, if you wanna you wanna hear some user feedback, got a lot for you. Tons. God. Okay. All right. Let's uh we're gonna we're gonna peel back uh yeah that a little bit. Um what else, Sararia? I mean Seedance 2.5, I think you were saying just dropped too.

0:33:48What changed in Seedance 2.5?

Play from 0:34:02 8:88:88 0:34:02 I I did see that. I wasn't sure is that it did that officially drop today? I was pulling something up on that. I s cause I cause somebody tweeted about it earlier this week, but maybe that was just like maybe there was like a Maybe a beta group or something. Let me let me let me pull this up here. Let's confirm. Okay, that's Dreamina. Let's go straight to the source. Can we find okay.

Play from 0:34:28 8:88:88 0:34:28 Does say it's live on Dreamina and BytePlus, so I'm guessing it's live for everyone. fourteen days of unlimited Seedance on Higgsfield, what a surprise. You don't say. Yeah, so I guess yeah, I guess official as of today slash yesterday. two point five. And I think what's what's what's the thing with that? I think it's the uh I think it's native thirty second gens.

Play from 0:34:53 8:88:88 0:34:53 Is that it what it's all about here? It can support up to fifty references, which is sick. That's crazy. Like, I I was like, how what do I even use for fifty references? You know, like I have to start thinking around like the and like Finn said, I want to go read the documentation and actually see what it can handle, what it can't handle, and then work we can push it. Cause it's if anyone hasn't used Seedance 2 in like a more meaningful way, it's very impressive when you start to get into it and what you could use as references and how you can use those references.

Play from 0:35:25 8:88:88 0:35:25 It's like To me, that's the main part of it. That's like the Nano Banana situation, you know, we had it was like, I can give it a couple different things, totally unconnected, and it can sort of tie it together without this crazy level of detail and instruction. But, you know, that being said, you're gonna get a lot of slop. we're gonna see we're gonna timeline's gonna be filled with slot. Yeah. we we've typically seen a jump, uh a leader change each time uh like one of these newer models comes out.

Play from 0:35:54 8:88:88 0:35:54 But Seedance was already at the top. And now you've got a uh built on on top of that. Um it doesn't sound like me and you have gone really deep in that yet, obviously, with this just being launched, but it'd be interesting to see like how bigger that delta got. And then I w I think about this from inside a company perspective. If you're gonna use this, like assuming you're like, I don't know, an enterprise, which I know this is, you know, video is still sort of like at the infancy stage of getting integrated and where and how and why we're gonna use it.

Play from 0:36:22 8:88:88 0:36:22 Um, but they're not gonna use the Chinese models. So they would be waiting for the O4. And I was just thinking about that. I'm like, I haven't heard nothing about that, but I would expect something here going back to our idea and strategy of like quarters, like thinking in quarters the way big enterprise companies think, that this would be sort of like a Q3 thing here, um, September, like August, September, would which would be what I would I would think.

Play from 0:36:54 8:88:88 0:36:54 And so I don't know. You know, I I'm curious to play around with this for sure. But um I think about this from like a an adoption standpoint. I mean, you're you're still gonna need somebody I on the American side, probably Google, I guess, at this point, really, to have a frontier tier model to get integrated with. Yeah. I'm I'm kinda curious as to what Kling's doing next, 'cause they've been quiet for a couple of months since three, so I think like they're probably coming up next and it'll probably be really good.

Play from 0:37:22 8:88:88 0:37:22 Just like, you know, my old reliable, like it can never sort of get away from Kling. I still love it. you know, even if Seedance can do a lot of great things, Kling's still just like simplicity and like do what you say and it and it listens. You know, Seedance is a little bit more complex. you spent more time equity there, you know, like so there's also to something to be said of like a place where you fe feel familiar. So it's like if two things were almost equally the same, you're gonna pick the one that you feel more comfortable and in depth with anyway, at the same time.

Play from 0:37:53 8:88:88 0:37:53 Sort of have a uh kinda have like a I don't want to say like an affinity for Kling. It's just been like they've they've also you know, Seedance has also been very nice. Kling Kling has treated me very well. Like with, you know, reach outs and they've sent me stuff and they're always communicative and they don't ever feel pushy, they don't ever feel like this is like some sort of one way agreement where it's like if I do this, then you know, they'll you know, I'm like on the hook to them.

Play from 0:38:20 8:88:88 0:38:20 Do you know what I mean? Like some of those companies Some of these companies we've partnered with, it it certainly feels like that. Like where it's like, Yeah. you scratch my back, I'll scratch your back. Yeah, like Weavy, not a transactional or weave, sorry, not a transactional relationship. Exceptional team. I feel like I'm part of the family over there. Right. Like they're they do it the right way. Like that's a it's a it's a relationship business, right? Like it's not just like I feel like Higgsfield's super transactional, you know, like where it's like again, them, there's a few others.

Play from 0:38:49 8:88:88 0:38:49 I feel like Luma's creative uh CPP got some some real heat for a while there. They were being crazy. Yeah. been super generous too, I should mention them as well. Like they're they're really great with their, you know, create a partner program and so is Flora. There's a few real good partner programs out there, that treat people like, you know, obviously we're doing a lot of marketing for them and they they act like why we're part of the team, which is great.

Play from 0:39:16 8:88:88 0:39:16 'Cause I mean, kind of we are. I mean, it's we're you gotta you gotta have good people in those seats internally to run those the right way. I mean, it's it can definitely feel transactional. You're right. There there are a lot more people or or entities or brands for that matter that feel more transactional than relationship, truly relationship. Yeah. So this one's interesting. I think we'll probably have more on this probably next week, right?

Play from 0:39:41 8:88:88 0:39:41 Or something like that when we get a little bit more into testing. I personally haven't been able to go like really far on it other than rerunning a couple generations, which haven't really noticed the difference to be honest, other than thirty seconds, but I haven't really pushed it to fifty references to see where that goes. But There was a there was a very quick maybe maybe we switched the Midjourney very, very quickly. There was uh they did have an office hours Wednesday, but it was like one of those real short 10, 15 minutes David hops on, hops off, like busy, heads down.

0:39:57What is new in Midjourney V8.2?

Play from 0:40:14 8:88:88 0:40:14 You know, they just did the acquisition last week, which we talked about with CoStar. That was their first acquisition. Um, I just I just basically broke this down to do a few notes that we just Should probably keep in mind. So V8.2, that's now live. That is not, you don't need the preview tag anymore. Um, you just V8.2. Um, I actually haven't run it a lot since that has not been not preview, but um I want to run more to really differentiate the difference in the final model that they released versus that preview one because I do uh from what I've read and heard, it's slightly different.

Play from 0:40:50 8:88:88 0:40:50 And then they just talked about V9 already in training. And they talked about edit models, you know, getting ready to be tested. So these are this is from the week before notes. one other reminder is they're talking about, you know, they're doing the UI stuff. Don't forget about alpha.Midjourney.com because that's where they use the testing ground for the things that they're going to push to the live site. So this may be a good reminder there to keep an eye out for when they do do the redesign, or maybe you're curious about that.

Play from 0:41:19 8:88:88 0:41:19 Um And then just mobile. Yeah. And then just mention that they're gonna do a you know, survey around the medical spa. That was it. That was all I all had to mention there. been still been jamming on it, pushing buttons. And it's it it feels good again. Um it feels really good again and I'm I'm liking what I'm getting. I'm having a lot of fun back on Midjourney and it feels it feels like I don't have to run like an agent system to just create.

Play from 0:41:48 8:88:88 0:41:48 You know, it's it's good. I don't know if you have you been have you been doing anything like outrageous Thought these were cool. I was running my I was running these these things again. the uh it's f it's like again, I forget that I used to do this and go all the way back. Um I I signed on to Minimax a few days ago because they released what's it called?

