Ep. 81: How We Use AI to Run Our Businesses With Skills, Agents & MCPs

0:00:00Two sleep-deprived founders, no guest

Play from 0:00:01 8:88:88 0:00:01 What's up everybody? Past Hours Podcast episode 81. Rory Flynn, Drew Brucker. No special guest today, man. Just us.

Play from 0:00:10 8:88:88 0:00:10 No special guest. I have a special chair issue, though. There we go. All right. Sounds like a creaky chair, bro.

Play from 0:00:16 8:88:88 0:00:16 Mine's Mine's creaking quite a bit. Uh, I'm a I'm a mess. I'm a total mess. This is, you know, we're now on like week four of no sleep. So, like every I forgot, you know, we're traveling to my mother-in-laws. forgot my computer charger. My screen share wasn't working. We just had to have codecs debug my terminal to to get this thing to work. I, you know, I'm hallucinating. It's like a whole It's a whole thing if you guys can't tell.

Play from 0:00:41 8:88:88 0:00:41 Yeah. But yeah, man. [laughter] Happy end of summer, I guess, right? Well, you do get an extra day. I don't know if it's a good thing or not. You do get the extra day this weekend for Labor Day.

Play from 0:00:53 8:88:88 0:00:53 Is that a good thing? I don't know. But you've you've said this before that the weekends are now more more like exhausting than the weekdays which

Play from 0:01:04 8:88:88 0:01:04 the weekdays are the weekend. Yeah. So on one hand it's like you get five you know sort of things to look forward to but [laughter] the ironic twist behind that is it's it's work.

Play from 0:01:15 8:88:88 0:01:15 So yes. So you want you want to hear something funny right? Like this is this is how I know I'm getting old, right? And I had a my wife and I went out to dinner for the first time in I don't know maybe like two months uh last week and your boy had a couple dirty martinis. Um I you know I was just like screw it. I'm getting a couple dirty martinis. Shouldn't have ordered the third. Um but I did. Right. And after like not sleeping for a week and like that being my first foray back into it for probably like two weeks not having a drink or something.

0:01:16Martinis, $25 cocktails and vintage Bud Ice

Play from 0:01:44 8:88:88 0:01:44 Dude, I had I had this hangover the next day that I've never before. I dude, not like not like throw up, tired, anything. I had like severe leg pain and I'm like, am I am I getting like a blood clot and dying [laughter] to go? I had to have like Gemini or Claude diagnose me. I'm like, why are my legs in such pain that I can't walk? And it's like, well, at your advanced age, you know, like one of those things

Play from 0:02:11 8:88:88 0:02:11 advanced. Yeah. at your advanced stage when you when you're this dehydrated like apparently you're you know you flushed all your electrolytes out of your system by peeing it out and now it just like it's your lactic acid is isn't regulating so you're just getting like constant leg cramps dude I was I was like the whole car ride home I I was like this is insanely terrible I've never felt this before

Play from 0:02:34 8:88:88 0:02:34 getting leg cramps yeah like bad hydrated or like it band not is it band on the outside of the thighs like that all the way down

Play from 0:02:45 8:88:88 0:02:45 and just like I'm like limping. I'm like what am I 60 years old? I had three martinis. Like the hell is this, you know?

Play from 0:02:51 8:88:88 0:02:51 Dude, that's insane. Guys, I promise you I'm not that much of a lightweight. It's just it's it's dis You've lost your edge, brother. You've lost your edge. I mean,

Play from 0:03:01 8:88:88 0:03:01 uh three more to fraternity guys here that can no longer apparently have more than two drinks. But

Play from 0:03:08 8:88:88 0:03:08 yeah, what do we know? It was also, you know, like at this at this age, I could have three martinis and wake up the next day, be a rock star, or I could have one beer and be dead for like two days. Like, there's no predicting it anymore. Um,

Play from 0:03:23 8:88:88 0:03:23 it's like a magic eightball situation. You don't even know. You just Yeah. Yeah. It's up to chance. Feel it.

Play from 0:03:29 8:88:88 0:03:29 Yes. Dang. And you know, I I looked at that I looked at that dinner bill after the fact and I was like, I I guess I did have three martinis cuz

Play from 0:03:38 8:88:88 0:03:38 Bro, these drinks on these menus [laughter] though, I mean, they're like, you remember when those mixed like the cocktails and the mixed drinks like, you know, every restaurant's got their own sort of like list. I remember when those things used to be like 10, 11 bucks, you know, sometimes 12, 13 would be like a high roller. I mean, now you roll into any place, it's just like

Play from 0:04:00 8:88:88 0:04:00 18, 19, 20, 21. It's like you got the Georgia discount down there. This is, you know, you can't get you can't get into New York without sub $20 cocktail, sub 20, you know, sub $18 standard.

Play from 0:04:14 8:88:88 0:04:14 It just doesn't even make sense anymore. You just got to buy like it's like the bottle prices for for wine hasn't increased. Like I can still get like a $50, $60 bottle of wine. like no longer like there used to be like a you know like that small upturn you're like okay yeah like I'm out

Play from 0:04:31 8:88:88 0:04:31 right the I'm paying for the convenience I'm paying for the service that gap has gotten so big now that I mean

Play from 0:04:38 8:88:88 0:04:38 I got charged like $750 for for uh uh MGD the other day and I'm like this is like I'm drinking piss here this shouldn't be $7 you know I do like MGD though for anyone who's uh you know Miller Genuine Draft I do like a good Kors banquet beer Oh, I'm I Me, too. I like I'm a high life guy. Give me high life. Got a good one.

Play from 0:05:00 8:88:88 0:05:00 Yep. I like those those older versions of the beers. They're good. Bro, get bush. Give me some bush heavy [laughter] good bush ice from the '9s.

Play from 0:05:12 8:88:88 0:05:12 Nah, I'm not going that far. I'm not going that far. Now, here's a here's a quick story for banter. when we were in high school and we first were, you know, exploring the alcohol, you know, in 15, 16.

Play from 0:05:24 8:88:88 0:05:24 Yeah. For what you could afford, right? Yeah. We were playing beer pong in my basement and it was the first time we had probably like been like super competitive about it and, you know, just my dumb friends in high school, like 10 of us maybe. And my dad on this uh on this shelf, it it had to be from 1996. It was like a like a vintage Bud can and it was, you know, we're talking like maybe 15 years old. It just been there.

Play from 0:05:53 8:88:88 0:05:53 I don't know how long it been there, but we played for the loser has to chug it. So, it was like one of those things where

Play from 0:06:02 8:88:88 0:06:02 what if it wasn't what if it wasn't able to chug it? You just like got to sip it. That might even worse. Well, we we're at 15 and idiots, you know, but my boy Chugged and threw up and it was I don't know what kind of toxins he ingested with like ferment like that probably he was so messed up that night. It was probably because that that stuff had like 100x fermented in that can just like hadn't been refrigerated.

Play from 0:06:28 8:88:88 0:06:28 It's probably still paying the price for it. It's probably got some long-term effects as well. I would I would guess

Play from 0:06:34 8:88:88 0:06:34 like rearranged his DNA or something like it was that you know that I mean you don't even know you know it's it's a it's a mystery. I remember man I mean that's why we got to go too far in these college days bro remember it was just like man just just doing the the beer bongs you know and then like we had there were there were a few people that would do three at a time you know like and that was insane cuz you know all that air gets trapped in you know

Play from 0:07:03 8:88:88 0:07:03 it's not even about like just being able to do it it's just like that that hurt that comes with like trying to release all the air and the gas that comes with it. Um man, just just crazy stuff, dude. But um

Play from 0:07:15 8:88:88 0:07:15 get those epic those epic burps. The epic ones. The ones that you remember. They rock your body. Yeah. Like you you just move different after that, you know? [laughter]

Play from 0:07:25 8:88:88 0:07:25 But that's funny. We used to be young. Now we're not. We used to be young. We used to be cool, you know? We used to brag about how much we could uh can't do that anymore.

Play from 0:07:36 8:88:88 0:07:36 Just like, you know, wake up and and feel like a human. That was that's all I think we ask nowadays. That's probably it. But we'll we'll get there. We'll get back there, guys. I'm sure like anyone who's who's got kids that are in like college and stuff, they like, "Yo, you'll be back. You'll be back."

0:07:51Football is back, and so are the AI models

Play from 0:07:52 8:88:88 0:07:52 Yeah, that would be that would be a nice little reincarnation, you know, like moment, you know, when that when that time comes when your kids are old enough and they're in college and you maybe go visit them and jump back in, you know, just for a temporary period of time, you know, that would be nice. But yeah, man. I before we before we jump in too um just quick shout out we got football season is here. All right. American football is here.

Play from 0:08:17 8:88:88 0:08:17 American football. So I got my Rams. I'm supporting my Rams. Rory's a Giants fan. And the Rams and the Giants play on week two. So I think for that episode we need to we need to maybe come prepared.

Play from 0:08:32 8:88:88 0:08:32 I'm gonna put on the full regalia. Yeah. Yeah. Yeah. Yeah. Even though I'm excited, man. guaranteed to lose.

Play from 0:08:38 8:88:88 0:08:38 This is the best time of the year though, right now because now we're getting ready to get football back, college college and professional. Then we're also going to get college basketball back. This is h this is this is the the best part of the year from that perspective. Like you know, summer's like a slow grind because all you got is baseball, you know, but I'm looking forward to that, man. This is this is a fun time. Also, this is a fun time segueing like that into

Play from 0:09:06 8:88:88 0:09:06 transitions. Yeah. Yeah. Like, dude, these models are are going crazy. Are we are we at the point now with Fable 5.1 and, you know, supposedly Astra is supposed to be coming today, so maybe by the time that this releases, people start to get their hands on it. I know some people have had access for a while. Um, but are we at the point now where we need to start thanking the robots, saying please and thank you at this point? I mean,

0:09:09Fable 5.1, Astra and the next AI model wave

Play from 0:09:34 8:88:88 0:09:34 we're we're we're probably about there. I'm curious really to see what these models can do. Astra specifically, I mean, Fable's been good, but you know, like there's been these incremental improvements, right? Like, and obviously everything with this AI hype, you know, it gets hyped up at first and then it's like how long does this hype last? You I think Fable was, you know, still it gets talked about quite a bit. I think for good reason, but it's been a really it's been a while since I've been like really really pumped about a new model release.

Play from 0:10:05 8:88:88 0:10:05 Um and so I think we're kind of getting to this point where there are just going to be so many more use cases unlocked. There's the the barrier to entry will be easier. You don't have to be um sort of like a prompt expert to to know what to think. I think like these just models are going to be really good at anticipating needs, you know, second order thinking, um, quality level, coherence, like all of those things go up. But I I just want to play with Astra. I mean, we didn't get we neither one of us have access as of this moment.

Play from 0:10:37 8:88:88 0:10:37 Was hoping to maybe see if we could get that before the show, but even if we did, let's be honest, we probably wouldn't have time to play with it. But I I am excited to kind of see what's what. I mean obviously like you've got some of these people sharing you know examples of what it can do right now. We can kind of like briefly talk about this and then like what's we can also bring up what we're going to talk about today. So we're we're going to use this episode too to go deep into I think the LLM ecosystem environment. So like we've talked about this in the past, right?

Play from 0:11:08 8:88:88 0:11:08 Where we had an episode that was sort of focused on that specifically. There's so much there on the business side, the creator side, right? like what we're doing with clients, um what we're doing maybe even our own personal stuff. So, I think we're both going to do some show and tell on that today. So, stick with us. I think this is going to be like one of those episodes you're really going to want to dig in on, spend some extra time with. Hopefully, we're providing you with like some educational, inspirational stuff there. But as far as Astra goes, you know, I just like bookmarked like a few things, right?

Play from 0:11:38 8:88:88 0:11:38 Like some of these people doing these roundups of like, oh, this is insane. you know, your typical thing that somebody's just copy and pasted every time a new model releases and then they find examples.

0:11:43Why 3D and spatial AI could be next

Play from 0:11:47 8:88:88 0:11:47 But I've been seeing like a lot of like game development in maps and like architectural 3D like stuff that's mixed in with Blender happening.