Play from 0:42:18 8:88:88 0:42:18 They're the Minimax three they released and the like H3. And Omni Reference again for anyone who's looking, I haven't again I've only tested a few things on that, but I went back and saw some of the stuff that I had done previously and then I was like, these are all like Midjourney V6 images. And I was like, I went back and found some of the the older gems that I used to do on Minimax with like, you Pikachu and Batman doing drugs and stuff like that.

Play from 0:42:45 8:88:88 0:42:45 And I was like, I the slop, the sl that was like pre slop. That was like the pre pre Sora. It was like fun, stupid and Minimax to do some great stuff. But you know, regardless, go back to your Midjourney V6, you know, images. There's a lot of good stuff in there and the prompts. But That looks inter looks like oil painting mixed with realism in some of those. sort of like this encaustic wax portraiture.

Play from 0:43:11 8:88:88 0:43:11 I was encaustic wax. Spit that one right out, didn't need a Yeah. Yeah. 'Cause I mean, I knew what you were talking about. I mean, yeah, it does it's got like this waxy bee wax sort of look on top of it. It's like a treatment. I was reading about it. That's just a it's like a pot like basically like a finish on on something. Huh. I like that one. it kinda it's it's kinda neat. It it like it it has sort of like these pops of highlights with because it's the wax and then it's also got like this this little texture that that shows up.

Play from 0:43:44 8:88:88 0:43:44 I think these are all those. Um And it's got the makes the lighting like reflect off it pretty cool. Like so this is again mentioned from Finn's podcast from the the show with Finn last week. I feel like texture is like the biggest differentiator like in prompting. When you go when you can type in encaustic wax, it obviously has a big influence on the image, right?

0:43:48How does texture improve AI prompting?

Play from 0:44:09 8:88:88 0:44:09 And that makes it, you know, just like he was talking about prompting with uh with the uh the fabric in the you know, interactions and things like that. think that's that's huge. And I think that's one of the tips that I I took back 'cause it's, you know, I'm a texture junkie. It's like you know what, I'm gonna research a little bit more of these that I can that I can sort of play with. Yeah. trying to like find I I think I found encaustic waxing because I was I was also trying to find styles that I have not seen or heard of before.

Play from 0:44:40 8:88:88 0:44:40 Like I was trying to like where where's sort of like the the white space or blue ocean of the styles that we just haven't seen yet? You know, and I just kind of started throwing some things around. There were a couple different ones I was doing. Uh this is a few weeks back now. I think this was another one. This curl Kirlian? Kirlian-style electrophotography. I thought was super cool. You know, like these are dope. Dude, make a mood board out of that.

Play from 0:45:06 8:88:88 0:45:06 Yeah. Yeah, these are fun. Um Or send me that prompt and I'll make a mood board out of it. yeah, there you go. All right, I'll send it to you. Thermal imaging, you know, it's cool. Like this one, like these feel like Roy Flynn colors right here. Um, let me see. What else? Do I have some other ones in here mixed? tell from our nineties basketball uniform discussion where a lot of my my color preferences come from, y you'll see it there.

Play from 0:45:35 8:88:88 0:45:35 Rewind back, you know, thirty minutes by Homestead, that's right. Don't use my personalization code. It's bad. This is so interesting. Like uh again I have any others here. Good textures. I see a lot of fabric play. I don't know if you're doing that on purpose, but these look like me too. Maximalism with Yeah. was exploded view object portraiture.

Play from 0:46:02 8:88:88 0:46:02 I like that. you know, so like shit, like I was like, I I don't even know what that is. Let's run it. You know, just again, like getting into that maybe latent space that you hadn't seen. what is this? Microfiche? Forensic evidence framing. Well, those look like more like stills, but that one's cool, cool. You can kind of see where that direction could go. This was cool. Like, I haven't seen a lot of these patent drawings, but I love those. Like these could be really good ah if you're wanting really good text, right?

Play from 0:46:28 8:88:88 0:46:28 And like I would love to try this in nano or GPT. The patent drawing stuff is always super cool to me because it's got like these really micro text pieces and it's also got like the handwritten fonts for like the figures and the call outs, you know? Like very and then it yeah, I it's dope. Um W what I've been liking a lot is this one token that works really well with eight point two. It's I've been using it for for layer depth on text.

Play from 0:46:55 8:88:88 0:46:55 So it doesn't look like it's been like graphic designed and like placed on a two D sort of flat surface where it's integrated into where it feels like it's in like the mid ground or it's slightly angled so it has like dimension and depth to it, like it's part of the scene, like floating in the scene. That one to me has been more like I've been I love that it can do that 'cause it's it's a hard effect to achieve. Um Yeah, otherwise it it's a lot of doctoring in the design, you know, in yeah.

Play from 0:47:24 8:88:88 0:47:24 So that's a that's a good one. I like the LiDAR. That's interesting. Yeah, that one could be another cool one to play with. I'm just seeing if there's if I have anything else in here. I ran a bunch of these. Newspaper halftone realism. I thought was kind of cool. Uh a couple of these turned out well. Uh what else? CRT phosphor burn. These are more textural, you know?

Play from 0:47:51 8:88:88 0:47:51 Photocopier, art. Um Do have anything else? What else you got? I've been finding that there's there's some really d like candidly the mood boards to me, I feel like they always take a little bit of time with these new models to sort of settle in for whatever reason. And I don't know if they do that on you know, if that that's the way it is.

Play from 0:48:16 8:88:88 0:48:16 I feel like some of them have a little bit more You ever notice that your mood boards have a little bit more power than other ones? Like some just like totally take over an image. Hmm. Do you like I memory? I mean, we went really deep in mood boards a while back. I they they've they've said they made changes, but I don't really have I I don't feel like I've noticed the changes so much. Um so like I think you're right.

Play from 0:48:44 8:88:88 0:48:44 Like there are some, and I'm wondering if that's just a matter of how many images you're putting in, right? Like. I I don't know. Remember when we had I mean, we did some really good ones with these when we did our mood board collection. Correct. That was so there's there's one or two that have like a really str like if I put them in any image it just totally dominates. Even if I have like eight mood boards attached.

Play from 0:49:09 8:88:88 0:49:09 Which is the only way I do things now. It's just could ma I could mash these more probably. I I really haven't played with those in a while, but yeah, I mean we'll s we'll see. I I I'm curious, you know, just wake me up on the edit models here. Um pretty much it. like it's it's having its swing back. I feel a lot more momentum back on the Midjourney side. I think people are like we've tell like we've said a few times, I think people are just kind of getting Yeah, just a little bit too comfortable with nano and ChatGPT, and it's like you're not it's not like stimulating enough.

0:49:25Why is Midjourney regaining momentum?

Play from 0:49:41 8:88:88 0:49:41 You know? It's like you know what it can do, you know what you have to use it for. It's not like throwing you a wild card. It's not it's not throwing you a wild card every once in a while, like the models used to, which I think is the again the biggest part. We've said this a hundred thousand times. But well. where these other models aren't even thinking about that. I mean, not yet. So You know what I have been seeing a lot of though is like I've been seeing we we probably mentioned I think we said this on the podcast with Salma.

0:50:04Why is AI entering its 3D moment?

Play from 0:50:12 8:88:88 0:50:12 like the the 3D moment for AI and it feels like it's it's really taking off now. feels like GPT-5.6 and then you know Fable sort of jump started that and I'm seeing a lot more of it. I'm seeing a lot of 3D integration, a lot of people either, you know, doing these uh Doing these depth maps for existing video and using them as references or building like, you know, CAD models and Three.js or doing all this other fun stuff with 3D I think we're gonna think we're probably gonna see a lot of cool stuff with that now.