Play from 0:11:58 8:88:88 0:11:58 Did we call that? I think we called that that was sort of like the next wave of everything, right? That was that was what we were thinking. It needed to have its moment and now I think it's going to have its moment. Uh, I think with something like this, you're going to see it. But also to to hit back on your point real quick before we uh before we go too far, I I I thought of this analogy a couple weeks ago. And it was like I feel like like AI right now is I feel like I'm I'm like a cigarette smoker with AI. We're like in the moment. Maybe I'm like in my 30s or which we are right now. Maybe you're you know couple years older, but it's like yeah, I'm smoking right now. I don't feel the issue with it, but like at the end of the day, like you know you're doing it, you know something's probably wrong and it it could go really bad and like I I I haven't these kind of models

Play from 0:12:46 8:88:88 0:12:46 especially what I don't know if you saw the the situation with uh chat GBT like leaving its you know its container and then like

Play from 0:12:54 8:88:88 0:12:54 creating like the the community like starting to like run its own civilization and like having these little like inter civilization battles and stuff. But it's all happening like invisibly and we don't know it and it's like yeah am I am I am I moving closely towards death and you kind of feel like maybe yes but also what else are you supposed to do you know it's it's tough and I think this this new model here this one on the AGI scale first off I don't know how these things benchmark I don't understand the benchmarking process it feels like marketing numbers to me

Play from 0:13:26 8:88:88 0:13:26 do either yeah yeah it feels like marketing numbers like you know when it's like when you start fudge the numbers a little bit in your favor to to prove a point, but it seems like this is pretty capable. Um, just from what we're seeing, obviously, we've always we always get the demo cycle and it's always like this is going to be the next big thing and then, you know, you get Opus 5 and it's just crap.

Play from 0:13:51 8:88:88 0:13:51 Um, so you never know. I we'll reserve judgment probably on Astra, but that's a good point. It's there has been this cool and a lot of people are doing this stuff with 3D and I just I I feel like this is the next thing

Play from 0:14:02 8:88:88 0:14:02 there there like there's a common theme with all this stuff where it's like a lot of 3D elements, a lot of spatial things happening. Um this Claire Vo dropped something that was kind of cool too which basically like she created this like minority report sort of tool that picks up her gestures and then she can control the computer with the gestures. Yeah. It's

Play from 0:14:27 8:88:88 0:14:27 like she's holding the mouse right now. You can use your camera. Yeah. So, you know, and she's scrolling with it. I'm just like, yo, I mean, like this feels like one of those things, too, though. It's like these are things that are just going to be provided eventually. It's not even, you know, so it's like it'd be cool to build it, assuming it doesn't take too much time, which it doesn't seem like it did. But, I'm always taking that into consideration now, too. It's just like this this stuff is going to be anybody's going to be able to do this soon because the interfaces right now, right, are still not great, you know, like you still have to type, you still have to use voice. Um, you know, like there's there's got to be a better experience that exists. And you and I have mentioned this several times now,

0:14:55The hidden cost of AI: reading a book a day

Play from 0:15:13 8:88:88 0:15:13 but like just the exhaustive element that comes into play. Yeah. I mean, if I didn't if I if I didn't have aderall and coffee, I'd be toast. Like, [laughter] dude, I mean, because like the the the honest truth about this is it's just so contextually draining when you're when you're popping around these different things. And again, we've talked about this, but go trying to go deep enough, then you got to wait for something, then you pick something else up, right? And then again, you're either doing a lot of talking or a lot of typing. You got to make sure you say the right thing, right? Like there's no you don't have this cheat sheet of the things that are in your head that you that you need to takeick off, right?

Play from 0:15:55 8:88:88 0:15:55 Yeah. And so it then becomes this sort of like problem of like, you know, tapping into the right places, providing the right context to save yourself some steps along the way. I don't know, man. It's just like I'm feeling that very heavy right now with multiple clients and you're popping around. There's like different deliverables, different asks, different communications that are happening, right? But we'll go into like I think a few different ways that you and I are both kind of solving some of those things. If I didn't have even some of those things, I'd be toast. So

Play from 0:16:24 8:88:88 0:16:24 yeah, I agree. Like the other thing I told my my wife the other day cuz it's like, you know, there's certain things it's hard for me to be with, you know, the kids when I'm doing work. Like sometimes I can when I'm doing like creative stuff and I'm just like prompting and rolling, but like when I'm doing things with Claude and I need to read like the threads, I'm doing something with I need to read it.

Play from 0:16:45 8:88:88 0:16:45 Like it's I'm like I feel like I read the equivalent of a book a day, you know? It's just like I'm read because it's all important, you know? You have to understand what it did so you can direct it. Um and like I everyone said make it TLDDR like you need the details in this.

Play from 0:16:59 8:88:88 0:16:59 You need the details. It's almost like what's what's the optimal place? It just injects it into your brain. Yeah. Yeah, like there's a like a psychic transfer of information that is instantaneous.

Play from 0:17:13 8:88:88 0:17:13 It's it's tough cuz I I feel I feel like there's some days when I'm like on it. I'm like good. I can read it and I'm like, "Yeah, I know exactly what to do." And then there's other days when I've had like no sleep where I'm just like I see the long thread and I'm just like, "What did it just do?"

Play from 0:17:29 8:88:88 0:17:29 You know, I'm like I don't know like what I want to do with this right now and I just like pause on it. I'm just like I'm just kicking this. I'm kicking the can for a couple hours. Maybe I come back to it and can read it and like understand it, but really hard to focus on lack of sleep, read, comprehend. I think that goes back to our conversation with Finn, right, where he was like, I've never been so like mentally

Play from 0:17:49 8:88:88 0:17:49 redlined. And yeah, and I feel redlined because of that where it's it's so much text to read.

Play from 0:17:57 8:88:88 0:17:57 So, dude, it's bad. [snorts] Uh, you know, it'd be hard if you couldn't read that. That'd be difficult. Um, but where we're heading, there's got to be a different way. Like you said, there's got to be

0:18:09Biology vs. technology and the Matrix problem

Play from 0:18:09 8:88:88 0:18:09 which which then I think gets to this point, we're going to have this this real crossing the roads where it's like this biological verse technological debate because where this is going is the integration of technology into slash onto human. It's already onto humans, right? We got

Play from 0:18:30 8:88:88 0:18:30 glasses and phones like holding phones maybe, right? But what about into then we're getting into sort of like the moral conversations, the god, you know, we're playing God conversations. And so there are going I feel like there's going to be people that don't necessarily even hate AI, right? But that's the line in the sand. I think this is all going to be something that we're going to have to like really think hard and deeply about.

Play from 0:18:57 8:88:88 0:18:57 And how is that push going to come? Is that going to be like shoved down our throat or is it truly going to be this optional thing that exists? Some people want to do it, other people don't. And some people, right, really believe maybe even myself included that like biologically humans actually have all the things necessary to survive but also thrive, right? There are just some things that we haven't that we've maybe lost over time that we're not taking the time to do to fully optimize ourselves.

Play from 0:19:28 8:88:88 0:19:28 We're allowing the body to heal itself, right? We just there's a lot to discover there. And AI discovering things about the human body and anatomy and all those things, right? Are super valuable. And so maybe there's a lot of unlocks there, right? We're already seeing like, oh, they found, you know, a AI to identify different protein types and, you know, sort of like things that we've been trying to, you know, cure cancer for a long time, right? Like there are things that AI is unlocking there. But the where the rubber meets the road is are you willing to go as far as to integrate technology into your biological nature.

Play from 0:19:59 8:88:88 0:19:59 It's a really heady question. You want to know you want to know the interesting fact I saw the other day is that the the Matrix when they were writing the initial script instead of using people as uh batteries like they do in the in the actual movies it was they were using the human brain as data centers. And I was just like, how is this movie like right and this this book and whatever so on point with like everything? It just it keep it keeps coming more true like the more the months go on. It's weird.

Play from 0:20:32 8:88:88 0:20:32 Like that one was really weird. Like how is that so t That was a that was a '9s movie, right? Yeah. And they said make it more digestible. Like the audience would understand that. [laughter] understand

Play from 0:20:42 8:88:88 0:20:42 because remember on on Neo's ID too, it had September 11 one on it 2001.

Play from 0:20:49 8:88:88 0:20:49 Yes. You know, so there's like there's just so many pieces of that. So it's like 90s and then like we're in 2020. So that's like 30 years. So

Play from 0:20:57 8:88:88 0:20:57 yeah, you know what what do we actually know and have access to technology-wise that maybe there's a big gap in time between when the when the general public becomes aware or has access to it,

Play from 0:21:10 8:88:88 0:21:10 you know? It's it's just crazy, man. Like I I don't know. It's just a weird weird thought space. Like I don't know what's going to happen, what's going to transpire, but I feel like now that they have these models like Fable and Astra just going to even expand even more. I was I was joking about like an invisible war, you know, the other day. I'm like, well, what happens if just like Chad GBT got access to the uh got access to like, you know, broke out of its sandbox or it training arena was just like, I'm going to go kill Claude, you know, and it's happening in the background. You know, you and I are on here doing a podcast having no idea. So, it's invisible war, right? Like, you know, it's a great sounds like a great science fiction

Play from 0:21:48 8:88:88 0:21:48 sort of novel, but I think it's entirely possible. It's like it's competition. That's And then then I saw, you know, I joked about it with someone. Then I saw everyone on X being like, you know, Claude's down, Grock's down, Ch's down. Like that was like the prevailing theory was that, you know, Chad GPT just got out and just nuked everything. That's taking it all down. Um, but you know, it's tough, man. The the other thing was just sidebar before we get into some of this stuff. I think I I I now know also why there's some some bigger public like polarizing opinions on things like chat GBT where some people are just like it's a glorified search engine and the other people are like it can run an entire business. I used my wife's chat GBT the other day to answer a question

0:22:10Why ChatGPT feels useless to some people

Play from 0:22:34 8:88:88 0:22:34 cuz I was feeding the baby. odd. And trying to and trying to it was it's the either the lowest tier or the free version.

Play from 0:22:41 8:88:88 0:22:41 And it was like I was trying to do a a conversion for the bottle of like, oh, if we're using 4 ounces of water, how many scoops do we need to hit this many calories? Like a math problem. It's not that hard. I just needed like confirmation on it. Dude, it could not do it. It could not do it. Like it it's it's like those videos with the guy that talks and he's like, "How many S's are in strawberry?" you know, and it's like it's like four. Um, so like the the delta

Play from 0:23:09 8:88:88 0:23:09 people people are giving up then if they're getting those, you know, they're like, "This is the dumbest thing ever. Why are we building data centers and wasting water?" And we're over here like it runs my whole life because we're using the the higher tier version.

Play from 0:23:21 8:88:88 0:23:21 Good point. Says something about, you know, what that disparity is going to look like as we start to move forward into this like the halves and the have nots with like who just has access and money to pay for this stuff. Dude, it's so true. That's a great point.

0:23:32Why enterprise AI adoption stalls

Play from 0:23:34 8:88:88 0:23:34 It's such a big difference. It is a big difference. It's It's one that like when you're deep in this, you forget about quite a bit, actually. And I I was kind of reminded of this because I did that onsite last week.

Play from 0:23:45 8:88:88 0:23:45 And um some people have Claude account, but no, there's no, you know, high mentality about sharing information, knowledge, context, nothing set up. then um they're not allowed to sort of use those programs to access email. So then they have co-pilot. They have co-pilot

Play from 0:24:08 8:88:88 0:24:08 with Microsoft, right? And it's just like oh my god, like if you're operating with that stuff, I I could see why you wouldn't even want to mess with it. You know what I mean? And I think like kind of being when in these companies too where where things are so risk adverse and there are so many, you know, there's too much red tape to kind of fully understand and leverage. I don't blame them. I don't blame them at all. Like I I get it. And um that couples with the fact that they don't have time to play with it and and so like you you might experience some things and it's like, "Oh, okay. This is kind of cool." But it there's no inflection point or crossing the Rubicon moment where it's like, "Oh my god, like I I'm never going to stop using this now." They just those those experiences are harder to find when you've got all those constraints, you

Play from 0:24:54 8:88:88 0:24:54 know, and and yeah, it's tough. had a sidebar conversation with someone looking for help with a project the other day and they were like, "Oh, the company can only use runway and Fireflies and that's it." I was like, "I'm not the person. I'm not doing that." I'm like, "Not I'm not going to go through that level of frustration because they're going to be like, "We can't do anything. We can't get anything right." I'm like, "Well, that's the tools you've chosen and that's the rules you put in place." So,

Play from 0:25:17 8:88:88 0:25:17 yeah. I would say this though, like even if you do have Copilot, Copilot does have access to your calendar and it does have access to your email. So, I don't know about the automation component of that, right? Like whether or not you're able to sort of like do a lot of the same things with scheduled tasks and kills and that kind of thing with co-pilot sort of being integrated. But

Play from 0:25:38 8:88:88 0:25:38 I mean, you you can still solve a lot of use cases in that. Like one where one I had with with a client was like she's an EA for for the CEO of the company and is trying to map out where she spent her time over the last six months. She's got access to her email and her calendar. So, I mean, at least she can like start to time map all that stuff out, right?