Play from 0:50:48 8:88:88 0:50:48 Yeah. a really good call. Like that's that's another one of those like big waves that we haven't really that hasn't reached the shore yet. You know, like that's one of those monster ones that is yeah, I think you're spot on with that. I I wonder when that happens. I mean, it doesn't feel like we're far off. No. It's been it's been pretty you know, I think anyone who has any experience in 3D, if you're able to sort of direct and control what you can do with you know, in a 3D tool, you can easily tell a LLM to do it.

Play from 0:51:20 8:88:88 0:51:20 'Cause a lot of the a lot of them, you know, a lot of the 3D softwares they take they ingest code. And, you know, if Claude or GPT can write code, they can do stuff. It's just uh I think having again subject matter expertise there, knowing exactly how to describe what you want, what changes to make, it's not as easy. I mean, you know, having the Three.js sort of rendering within Claude and GPT and them just doing it like live in your p in your chat is insane.

Play from 0:51:46 8:88:88 0:51:46 But also the the brow the Blender MCP and yeah, I think I think Cinema 4D has one and a few other tools have one. But I I like, and I would say my experience with this, like on the photography side, is this has a really big short tail effect for those. Like if if anybody 3D is listening to this, the earlier you get on that wave, the more value you're gonna get out of that, especially short tail, because people aren't gonna know the things that you know yet.

Play from 0:52:17 8:88:88 0:52:17 They're not gonna know the best way to to to communicate, to prompt it. To engineer it, to think about it, right? The more examples and people that start to talk about it, eventually that that that moat sort of dries up a little bit. But there is still long tail too, right? Like so I'm the example I'm giving there is relative to photography in my sense of when I first started using Mid-Journey, I instantly was thinking about camera angles and the camera, maybe putting the camera in the prompt or the lens in the prompt or you know, uh the type of color that was, you know, either color grading or maybe it's got sort of an effect of film, you know, film grain or noise or shutter speed or direct, you know, like all of those things that were SME things that I experienced on the photography side.

Play from 0:53:06 8:88:88 0:53:06 Now a lot of that now is common knowledge, right? But there still are long tail benefits of having that experience. Right. Like you're still thinking about it from a photography standpoint. Like I instantly just go and I think about like even photographers now, like we've talked about this before, but it's that that net effect of like, what if I were to go out to a camera today and I was doing a shoot and let's just say it was a high stakes shooter, I was getting paid for it.

Play from 0:53:33 8:88:88 0:53:33 I've got real information capturing the camera. What if I didn't get The camera angle that I wanted, or what if something about the lighting is just a little bit off? I can now use AI to offset those things. This is now a part of the workflow. It doesn't have to be this, I'm just going to create AI images instead of photography. No, it's like how what are the pieces, the levers that I'm going to turn that use AI and infuse that into my process, integrate it in.

Play from 0:53:58 8:88:88 0:53:58 And so I think like if you're a 3D, you know, like we saw the photography, we saw the film stuff now, like to your point, like. 3D, my biggest advice if anybody's 3D is just like figure out what those short tail things are, do it early, then figure out the long tail things because there is a lot of value that you can provide as an SME. It's uh to me like it's the can it's the ultimate control that everyone's been looking for. Right? Like whenever anyone's like I want this like extreme detailed level of control, like this is it.

Play from 0:54:28 8:88:88 0:54:28 This is how you get it. And you know, when these models get good enough, like they're building reality, right? Like that's the whole thing. Like some of this stuff with the with the CAD models and whatever they're able to do and you can like interact with it, it's so crazy that you can build this. This one I thought was interesting. Someone used a a gun reference. I hope we don't get dinged for that. I'll take it off the screen. But they, you know, they they built like a 3D model of a of a you know, of a W E A P O N.

Play from 0:54:53 8:88:88 0:54:53 V two's gonna screen screen us for that. But I mean that's that's also the other side of it. Right? Like that's the dangerous side. Is that like that that now, if you have a 3D printer, I know you can buy I know you can get this stuff online, guys. I you I I've spent enough time on the dark side of the internet to t to know like what's possible. But the it's it's just kind of insane that it can build this stuff now because I've been I've been sort of following this. Like this, some of this stuff is crazy.

Play from 0:55:20 8:88:88 0:55:20 Right? Like that it can sort of dev and sculpt these characters and then rig them, that they can actually move this fluid and blender is really solid when it no when you know what you're doing. It's super complex. But I've just seen some really cool stuff. Did it once you once you get in there, you start to realize like how complex it is, and you realize why the 3D artists sort of hated us in the beginning. Because I'm like.

Play from 0:55:46 8:88:88 0:55:46 Yeah, anyone who can do this is like super intense and meticulous and smart and really talented to get to this level of, you know, that we're posting stuff. And I'm like, I get I see where the hatred comes from, but I feel like now they have some stuff. Now it's you guys you guys have a lot of stuff to use. Yeah. one of one of those creative crafts too that there it just takes so long to master and there's a long journey there. And, you know, to take away from that journey or to experience somebody maybe that's shortcutting that journey doesn't probably feel so good either.

Play from 0:56:15 8:88:88 0:56:15 Right. Like it it you know, you feel maybe a little bit of insecurity when that happens at the same time. Totally. I I I would feel the same way. Right. Like I felt and that's that's because it's such a meticulous process. You go in there and realize all the things you can do, and then like it's like, then there's the nodes, and then there's the sculpting, and then there's the materials, and then there's the lighting, and then there's the staging. And it's like, okay. There's like you're building fake reality in there, which is it has to be precise most of the time.

Play from 0:56:43 8:88:88 0:56:43 And that's what they're doing. But I think it's uh I just think it's like finally sort of the moment. For 3D, and I'm trying to find some of the other examples I I pulled here. But people are building, you know, image to 3D. They're building sort of text to 3D. I like that example too that you had 'cause I also had a bookmarked, I think Javi posted it, but it was that car and it's like Gaussian splat where you can click it and almost basically like, you know, manipulate the car, yeah, movement and I think it's go up maybe a little bit more.

Play from 0:57:17 8:88:88 0:57:17 But that like that was super neat. I was like, Yeah, there you go. That's what we're and like the interaction of everything that's happening around it too, not just moving the car, like all this other stuff gets whacked. Which is I don't know if you saw this. I don't know if you saw it. Sorry to take us off off track here. But I don't know if you saw this SpaceX video the other day of this thing, uh of this rocket landing.

Play from 0:57:45 8:88:88 0:57:45 Let's see if I if they have it here. I sent it, I mean I would I got. Maybe have to search space SpaceX rocket. Let's see. I mean I c I'm trying to get trying to find the oh god, where is this video now? Not gonna be able to find it. Space, you know, rocket, ocean, ocean.

0:58:12Can people still detect synthetic video?

Play from 0:58:13 8:88:88 0:58:13 Maybe this is it? Yeah. Like, I'm looking at this. Hold on. I know I ruined it by saying what I thought. but but there's no way that that's real. This is real. So this is apparently this is real footage. No way. I don't buy that. Are you getting the audio on that by the way? Let me let me let me cut it out because it's loud.

Play from 0:58:40 8:88:88 0:58:40 So first thing I'm looking like 'cause now you scrutinize everything, right? That cannot be real. I don't know if it I don't know if it is or not, but this is what I'm saying. It's like now we're at this point where I have to scrutinize everything because we know what's possible. Right, there's a few things here that that look suspect to me. Like I see these, you know, first off, the color of this seems way too good for for this. I mean, it seems like it's either retouched, maybe it is retouched.