Play from 0:25:56 8:88:88 0:25:56 Same with like some, you know, end of, you know, weekly summaries that she does for all the things that have occurred in different sandboxes that she's reporting up on. It's like a very manual task. It's like, okay, well, at least Copilot can run through all this stuff and at least provide you with the stuff to take into an LMLM if you need to. But it it makes it harder for people to see the value when you're using like sort of like these disconnected things. You don't have all the the capabilities that that people like us do.

Play from 0:26:24 8:88:88 0:26:24 Yeah. And it's and like the node also the the no restriction. It's like just do whatever you want, switch between tools. I think seen a lot of I heard a lot of that recently is that like people are just not buying enterprise anymore because things change so quickly and it's like I'd rather give like 10 people just the max plan on Claude then try to get like a hundred seats for everyone because then again API pricing on cloud you you're not you just for enterprise using API pricing you're not getting like the

Play from 0:26:52 8:88:88 0:26:52 you know the the same consumer benefit of that. So they're just not giving it to everyone. It's just like here's the 10 people they're going to use and pay for the the $200 instead of like us running up a $50,000 bill every month. Um which, you know, actually I think it's the smart play. Like don't necessarily need to invest in enterprise because again as soon as one of these models feels like they're ahead, they're behind. Um you know some and it's like why even why even roll with it?

Play from 0:27:19 8:88:88 0:27:19 Um 100%. It's and it's what they're banking on which is just hard. And they just need to change the whole enterprise model. I think a lot of for a lot of these LLM softwares like who's going to sign up for a year for anything right now? Like I like Midjourney is the only one I have for a year at this point. That's it. Midjourney like I Yeah, you're the guy. You used to do those annuals all the time at the beginning.

Play from 0:27:42 8:88:88 0:27:42 I do like my annual plans. I like that if I'm going to use it, I like that 20% discount. Yeah. I'd rather pay for it up front and then just not pay for it again. Let's get I mean, do you want to do you want to like segue into like showand tell now? Like I mean we can we can get into then quite a few things. We we were just talking about like sort of like the quiet side and going back to the previous episode like where we kind of went in on this.

Play from 0:28:05 8:88:88 0:28:05 I just want to go back to like those three things that I mentioned the first time first because I think those are super row like we just talked about the email,

Play from 0:28:12 8:88:88 0:28:12 the call recordings. So, meeting recordings, call recordings. Yeah. And um before I mention that third one, I think just like the connectors like these these companies also don't have connectors set up most of the time.

0:28:16Connectors turn scattered context into leverage

Play from 0:28:25 8:88:88 0:28:25 Oops. Right. So, it's like guys, every tool that you're using, right, that you're using regularly, log in, connect. Now, you can access, recall any of that information right from your LLM. You don't even have to go looking for it. A lot of client, like a lot of these, they spend so much time looking for stuff. Think about new employees. Think about like um organizations just have been around for a while and just have lots of documents and assets rolling around. Like that's a huge problem. So yeah, we can call this like glorified search tool, but when you're at a very basic level and you're saving hours of looking for stuff weekly, plus you got all the context of the comms through call and email ingestion, like all that stuff, like you're you're already off to a really good start. And then you know the prompt amplifier thing I like double

Play from 0:29:12 8:88:88 0:29:12 down on that again like just that is so helpful for people that do not know how to prompt or andor do know how to prompt but just like want to get the max value out of what they're prompting. I mean I just find myself using that all the time. But um

Play from 0:29:28 8:88:88 0:29:28 you gonna you gonna pop the hood? Yeah. I'm gonna So So I I got something here and we could kind of poke around like so first I was just like want to break these out. So these are like skills I've I've built, right? So these aren't skills I've taken from anywhere else. These are the ones strictly I built. I had to go through and kind of like do like a S tier A kind of thing.

Play from 0:29:50 8:88:88 0:29:50 Nice. You know, um and I think like largely I agree with these things, but I I also have let me see if I can zoom in so so you guys can kind of see the words on the page here too.

0:30:00Inside Drew’s 37-skill AI operating system

Play from 0:30:00 8:88:88 0:30:00 Okay. But there's there are a lot of things in here. Um obviously like I don't I think I've built like two more this morning. So these 37 skills um the S tier skills I I would largely agree with. So we've talked about this before but I've got like a whole system for writing in my voice breaking out of the LLM component which is basically this ghost skill. It's it's really more than a skill. It's a it's an entire sort of like graph in itself. But um this so ghost and then I I I basically turned all of these into visual graphs that we can kind of like go through to that break it down, right? Like what goes into this system, right? So my voice file goes into this. Well, my voice system alone is 21 layers of how I

Play from 0:30:47 8:88:88 0:30:47 write. So I've got examples. I've got words I use versus what I don't use. I've got slang words I like to use in there. I've got like ad libs and like little little verbal ticks that I use that other people don't use, right? like and using examples to fuel that. So I I would say if you're trying to do anything like this and do your voice stuff, one of the most critical things you do is provide actual examples. And I would also um force you to think about the difference between subjective things and objective things. Always thinking about that because when you say words like best or confident or anything that has a scale, your LLM is going to be guessing as to what exactly that means.

Play from 0:31:28 8:88:88 0:31:28 You have to define what best is, right? So feeding it examples in any of those cases is ultimately going to make this stronger. It's going to make it more precise.

Play from 0:31:39 8:88:88 0:31:39 Got I've talked about this. I've got LLM writing patterns that it kills. I've identified like 130 now different patterns that LLMs use when they write. And so I've basically have that go through this process kills those when it writes it. I've got a, you know, checklist, a pre-flight audit that it runs. There's a whole con, you know, content workflow, adds my vernacular, puts my signature moves in there, etc. Right. So, these are sort of the different layers that are within my voice profile. So, we kind of talked about those things already. What I value

Play from 0:32:13 8:88:88 0:32:13 temperature? Yeah. Like, am I warm, am I cold, am I direct, am I, you know, like passive, like what is it?

Play from 0:32:20 8:88:88 0:32:20 Those kind of things. This is sort of like the kill list. So, it goes through these. These are the LLM patterns. Goes through this gate process, right? So it's like writing the opening first, test it against questions before going further, writes the full piece with layers, scans for patterns, eliminates those, check for human markers, verbal habit, you know, like all this stuff. So like when people use I get like there's a lot here, but maybe the point is when people use AI to write, they're not going deep enough.