Play from 0:59:07 8:88:88 0:59:07 Um, but like the steam right here coming out and the smoke, like the boosters, like that feels a little weird, even though the physics Could be right. The flame being, you know, I I guess the flame is that this this is what I don't get. Is like, why is the camera on on top of the drone? Like, why are we getting the wings? Normally cameras on the bottom of the drone. Like dangling. So that's weird. Right? Um, and then we're watching it and like I don't feel like the physics of the water makes sense of when it hits.

Play from 0:59:37 8:88:88 0:59:37 And I could be totally wrong. We could be totally overanalyzing this, right? But this is like the the world that we live in now, right? It's uh the water's all right. speed one more time. So it kind of looks like it's shot in sixty frame, you know, 60 FPS. Like I don't know if the water here maybe like the more I look at it, I I don't I don't know. Then like when it hits here, that looks good.

Play from 1:00:05 8:88:88 1:00:05 I I'm so c I'm so conflicted by this. problem with the first half of it the most. So because go like go back to the beginning one more time. This is fun. This is kind of a fun game. I like I feel like it turns upright way too fast there. Like that first five seconds, the like how fast it ships shifts upright feels very quick to me. Right? But also just like the look at the coloration of it and like the details.

Play from 1:00:33 8:88:88 1:00:33 Like is that just those little st like the like s no, just like evaporates too quick. Hold on, did I just see? Where is it? Is there is there missing shadows here? We have the lights coming from this direction. Look at us trying to play like forensic analysis here. And then uh okay, so the sun is coming from this direction. Where if it was rendered in a 3D engine, you would be able to see it. So you get the shadow right there. Huh. It's it's like here, this looks fake to me.

Play from 1:01:01 8:88:88 1:01:01 Like this this angle right here. It doesn't look it looks like it's simulated. I don't know. This is this is why I'm starting to get nervous. Cause it's i it's probably totally real. You know what I mean? And now we're now we're we're sitting here like, is this is this our new reality of like trying to cipher everything? go. This is a nice little uh segue.

1:01:21How does NVIDIA detect fake video?

Play from 1:01:26 8:88:88 1:01:26 But there was a there was a news release, I think, last week, but in NVIDIA introduced a real-time synthetic video detection. So it evaluates video frame by frame, reports 92% accuracy on uncompressed video, uh, with performance declining as compression increases and processing fast enough for some live stream applications. Uh Why should you care while detection is moving closer to infrastructure rather than after the fact moderation?

Play from 1:01:56 8:88:88 1:01:56 So I think that's an interesting point there. What is the name of the product? I'm just trying to see if we can find a video of it real quick. here, let me see if I can pull Let pull this up real quick. I got the article up here. Me let me stop sharing. Am I even sharing? No, I am not sharing.

Play from 1:02:21 8:88:88 1:02:21 Nice. is just saying advances graphic simulation with a genetic physical AI, blah, blah, blah. Okay, they got a whole keynote in here. Let's see if there's like a section specifically toward that or not. 'Cause this is this is gonna become more popular. You know, we're gonna question everything. I already do question everything. here it is.

Play from 1:02:49 8:88:88 1:02:49 Helps newsrooms detect synthetic video. Good. Cause it's getting too good now. So NVIDIA's doing something there. And again, if they are doing something there, and I thought sort of what we pointed out in terms of like what that is here, sounds like it's already decent. So it's only gonna get better. I mean, but we we we need it.

Play from 1:03:15 8:88:88 1:03:15 I mean, more more timely than ever right now. So this is gonna be a key part. But the the point too is like this piece was interesting. Detection moving closer to infant structure rather than after the fact moderation. Huh. How does that work? I don't know. I think it's gotta be this has gotta be integrated into platforms. Yeah, I d I I don't know.

Play from 1:03:40 8:88:88 1:03:40 I mean How are you gonna moderate that? It's just like I these gu again, like LinkedIn will be the first one to do it and then they'll have no content anymore. Then they'll have to open it up. If if LinkedIn didn't have AI, if there was no AI, LinkedIn would have like probably ninety percent less posts. I mean, that's what it was like when we were on there, right?

Play from 1:04:05 8:88:88 1:04:05 And it's the hardest part I think is like w if I were to it's one is like knowing what you're gonna talk about. And then, you know, when we were creating posts, when I even when you're thinking about like even if you're not doing a carousel, I mean, I used to spend, you know, like an hour or two on a post, you know, just writing out the idea first, doing like a pass of like, I don't care how it sounds, I just want to get all my ideas down, then like let me move, you know, move around, revise, edit, blah, blah, blah.

Play from 1:04:33 8:88:88 1:04:33 Let me find a good hook. Right. Like all those things and then post. Yeah. But I mean, like with the AI thing, right? Like people are able to shortcut that. They don't they don't need the idea. A lot of the ideas are already fed. They don't have the you know, they don't have to be great writers because they're mostly just taking what it comes up with, you know. Um, and logically it flows, but you know, the the AI patterns and verbiage and all those things are there.

Play from 1:04:58 8:88:88 1:04:58 So There's a lot of there's a lot of engagement pods on there. And it's not even just engagement pods, it's like companies that are running like the full pod that just like create all your content for you, but it's all just content that they've used for everyone else, just retailered to you, and then you get in a group and then whatever all their other clients are, they just it's like they all just sort of like cross pollinate with each other and like that's how it moves. It's just yeah.

Play from 1:05:24 8:88:88 1:05:24 bro. I mean, it's just like, Okay, you know, you're in this pod and then, you know, we'll like your post, but then you gotta like the last ten and leave a comment. And it's just like this is grouse. That's so they do it for the people who are also writing your posts are also doing that for you. Right? So it's it's like they're doing everything. And it's it's it's it's you know, LinkedIn's a LinkedIn's a a dirty dirty beast now.

Play from 1:05:50 8:88:88 1:05:50 But we've created And I feel like I said, I feel bad. Like I want to like when people I couldn't believe this is AI, this UGC video I created. I'm like, I can write ten instances in the first five seconds where it's AI. And I'm like, I don't wanna do that. I don't wanna be that police I don't wanna be the policeman of your of your content. Right? Like I hate when people do that. You should you shouldn't do this with your content. Like you know, go do your own thing then. Like I don't give a shit. I don't care what you think.

Play from 1:06:16 8:88:88 1:06:16 Like you're not this I'm I'm not you. Which which is why some of these people I'm just like I I see it. I'm just like let them just like let them have their fun. Let them bury themselves with this stuff 'cause it's just you don't You don't want to become like the elitist people that we were trying to sort of circumvent, right? Yeah. I'm like, don't do that. Or like, this isn't the full way to be thinking about this.

Play from 1:06:41 8:88:88 1:06:41 You know, or just kinda like like you said, just like policing. It's like, yeah. 'Cause I feel like I used to just cheer every single person who was doing stuff with AI and trying, right? And I liked it because people were open and honest about how they were trying it, what they were you know, what they were seeing, like it wasn't good or like whatever, I didn't like this. Now it's just like it's the best thing to ever I'd fifty X my, you know, in my gains with y you know, just all the standard talking points and just acting like they've they've done this for ten years and it's been around for three.

1:07:10Why does speed devalue AI creative work?

Play from 1:07:14 8:88:88 1:07:14 You know. we've spent so much time with this too that we know it like in some cases this is all hyperbole. I mean, can you do this quickly? Yes. Did it could it have taken you five minutes? No, it's not it's never as easy as you're making it sound. And also by you making it sound super easy, you're devaluing the the end result. Like I mean, that part I that I think that part is only good when the it's a new capability or new technology.