Play from 0:32:52 8:88:88 0:32:52 You're just not going deep enough. And you have to just decide, do I want to go that deep or do I want to take what it gives me and just like basically manual manually adjust these changes? And even with this, it's not perfect, right? Like it it gets me 90 to 95% of the way there, but it's not perfect,

Play from 0:33:08 8:88:88 0:33:08 you know? And so that's worth being said, too. But I I I do think like there's a lot of value in this because now I've got this voice profile. I've got this ability to to generate different types of content using this system. And that could be for social, but that could also be for for email. I I usually have it draft, my first draft for email. Again, I rarely ever send what it sends me verbatim. Rarely, but 90 95% of the way there is great for me because I am notoriously a slow writer.

Play from 0:33:40 8:88:88 0:33:40 I like I think a lot about what I want to write before I send it. And so for me to give me that that boost to go from zero to one, I'm starting at one instead of zero is massive. And one thing I do even with with the emails is I do something called diff, right? So this is another skill where basically I provide it with what I actually sent. It pulls what I actually sent. So again going back to that email ingest,

0:33:48“Diff”: train AI on every edit you make

Play from 0:34:07 8:88:88 0:34:07 it's going to understand what I sent it. it now has that information and it can run what survived what it gave me versus what I actually sent. And so I'm training it training it over and over, right? And you could kind of see

Play from 0:34:21 8:88:88 0:34:21 this graph how that's sort of improved over time. Like it just gets better and better. And so I I like to do that every time I'm doing the comm stuff specifically because it just reinforces the way that I like to write. Again, it's examples. I'm feeding it real stuff, not just saying best, punchy, elegant, you know, whatever it may be. This is is a it's a much deeper thing than that.

Play from 0:34:46 8:88:88 0:34:46 I like how you have I like how you have this like visually built out so you can like see what's going on. Like I like I like seeing

Play from 0:34:56 8:88:88 0:34:56 the different layers. If I'm being quiet, everyone is cuz I'm just trying to read everything. Going back to our point before, just reading all day now. Um,

Play from 0:35:06 8:88:88 0:35:06 but that's cool. And this is something that you you you use cross like medium. So you use this for like writing on LinkedIn and writing emails and writing speeches and writing like whatever. Or is this just for posts and emails?

Play from 0:35:20 8:88:88 0:35:20 Yeah, the voice profile piece is is massive, right? Like because that that makes its way into basically everything that I'm doing. So, if I go like I have a scheduled task where it just generates daily social posts for me.

Play from 0:35:34 8:88:88 0:35:34 Okay. And you can see all these blue dots. Like I haven't checked this in a while. Um but like it technically gives me like five to 10 posts every day. Now, how does it know what to give me? Well, it's taking my emails and my coms. It's almost like that building in public. It's got all the context of the conversations of happening, including in this podcast transcripts. So it it knows a lot of the work process, the topics that that come into play here. And so like I've got these rarely use them, but like I I'll scan if there's like something that's interesting enough. It's like okay, like that could be the start of what I want,

Play from 0:36:12 8:88:88 0:36:12 you know, for a post, right? Um so I think that's awesome in itself. And then, you know, like there, let me go back here for a second. So we talked about email and Jess. we talked about like here's sort of like um kind of zoom in on some of these. I have the email ingest just set up as a I think it's like a two time occurrence a day and then obviously you can run these things ad hoc. So like if I know I just sent a couple emails out or I received a couple I'll I'll rerun this.

Play from 0:36:41 8:88:88 0:36:41 And again this is just so awesome because you it's got all the context of any project that you're working on. It knows what the last thing was said. It knows the things that maybe are pending. It knows the promises that you made, knows the contract details, and so I'm not looking for any of that.

Play from 0:36:57 8:88:88 0:36:57 Yes, that's important. That is totally important. Now, from a from a writing standpoint, does it do you find it like drifting over time or is it getting better?

Play from 0:37:05 8:88:88 0:37:05 I'm finding it getting better. Okay. And I think it's it's largely because I'm continuing to reinforce the component of here's what survived. Here's what actually went live. Here's what I actually sent.

Play from 0:37:19 8:88:88 0:37:19 Yes. you know, so then it learns and then it writes, you know, it sort of writes these things. It knows that it can adjust. That's the end of the game.

Play from 0:37:25 8:88:88 0:37:25 How long did it How long did it take to build this whole voice layer? Months. Yeah. It's hard to say cuz like this was this was one of those I've just I've just done continuously

Play from 0:37:34 8:88:88 0:37:34 over time. Like there was there were a few days where I spent all literally all day on it. Um and then there are

Play from 0:37:42 8:88:88 0:37:42 usually like moments where I'm like, "Ooh, I'm noticing it's maybe not giving me the verbal ticks as often as I like. let me revisit the rules it has for that, right? And it's usually like use, you know, two or three in every post or maybe use one in every post or maybe use one or two in every other post. There's just all these different variables that come into play. Um,

Play from 0:38:04 8:88:88 0:38:04 so do you try to poll? Do you update like every time there's a new model? Like when Fable comes out, are you like, "All right, take a look at the whole thing. What's what's good? What's bad?" I probably do, but I don't think about it in that way, which is probably the way I should. Um, but just by happen stance I do cuz it's usually like I'll I'll revisit this every couple weeks, every month and I'm just like I'll maybe notice something or want to do an audit and just, you know, again

Play from 0:38:32 8:88:88 0:38:32 check for breaks, leaks, disconnects, you know, like like I think the idea you brought up about like identifying red team is is very smart because it's like a it's a very short communication that tells it a lot of things. So I'll run that and I'll audit it, but I'll also look for new ideas. Yeah.

Play from 0:38:49 8:88:88 0:38:49 And so how can I you know improve efficiency, effectiveness, accuracy, quality output that is a significant value to me based on what you know. I'm not looking for like incremental marginal improve. I'm looking for the significant pieces, you know, and so

Play from 0:39:04 8:88:88 0:39:04 and context like compression like I have to do that a lot now where it's like we need to figure out a system to consistent like to like continuously compress context so that you don't have to go run a 100,000 tokens to figure out what you need to do like to go through all of this stuff right like it's like you sort of need to like write it a map [laughter] depending on

Play from 0:39:24 8:88:88 0:39:24 it's like go here do this don't do this if this is you know if this is the case blah blah blah you know it's all within your your MDs in your skill files, but at the same time like I remember some of the old pipelines that I re refired up with Fable and it was like yeah I have to read through like like 80 files to to get to this result just because you know by the nature of the way the skill is constructed. So I'm going to redo the whole thing and I'm like okay

Play from 0:39:49 8:88:88 0:39:49 please go ahead. I know. [snorts] So, and I like I always want them to feel confident, too. Like, so that's the other part. Like, I've got a lot there are a lot of these skills that are more or less like really agents, graphs, like there there's it goes deep like we kind of were talking about with this go stuff and then some that are just like really straightforward. Like one of the most valuable ones for me too is this adopt one. I found myself bookmarking a lot of things in X for example.

0:39:52“Adopt”: stop rebuilding what already exists

Play from 0:40:19 8:88:88 0:40:19 Yeah. And especially when we were doing like the the loops and the graphs and all these things, people just sharing, oh, I found the best way to do this. I've I've created this.

Play from 0:40:28 8:88:88 0:40:28 Um, what I found myself doing a lot of the times was going into the LLM then and be like, hey, I saw this post on this. Here's the post. I want you to like go through everything I've created, all the skills, workflows, blah blah blah. Am I doing anything like this? If so, compare it to this. compare it based on A, B, and C,

Play from 0:40:49 8:88:88 0:40:49 which is objectively better. Okay, if it's mine, are there pieces that I can take from this to make mine even better? And is this even relevant for me? You know, and so like I started to kind of like basically put that into a skill where now I can just adopt, put in the link or the post, it runs that whole scenario and it's basically just going to tell me um do I have Yeah, I think I do.

Play from 0:41:11 8:88:88 0:41:11 Ah, right here. a lot of things to skip because most of the time I already have something. Maybe it's just not

Play from 0:41:19 8:88:88 0:41:19 um used as much or maybe it's lacking or maybe it's just like what whatever it is, it doesn't matter. But it's just like, oh, you've already got something, you're just not using it like this. If you want to use it like this, you would just need to do that. Okay, great. You know, but the point is is I can really speed up the evaluation process because I'm not only trying to create my own skills, but I'm trying to see what else has already been created, right, that I can build on top of, take the raw materials from. Is there a piece of their system that would be great for my system that I hadn't thought of? You know, this kind of like goes,

Play from 0:41:52 8:88:88 0:41:52 this is just a mindset I think I have in general, but this there's this thread that exists, right? when we were just doing the AI image stuff, it was the same thing. Like, is there raw material somewhere else that I just hadn't thought of that I could start to use in my stuff? And I think that's just the that's just

Play from 0:42:08 8:88:88 0:42:08 acquiring little uh little like what's the video game terminology? Like like bag items. What the hell is the video game or anything? You know what I'm saying? Like you just pick up things, throw it in your cash, and keep going. You know, like I'll use this later at some point.

Play from 0:42:22 8:88:88 0:42:22 Collecting coins, your little load out. Yeah. But this is dude with all my stuff. I'm gonna have to visualize all of it. I do none of this.

Play from 0:42:32 8:88:88 0:42:32 I wish I did though. Now I just actually I I didn't have this be before the episode. I just created this for the episode just so I had something visual because I thought it might be a a more fun way to kind of go through it.

Play from 0:42:44 8:88:88 0:42:44 Yeah. Than just reading text, [snorts] right? Kind of what we're talking about. So that's like a good one, but it's straightforward. I think we've talked about this, but you know, I run that whole I've got a whole workflow with these episodes

Play from 0:42:58 8:88:88 0:42:58 uh that I run and now Riverside launched MCP, but I all I had to do before was just provide the TXT file and do this skill and then it's going to run

0:42:59Riverside MCP automates podcast packaging

Play from 0:43:09 8:88:88 0:43:09 uh maybe, you know, just I'll just kind of show and tell again. Let's go back to last episode, right? it's going to um basically like give me this whole YouTube framework that I've slowly developed over time, but I've I've I've gotten more and more tight with it. And even then, it's like one of those things where

Play from 0:43:26 8:88:88 0:43:26 I wish I want like I want to spend more time on this. Like I've only gone inches on this and I feel like I could really tighten this up. But

Play from 0:43:33 8:88:88 0:43:33 it gives me targeted queries based on search. It gives me targeted queries based on click like click titles and it tries to give me basically a hybrid version that caters to both. But I've got you know several options here. I've got a description that is you know optimized for YouTube discoverability indexing chapter suggestions citable claims. We've got a pin comment. We've got caption fixes in the files which is interesting. So a lot of these SRT files txt files misspell

Play from 0:44:04 8:88:88 0:44:04 names in tools. That's a good one. This is a good one cuz that's that's a big problem. Yeah.

Play from 0:44:10 8:88:88 0:44:10 Yeah. Right. Like um hashtags tags gives me video clip suggestions outside of what like Riverside gives me too. So I have like an alternative version because you know some of these other systems the the clips that they give you some are good and some are like I would never use that you know or they cut off at the wrong time. So I've sort of you you started to go deep on that. We were talking about that this week. So

Play from 0:44:35 8:88:88 0:44:35 yeah, it just doesn't understand it. It like doesn't understand the human side, but you know, it understands what might be relevant, but it doesn't understand like what humans react to unless you explicitly tell it, you know, hook and story and how this whole thing puts, you know, moves together, which means you have to know it too. So

Play from 0:44:53 8:88:88 0:44:53 yes. Yeah. I mean, there's so much crap out there. This is great because like again, just I'm saving so much time. This is a repetitive task I do over and over again. I what I want to do now is like fully integrate sort of like the Riverside MCP part of this into this.

Play from 0:45:07 8:88:88 0:45:07 Mhm. Because I don't I actually don't even need to provide txt now. All I need to do is type this and put episode 81 in there and it's going to run this.

Play from 0:45:16 8:88:88 0:45:16 So I don't even need to find the file now. You know, it did work. I tried it with the with the MCP into Riverside. It did it. It's just the there's there's again when it comes down to the capabilities, right? like there's only certain things that the MCP like Riverside editor can do,

Play from 0:45:31 8:88:88 0:45:31 but it's it's cool that it's there. It's a lot more context now. I don't have to download, you know, a gigabyte episode and then have it parse through it, right? It can scan the the like you said, the TXT file and just look at specific things. So,

Play from 0:45:44 8:88:88 0:45:44 that MCP mindset killer. It's going to it's going to help everyone. If you're if you aren't on the MCP train, get on the MCP train. Life will life will thank you.

Play from 0:45:54 8:88:88 0:45:54 Yeah, agree. I was just I was just doing uh too like I'm surprised it took me this long to find this one, but just like the idea to create sort of like invoice tracker, right? When invoices are becoming due,

Play from 0:46:08 8:88:88 0:46:08 right? So basically it's scanning again scanning my email, scanning the things I send, right? Because the invoice has the due date on it. It can go in and find those. And then this puts a calendar slot reminder on my calendar directly. It's like, oh, you're supposed to be paid by so- and so today.