Play from 1:07:44 8:88:88 1:07:44 And and so like we had that moment with with with images and video. And it's like, man, look what we can, you know, like do. And this only takes five minutes. And I think like that there is some value in communicating that. But if that's your strategy of just all about speed and you're just gonna like that really long term devalues this whole thing. Right. Um it's not just about the speed, right? Like there are just so many different elements into it. We we've talked about it last week with Finn, which is like There are and the week before and with with Salma, and we're just like, there are gonna be parts of this where we actually wanna take back in control now.

Play from 1:08:17 8:88:88 1:08:17 It's not even about the speed anymore. We wanna take the speed where we wanna give it speed. We wanna, you know, there are gonna be times where things are gonna take longer, but we wanna do that intentionally, you know, and and then you got like the taste factor and you got the distribution. Like these are just things that like it's not just about the speed. That's uh I don't know, I've been I've been thinking about this more often, right? Uh there's there's something there's something to be said about what position we sit in and like how much exposure we have to this, 'cause if I'm a lunch lady, right, and I see this stuff about AI on the news and it says this is gonna take your jobs and we're gonna build robots and you're not gonna be a lunch lady anymore, right?

Play from 1:09:00 8:88:88 1:09:00 Like and then you go on social media and all you see is just crap. It's just like, you know. It's just all this stuff that's just like I'd use this, I built this in five minutes. I'm this is the best thing ever. But like no one actually has exposure to like the depth and like infrastructure that can be built, like we've seen, you know, like being inside of companies and seeing how they're deployed and the really smart people that work on it in different ways than us. And like when you can take months long processes that are intricate and have checks and balances and have like really quality sit like assurance built into it.

Play from 1:09:34 8:88:88 1:09:34 When you see that, you think about it differently than when you just see what's on social media. Right? 'Cause it's like it's and we're sitting here like trying to defend that while the person who's not necessarily in that is looking at, you know, Pikachu smoking weed on Instagram saying like, Why are we wasting water on this? You know, like this is I've I just feel like there's there's I know we always say we're in a bubble, but like really thinking about it, you haven't just There's no exposure.

Play from 1:10:03 8:88:88 1:10:03 Like, does the lunch lady outside of like, we can automate your entire meal planning, recipe, schedule, like inventory management, like what if they saw that and they saw what they're able to do, like, you know, telling telling Claude yesterday that I need to find a very specific part that for something that broke on my one of my, you know, consumer electronics and I need to find it today because I need to use it today. I need to find somewhere that delivers. I need to find someone that delivers in my area. I need you to track, go search a ton of websites, find the inventory.

Play from 1:10:31 8:88:88 1:10:31 You know, make sure it's on DoorDash or whatever and it can be delivered within three hours. You know, it came back and did all that, but like don't think your average person knows that it has the capability to go do that, right? Like it doesn't have the capability to go hyper target things in the like and it's hard to combat that as a useful way for AI. Like to it's hard to argue that that's not value adding to your life, right? Because yes, it's about speed, but it's also there are there's nothing in that process that is slop for lack of a better bucket here.

Play from 1:11:07 8:88:88 1:11:07 Uh and what I mean by that is like that taste or the judgment layer, or you brought up a really interesting point, which and I think you'll hear this a lot more. Which is this evals, like evaluation built into whatever system you're building. And you like you'll you'll start to see that now, like even with loops, you like like hearing about loops, which you know is basically like a part of your system that evaluates and it keeps going until it hits a threshold, right?

1:11:13How do eval loops prevent AI slop?

Play from 1:11:37 8:88:88 1:11:37 That's based on judgment and taste and you know, uh a a quality assurance. Right. And that prevents slop, assuming that you're you're you're evaluating for the right things. Right. So it's like moving upstream, like thinking about things like that, like that will be a bigger part, especially when there is these um these these big differences, these deltas between just mediocre, which is where a lot of people operate, versus like great outputs, which has to do with the SMEs, which has to do with context, which has to do with right, like getting directly to the source.

Play from 1:12:13 8:88:88 1:12:13 Um, and so I think like that's really interesting. Like speed only matters in those circumstances, assuming you've got the righty vowels and context and things built into that. Right. Like a perfect a perfect example of that. Like before this, we were just talking about like Content, right? Um, if people want to create content at scale, they can, right? They can get the slop, they can get topics that they're interested in.

Play from 1:12:40 8:88:88 1:12:40 But like if you really want to refine that process, you need to go to you need to find the right, yeah, you gotta go deeper, you gotta find the right topics. So how do you how might you do that? So one one one of the ways I do that is I have calls recorded with clients or prospects, and now I've got that transcript. And now I've got the context. It can pull ideas, content seeds is what I call them, out of those things of things I could talk about.

1:13:00How can calls create original content?

Play from 1:13:08 8:88:88 1:13:08 And that's coming from a per first person point of view. So my point of view is now out of the mediocre box. It's now in a very specific, unique box to me, right? Because I can talk about this client, this problem that they had. Here's what we talked about, here's how we solved it. And so, like it's like that sort of thing. And then also another example that Rory is like. The other thing I was showing you, which is like this idea of like, we'll take this transcript. We got to post this on YouTube.

Play from 1:13:34 8:88:88 1:13:34 What if it pulls all the context out of this transcript and that gives me content for us to post on LinkedIn or X or whatever? Those are coming straight from the source, right? Like that's that true source of information where we're eliminating a lot of that mediocre common content and it strictly comes from our words of mouth. We're subject matter experts. Boom. Right. So speed only matters if you solve for those things, is I guess my point. And you've got an evaluation threshold that you've established.

Play from 1:14:03 8:88:88 1:14:03 So it's so it's so deep now, like number one from a competitive standpoint, I feel like you have to go a little bit further, right? If everyone when everyone had a custom GPT, it's like, Well, how do I string three custom GPTs together? When everyone has one Claude skill that does this, how do I build a whole pipeline? When everyone has a pipeline, how do I have agents? Right? Like you just keep sort of going to make it further and deeper and bigger and you start to understand like How intense our brains actually are when we think about this stuff and you realize how much goes into writing a post and what context is needed versus what experience is needed versus context of what's worked before in the past versus what's never worked, you know, what's what it's never seen before.

Play from 1:14:42 8:88:88 1:14:42 So there's like this whole you know, the eval process is probably what's going on in your brain. Right. And it needs context too to understand what happened prior and why that's good. So there's you know, you It's not e that's why these that's why everyone is selling or shilling some content system. Right? It's like, I can I can write your LinkedIn posts or I can write your X posts and it'll be good. It's like, no, it's not good because you're doing very baseline stuff.

Play from 1:15:07 8:88:88 1:15:07 You're not doing anything that would make this remotely interesting for anyone. You know, it's nothing nothing popping out of there is out of the distribution. We'll call back to to last week. Yeah. right? Like going back to like if if you were to cause I I am working on something like that, but if I were to if I were to actually productize it, it would have to be yeah, it would have to it would have to really leverage the the unique aspect of whoever it's serving for, right?

Play from 1:15:37 8:88:88 1:15:37 Because It it needs to be built off those things in order to truly write in your voice from your point of view. And even then it's not going to give you a hundred percent, right? So like we can go down this whole path of agents and loops and graphs and the evals and like wherever it's going next. But what I'm starting to figure out now is there are just some pieces. Like as much as we just wanna have like this full circle closed loop that is perfectly automated, there's always like a few little things that just like I I can't.

Play from 1:16:07 8:88:88 1:16:07 think I can solve for that. Like I I need to be a part of this and that, even if I don't want to be, even if I want to make that choice. So like I'm just finding there are still breadcrumbs out there that are variables that can't be automated. And people try to automate opinions, right? Like they're trying to automate their basically their opinions where they're having AI write this and whether they believe it or not, it's w it's weird to me that you're giving away that much power to it.

1:16:20Why can't AI automate your opinion?

Play from 1:16:33 8:88:88 1:16:33 Where it's like, here's this content that I'm gonna write. And really it's just I don't know, I don't know the best way to explain this. I probably need you know, better better vocabulary. But it's it's like when you it's like, I'm gonna write about this topic, like this is a topic that's interesting, you know, go write this post, right? It writes the post, but It's really surface level because there's actually no it's just it's it's optimized for writing, not for not for thinking, if that may.

Play from 1:16:59 8:88:88 1:16:59 I I don't know if I'm saying this the right. Do you get what I'm saying? Where it's it's like s structured formulaic writing prose, like it's not thinking of like how do I inject emotion or like a curveball in here or like it's just following the rules that you've provided for it. It's not it's not, you know, abbreviating words on purpose to like chop up a sentence or make your eye go somewhere or like unless you tell it to.

Play from 1:17:25 8:88:88 1:17:25 But I think I'm not I'm not fully explaining this one the way that I mean it. I'm just uh saying like it's it's coming up with an idea and it's just writing because that's the way it writes, not because it has a feeling on it or an opinion on it that you know I it's so hard. I gotta I gotta flush this one out, guys. I gotta flush this this out. I can't I can't get it out. Yeah. though. I like that uh I like that process. I I like that angle of the way that you're thinking about it.

Play from 1:17:51 8:88:88 1:17:51 I think that's the the line for me is like formatting the output for thinking, not just the writing. Um and I've just run this experiment with like this particular system so many times where that is so true to me. Like I I I've seen that. So it ultimately comes down to. And that's why the the unique aspect is so important. Because if you've talked about a problem that you've solved, it knows how you thought about it, right?

Play from 1:18:19 8:88:88 1:18:19 Because you've probably talked out that process, which kind of gets you closer to what you're saying, which is format it for the thinking versus just the writing. And that's when, like, I just think this writing thing is like probably one of the toughest. Things to crack here because you are not just like you have to think beyond like yes, you have to think about the writing, but yes, you have to think about the thinking. Right.

Play from 1:18:44 8:88:88 1:18:44 So you have to set up these rules of here's how I write, you gotta give it examples. You here are the like I I have a whole kill list of AI LLM writing patterns that I weigh it against, right? So that's an eval thing. You know, like there's a so like there's just many, many parts to this. Like the the here are verbal ticks that I say, right? Like here are phrases I use. Here are words I never say, and here are the words I usually say instead of it. Like I'm not ever gonna say however, but maybe I'll say but, right?

Play from 1:19:10 8:88:88 1:19:10 Like, so you know, it like those are like you've really got like this big problem, you know? And um, you know, maybe one of these models will get there, but like I the deeper I go, the the more I realize there's just more and more things that you need to have um really contextualized for lack of a better term, in order for it to operate at the highest level. Like. I think there's if we go if I I have to go back and look at this, that we maybe 20 episodes ago we looked at that that comedy writing pipeline that I had sort of just went nuts on with Opus 4.6 when I was like, I can go crazy on this if I want to.

Play from 1:19:47 8:88:88 1:19:47 And there was so many little things that I took out of that whole system that made the entire writing system better, which I don't really write that much anymore. But when I want to write something, it's good because I stripped it. I mean I built the whole thing and I stripped it down into parts. two to like have them. Like remember we talked about the analogy engine, right? Like that thing is great to inject in. The consensus engine, which is finding like the the polar opposite sides of both, you know, both sides of the conversation, and then finding like a common middle ground where we can make fun of both.

Play from 1:20:17 8:88:88 1:20:17 Like that was to me there is like an interesting thought pattern in that or process in that 'cause my writing comes out less. Or Yes. I the idea is that the model would never suggest that particularly. You know? Like that would be something you would have had to note, like an observation, and then you would have had to contextualize it somewhere in order for it to then present that idea to or know how to build on top of. Also the writing when people are saying like we're gonna write a post.

Play from 1:20:45 8:88:88 1:20:45 Like I think that it, you know, it triggers that it triggers that sort of mentality like as opposed. Correct. And it's gonna go there. And it's like, I'm writing a LinkedIn post, I'm writing a social post, I'm writing a script for an ad. Yeah. So you gotta it's like you have to take it in different directions. Like I'm writing, you know, an editorial opinion article or something like that, or I'm writing, you know, blah blah blah or I just wanna have like I wanna have like a social social dialect simulation or social dialogue simulation.

Play from 1:21:12 8:88:88 1:21:12 Like there's a lot of different ways you can probably trigger different words and context and emotion out of it, but we're only using like the surface level stuff because it's just so easy to do it. Write an email, write a social post. I'm gonna build this whole content system for LinkedIn. And now it's all it's doing is thinking LinkedIn, LinkedIn, LinkedIn. How do I write professionally? How do I do this? How do I, you know, structure it within the character cons you know content? No, it's not thinking about like what to actually say, it's thinking about how to actually say it.

Play from 1:21:38 8:88:88 1:21:38 And that's that's the Or if you don't have the opinions, right, that's why everyone's stuff looks exactly the same now. It's just like reading the hot topics. Here, these are the hot topics that I'm seeing everyone post about. Write write a LinkedIn post. And whatever skills they're using that are still pumping out the same thing as everyone else. Right. So I think we're like the reason we're talking about this is because this is this is when you get deep enough in here, some of you will know this, but it's just like you'll you'll know exactly what we're talking about now.

1:21:54What should AI content systems ingest?

Play from 1:22:08 8:88:88 1:22:08 And you s and you see what's gonna happen here if you don't fix those things. Um and so like I I would just say like this is these use cases aren't gonna blow you away. But what I would suggest are the are like a couple of the best things you could do right now. That don't take a lot of work is ingest your calls, right? So granola, fathom, whatever it is, have something set up for that. Schedule task, it's gonna pull those calls in every day, run it, right?

Play from 1:22:37 8:88:88 1:22:37 Learns off of those. Now you've got, again, content ideas, you've got the way that you talk about things, topics, angles, right? That are relevant to you. These are gonna compound over time, right? Day one, it's not gonna be this mind-blowing thing, but. quarter or two from now, a year, like it knows all these things about you, assuming you're good with that, right? Like that you can now have as leverage. Number what so that's number one.

Play from 1:23:02 8:88:88 1:23:02 Number two, do the same thing for emails, ebo and jest. Like that's another one I set up. So pulls my emails. Now this one I had to set up like I don't want it to pull all my emails. I want them to identify people that are prospects or clients or there's there's a relationship opportunity there. Mm-hmm. I also have a system where it's like I'm approving, hey, this should this person should be on this. No, this person is not, but it's actually done a really good job of kind of contextualizing what it thinks it should categorize.

Play from 1:23:31 8:88:88 1:23:31 So now I'm just going, yes, no, yes, no. But like those two things alone, perfect for giving you, for for just injecting mass amounts of context, personalizing this, giving your unique expertise, leveraging the SME stuff, right? Content seeds, blah, blah, blah, blah, blah. Those are like, do that. And then the only like one other thing I would say, Rory, too, like these would be just like table stakes for me. Anybody listen to this would be just a prompt amplifier skill.

Play from 1:24:00 8:88:88 1:24:00 So mine started as just like a skill. Now it's turned into like this whole graph where it's connecting to other skills to find the right context or the right thing to do based on that. But it's essentially I am prompting like normal, assuming I've given it enough context. It's going to basically amplify that prompt and give it the best optimized, informed, specific, clear prompt it can for it to run it.

Play from 1:24:27 8:88:88 1:24:27 And I can I can even just say dash dash review and it'll show me exactly how it amplified it before I say go. But like that alleviates a lot of pressure from me and mental exhaustion of trying to think of every single thing. Cause now I can just type the the bare minimum in in terms of context. And it usually comes up with things I'm not thinking about, I couldn't elaborate on. And boom, like shortcut process there that again, like has insane value and it's not like anything like crazy.

Play from 1:24:55 8:88:88 1:24:55 Anyone can do this. Like. I th I think maybe a fun episode. You guys can say otherwise in the comments. I think it's fun. Is like the non-creative stuff that we use AI for, like that all this like supplemental admin stuff, which is really once you do like once I see what you do and I'm like, that's cool. Like, you know, it's it's different because it's not like I'm creating these videos or these images that are awesome. It's like these are these little things that l legitimately allow us to do what we do is like this stuff.

Play from 1:25:23 8:88:88 1:25:23 So like there's you know, going back to before the first time I got on stage Right, like three years ago, it was a lot of ops stuff. It wasn't just image and it was the the one thing was the sales calls, what you're talking about. Like we were pulling we were making optimizations from sales calls with ChatGPT three point five. Because it was we could dump all that data in there and then say, like, you know, if I was taking twenty sales calls a week, which I was for the agency at the time, we're stripping out everything.

1:25:27How can AI improve sales calls?

Play from 1:25:50 8:88:88 1:25:50 So we'd strip out all the here make a list make a list of all the questions, right? Like this is back in three point five. I could do Million different things with this now. But make a list of all the questions, and then basically we'd we'd run that list and we put it into a spreadsheet and then we'd have it go and then optimize for like what the actual underlying real question was. You know, like what that is that a price objection, is that a this is that of this? Is that you know timing, whatever? Um, or if it's just like a really good question.

Play from 1:26:18 8:88:88 1:26:18 And I remember we built a lot of stuff with just the questions. Like we built whatever questions were at this asked the most, we built it into the sales script. So we were basically answering questions before they were asked. Then we'd start to build that into outbound marketing. Whatever we f we felt like was the most asked question, that would be forefront in the marketing. And it was like we were cheating at the time. Like people didn't know that we that we were cheating, right? Yeah. because most of these companies don't have the stuff set up. So keep going.

Play from 1:26:44 8:88:88 1:26:44 Yeah. And it was like there was that. And then we realized, if we dumped in all the the call data and the sentiment analysis stuff that we could get from Fireflies, it would tell us when people were starting to either bridge or lose interest. Like typical sales call process. Most people have been on a sales call here, I'm assuming, right? You go through the disc like the niceties, the discovery phase, and then like the pitch, right? So typically I would notice that right before the pitch, you'd see the tonality change, everything would It's like sort of slow down to a to a halt, right?

Play from 1:27:15 8:88:88 1:27:15 From like an excitement standpoint. So I just decided one day, let's just flip the price. Like, let's go to the beginning. We'll do the niceties. Instead of like discovery of just asking a bunch of questions, like, let's just do an audit here. Like, let's just audit an account. Just go in and find stuff totally unstructured. Let's see how they handle it. Right. And we're just talking about their problems, not like an interview where it feels like you're on the defensive already. And then right after I'd get that, I would just drop price right in the middle of the call.

Play from 1:27:42 8:88:88 1:27:42 So it would be like, you just heard everything that I saw. You just it's like you and I were vibing. Here's the price. And then I'd go, like, yeah, totally. We can help you with that. 100% typically what we do price wise is this. Then I'd go into all so this is how it works though when we when we actually work together. So like bury it right there and then go into here's who you'd work with. This is typically how we function, like we talk twice a week. You're gonna be here. Once you sign up, like this is how it goes. Like you get this on like everything had been taken care of.

Play from 1:28:09 8:88:88 1:28:09 I'm like, so at the end, I'm like, any questions? And they're like, Uh no, like that's great. And then it ends on I told them exactly how we were going to like work together instead of here's my price, like make a decision right now, like on the spot. So it it left like it just feels more comfortable at the end. You don't feel like you're like uh required to do something on both sides. And it was just like I'll send you, you know, I'll re record the audit for you, I'll send it back over.

Play from 1:28:38 8:88:88 1:28:38 You tell me what you think. You wanna jump on the phone next week and talk about it, up to you. If not, take this. Take this thing and leave it. You know, take it, the audit and have it. It's for you to use. And sales jumped insane because we were just taking this AI data and we were like optimizing it. So long story short, it's like you can get creative with what you see. You don't have to just, you know, use AI at its word. It's like it would have never told me to drop price in the middle of the call. I was like, it's just getting here.

Play from 1:29:04 8:88:88 1:29:04 Let's just see what happens if we drop price in the middle of call. So there's a ton of stuff. And we never, we've probably never done a sales W Yeah, we I I think somebody was wasn't somebody asking about um like more of like our business process anyway. Like they were just like, Can you guys get into like maybe the business stuff or like maybe how, you know, with prospects or clients or maybe yeah, like CRM stu you know, pipeline. I like we we could get into I'm sure you I mean we're both me and you are both doing our own thing.

Play from 1:29:33 8:88:88 1:29:33 So we've we've got things there. Um and so if that is appetizing, let us know. I mean we're we're happy to kinda kinda go into that route. And um There's a lot. I mean, there's a lot, dude. There's a lot, there's a lot here, man. I I I think at this point our your prompts shouldn't be starting you your prompts shouldn't be starting at zero. There should be a context file at a bare minimum that you're working with almost always, right?

1:29:41Why should prompts stop starting at zero?

Play from 1:29:58 8:88:88 1:29:58 Unless it's just like a simple like Google level query that is a one time thing. There are still times we all use LLMs for that. But every everything hypothetically should be compounding, meaning there's a context file at a bare minimum. And then you've got a system or you know, agents or what what whatever it is. And I think the way that maybe like a quick way you're gonna notice this is like your prompt should be getting probably shorter now because you could call on a skill, you could call on an agent, right?

Play from 1:30:24 8:88:88 1:30:24 Or it's already recognizing a process that exists. Um, or it's you know, like I I can mention three skills like in a prompt, right? And I'm already saving probably 80% of what I would have had to type in order to get that. So I think like that's one easy way you know. That you're starting to make some real progress there if if you guys are starting to think about those things. And I don't know. I want to, I don't know. We could, we we, we could, yeah, we could go, we could go hard on this.

Play from 1:30:52 8:88:88 1:30:52 I just think like too, like one other thing to to note, if you are getting started with that, like in the skill stuff, and we we can we can really go deep on that. But it's just like also audit the skills that you're putting in there. There's a lot of bad skills that are probably out there, and even if they are good. Always ask how you can improve them or make them even more relevant to you. Right. Like benchmarking against what already exists, build on top of. Like I I view all that as raw material that I can make even better.

1:31:08How should AI skills be customized?

Play from 1:31:19 8:88:88 1:31:19 If I see a skill out there that I think I would use, I'm like, this is good. Now I want to hyper tailor this to me, or I want to like even five, 10X this. How could I do that? ID8 with me. I'm interested in these ideas. Let's build that. Right. Like Most of the time the skills aren't built for you. Right? Like there's someone else someone else open sourced them, but it was for their specific task. And yeah, there's general ones, but that's what everyone else is using. You kind of can just take it and customize it and be like, I like this idea.

Play from 1:31:46 8:88:88 1:31:46 Like what else can we sort of how else can we take this? You know, based on knowledge of me or whatever. And this is very simplistic, by the way. 'cause most of the stuff I have now is just it's just if I've done something once and it shows up on my radar again and I have to do it again. that I'm building a process. I'm just I'm just not doing the same thing twice ever again. Like where dude it it's taken on like eight iterations of of where it's gone to like trim it to trim it to trim it to trim it.

Play from 1:32:21 8:88:88 1:32:21 And it's all still the same levels of control. I'm just d operating it in a different way. Same process. Basically the same exact process it's been for eight months. You know, base characters Environment, composition, editing, right? It's just like it's n new tools, new iterations, new ways to f sort of figure out to build the skills around it. But it's that's one thing, right? Like I I don't even think we realize probably how many things we have set up that help us day to day.

Play from 1:32:48 8:88:88 1:32:48 There's no way to be able to do this five years ago, would I would I get it done now. I mean, like like if you're in house somewhere and you're not doing this day to day, I I you just like I I'm spending whole days sometimes, like like even just this past week. I probably spent two or three days just auditing my entire ecosystem and like fixing leaks or breaks or optimizing or you know, like it's like, dude, you know, like that's not sexy at all, but like those are rabbit holes that also improve the health of your environment and give you better output.

Play from 1:33:20 8:88:88 1:33:20 So it's like If you're busy and you're in-house and we're doing meetings and you got other projects going, like you just don't have the time to do this shit. You just don't. Yeah. I think it's I I think like if you are doing this stuff, like don't over engineer and go crazy. Like you're gonna I'd say like scrap things as much as you build 'Cause if you try to keep operating and consistently using this stuff when it's not, when you don't use it, it's just like get rid of it because it's I've built a ton of stuff where I'm like, this is great for the immediate, but then you know, there's certain things that I keep coming back to.

1:33:29When should AI workflows be deleted?

Play from 1:33:52 8:88:88 1:33:52 Like I have a I built a local a local uh social media video downloader that I can just run, just drop a video into my little tool anytime I want to. Instagram, Twitter, YouTube, whatever. So I can just download it because you don't want to use those sketchy sites online. Right? Like that was something that I built on a whim, like I don't want to use this sketchy site anymore. Can you build this? And it was like, yes. But I use it all the time, you know. And then I was like, wait, well what do I what do I store this stuff? This is just gonna like randomly downloading my downloads.

Play from 1:34:18 8:88:88 1:34:18 And you start to build a little bit more. It's like, it should go here. Then what should I do with it once it's there? But that's where I think there's a trap in that also. is also a trap. Yep. Yep. I could just stop at that downloader and that's a little tool and that's all I need. But like I'm still kind of like, What do I what else gonna do? What else gonna do? That's it did a long time ago. Yeah. remembering all the things that you've created. Like the this this, you know, these dashboards and the interfaces aren't really built yet to to completely give you that visual of everything that you've created.

Play from 1:34:48 8:88:88 1:34:48 Sure, you've got all the skills, but like you forget about them. And then, you know, they might get stale and they might not work as great anymore because you've changed other things, right? Like you're always quality checking and I'm always asking to double check work because I mean it's And then you gotta log you wanna log those lessons so they don't make the same mistake again. You know, like, dude, it's just it's infinite. It makes you go crazy. But I yeah, maybe we do a whole episode on that. Like, that would be that would be nuts.

1:35:12How do you build an AI skill vault?

Play from 1:35:15 8:88:88 1:35:15 I now have built a pipeline to deconstruct all of my skills into individual skills to just bank them and archive them so that like if I'm saying like to the orchestrator, my my little agent orchestrator that I utilize. I want to build this. Also then like just pop through, you know, the archive over here. And I have it I had to build like a like a catalog in the archive for so if I build like this, you know, this crazy pipeline and maybe there's like six steps to it, but there's probably like maybe eight or nine supporting skills in it.

Play from 1:35:46 8:88:88 1:35:46 It'll strip those eight or nine supporting skills and isolate them so they can be used interchangeably in any other pipeline that I want. But then I had to because it was taking too much, you know, you learn this stuff. It's way too much token context when just reading through the whole thing. I had to you know codify a system for it to read through and understand what's in there first so it can always pull from this stuff. Because that's like the stuff that's been approved. Like I only do that if it's approved. Like it works. It's worked ten times. works really quick and it's good.

Play from 1:36:12 8:88:88 1:36:12 Then it goes into this little you know, like my little seed vault. Like one of those things that you know, the the the doomsday vaults that have like the seeds are Yeah. It's just like keep it back there for you know for a rainy day when I need it. But it's you go crazy like we all realize how insane we are when you open up our skill files. Open your skill files. Let's see how crazy you are like a psychopath right now.

Play from 1:36:37 8:88:88 1:36:37 Yeah. Yeah, I'm on the hunt, man. This is crazy. It's true. We maybe we should do that episode. It'd probably be fun. Probably just be like I don't even remember building all any of this stuff. I think it'll prove our point too. Like there are there are things that like are absolutely 10x worth their value. There are other things that we created spent a lot of time on that we just don't even use and ended up being like probably worthless. Well, I mean, to maybe to us, but um, yeah. Anyway, Rory, I'm thinking like we should call out Buc-ee's, we should call out some Buc-ee's subscribe.

Play from 1:37:09 8:88:88 1:37:09 Like, we need some Buc-ee's employee subscribers. Yeah. I think that's who we have to ask this time. Ask your you know, if you're in any of those states that Buc-ee's is available. Where's the closest Buc-ee's to you? Now that I ask. Cause I'm not really sure of what their their placement. Like okay. Like how far is that for you? By car five hours? Yeah, five hours, five and a half hours, six hours maybe?

Play from 1:37:37 8:88:88 1:37:37 Depending on traffic. Yeah. We got like we got some. I think the close the closest one I think right now is like an hour. Yeah, you have to have some within driving distance of you. That there's no space here for a Buc-ee's like you couldn't even fit one in the the tri state area. wondering where the the closest one will be. There's no there's no open land in these three states here. But that's that's a good one. We should shout out the Buc-ee's employee. But you know, because we have made it this far, and you've made it this far, typically, you know, if you make it this far too, you also get like the random stream of consciousness nuggets at the end of there.

Play from 1:38:10 8:88:88 1:38:10 Yeah. The best nuggets. So that's we don't do that on purpose. We just get to it at some point. Figure out what we actually want to say by the end of the episode. Um, but that being said, if you're still here, you gotta do just like one thing. Let's say one. Like it, subscribe it, tell somebody, tell your Buc-ee's you know your local Buc-ee's employee. maybe Oh, did we? So like we we we passed fifty K.

Play from 1:38:38 8:88:88 1:38:38 Next milestone, a hundred. And uh I think we just went over fifty two. So please help us out. Please help us out. I think I think we can get there before the end of the year. I think so. I think so. You because if you guys want to see more of us, that'd be cool. 'Cause then we could probably record more than once a week. If that's the case. See that one there. But if you you know, you guys have been doing great. Just number one, we'll say that.

Play from 1:39:05 8:88:88 1:39:05 Uh the the amount of comments and DMs and things that we've got we've we've been like re inspired by it 'cause it we we don't realize how many people listen and what you know, we we only see the number. That doesn't like really have any tangible feeling to it. So if you say something, it makes us feel, you know. We're like we're doing something important. But it's been good. We appreciate the feedback on that stuff. We appreciate the ideas, the guest responses, the liking, the sharing, the telling your quantum physicist, 'cause it all helps us like stay motivated.

Play from 1:39:34 8:88:88 1:39:34 Yeah. that was maybe that was the delta like we we we got those two thousand it was like the quantum physicists heard what was going on. Yeah. But think bigger and better things coming up. Let's wrap it up. baby. I think we like I'm excited about the future episodes too, the gas, like we mentioned at the beginning. So um yeah, man, this is a great way to end it. And um hope you guys have a great weekend and we will see you in the next one. Peace.