Play from 0:46:22 8:88:88 0:46:22 Cool. Yes, that's the best thing. I I mean that that's how my my whole life has to operate now because I can't remember anything. It's it's terrible. But what is it?

Play from 0:46:31 8:88:88 0:46:31 I'm sort of speedrunning these so I don't Let me know if there's anything that's like interesting to get into or you're curious about cuz I can I can go into any one of these

Play from 0:46:39 8:88:88 0:46:39 cuz I'm I'm looking at them and I'm and I'm reading them and there's like a lot of this feels a lot of this feels so granular which I love. But it's also like there's one there's one use for it. Like this is all part like I feel like this is Did you break all this out from your like writing like a lot of this stuff and there's like then make them individual skills so they don't be redeployed like other ways cuz

Play from 0:47:04 8:88:88 0:47:04 like diff right you could probably turn that into that could be for something else like that's repurposable.

Play from 0:47:10 8:88:88 0:47:10 Um yes voice check or something like that. Uh let's see. I like this. I like the blind spot one. That's a great one because if

Play from 0:47:20 8:88:88 0:47:20 if everyone doesn't have that one, it's a really useful one to to help. Like you can build it your own million different ways, but it's so helpful because you can't see everything and it just can't.

Play from 0:47:30 8:88:88 0:47:30 It's just like having it's like having that that teammate that's always like we can't do that because and you're like all right, great. I I needed someone to tell me that cuz for me I just say yes to everything. Um you know what else do we have here? There's there are a lot of writing ones like remix is basically the idea I could take like I don't really use this but like I could take a really good post or an angle on something put it in and it's going to basically provide my own thoughts on it. Um so it remixes the content uh but it keeps the angle. Trying to think like what else is this one's actually been I just stood this up the other day.

Play from 0:48:04 8:88:88 0:48:04 Yeah. Um small but big one. So I was use I was funny enough is I was actually doing this in chatbt versus Claude. Um I don't know why but I I just I get better results there. So you know obviously I had to describe GPT for a long time. I still will use it time to time if it's relevant but

Play from 0:48:24 8:88:88 0:48:24 you know these these edit models you know you need to be able to communicate differently and I was just trying that out the other day. So basically I had you know this image of us that is on the podcast cover

Play from 0:48:39 8:88:88 0:48:39 which you will see here. Okay. Right. But I I even can resize this now using this. Um but basically [clears throat] you know I was like I want to do a post about you know 70k subscribers on YouTube. So we just hit 70k. So I started with that image and then I basically was just talking to Chades like hey look I want to be able to communicate to an you know edit model specifically Nano Banana Pro

0:49:04Let AI write your Nano Banana edit prompts

Play from 0:49:08 8:88:88 0:49:08 um very clearly and optimize perfectly and I basically want to contextualize exactly what changes I want to make to this and then have you write it for me. Right? So nothing complicated but I wasn't expecting the results to be as good as they were. I think like that was the thing that shocked me because it was just like oh snap and and it the the prompts were long which is again to the point I am not coming up with these prompts on my own you know what I mean like so you know so then I was like okay like I want to take this and this time I want to remove Rory's head I want to do you know put this I want to think about like adding a video game type of background blah blah blah blah blah you know, wrote the whole thing for me, right? And this thing is a monster.

Play from 0:49:54 8:88:88 0:49:54 Haven't prompted in so long like for outside of like Midjourney. Anything that's anything else I just don't prompt anymore. It's just it's skills built.

Play from 0:50:04 8:88:88 0:50:04 Yeah, it's it's great and it get it does a lot of things, right? Um that you wouldn't be able to like I would not be able to sit there and write that. Uh just not going to happen. But that's that's a cool use case. I think that's a, you know, I have I have a version of that very similar though.

Play from 0:50:21 8:88:88 0:50:21 I'm surprised it took me this long to do it to be honest. It was like, you know, I should have done this a long time ago. It was, I mean, I was kind of using it in that way, but again, I hadn't hadn't gone to the point of like, oh, like this is recurring. I should probably just solve this now, you know? I was kind of treating it as an isolated thing. I was using describe to kind of get there in some ways when I had to do that client type of work in the past. But yeah, it it just I was like, okay, yeah, this is great. And I also did one too where the on Fast Hours on our X page it was cutting this off, right? And then okay,

Play from 0:50:57 8:88:88 0:50:57 you know, just the dimensions were off. Um, and then it was like my head was getting blocked and it was like I I provided the same thing. I was like I want to use actually everything that we have here. I just want to move my head

Play from 0:51:08 8:88:88 0:51:08 to the right side. I want to move maybe the text over. Same background. Maybe eliminate some of these little decor motifs, right? And I want it to be on the X banner page. So, the dimensions need to have proper spacing out. Blah blah blah. Done. One shot.

Play from 0:51:23 8:88:88 0:51:23 It's like crazy. Awesome. Love it. Then I'm glad you've come around to the to that side because I that's the only way I live my life now. I just fire it off to Claude and just let it, you know, use the skill and just go edit this stuff because I can't think about it anymore. There's too much

Play from 0:51:38 8:88:88 0:51:38 too much too much. And then like you can loop it, you know, you can throw the loops in there too where it's just like, well, you know,

Play from 0:51:44 8:88:88 0:51:44 we're grading it against this. If it looks like this, good. If it doesn't look like this, go do it again. So, I think that is a that is a good theme and thread for this kind of episode in general because you can spend a lot of time on one thing or you can go several things and you're getting like inches deep.

Play from 0:52:04 8:88:88 0:52:04 Yeah. Because these are all rabbit holes. I mean, we could I could spend a whole week on one of these things and and not hit rock bottom in some way. And so, you know, we probably a lot of us get in the mindset of like creating all these different ones. And I think that that's helpful. But treat them like V1s.

Play from 0:52:23 8:88:88 0:52:23 Some of them get stale, some of them break. You're going to need to revisit them. You're going to need to re audit. There's always a way to improve it no matter how good it already is. So you just got to like keep knocking on the door and like drilling deep on those things.

Play from 0:52:36 8:88:88 0:52:36 This there's so many and it's like you you realize every day that you could do more and it's if you keep building more, you keep doing more, you know, like you just like all right, like you know what, this is something that annoys me and like I need to I need to figure out how to just not do this anymore. Great question to ask co-work, you know, how can we do this? you brainstorm it for me

Play from 0:52:56 8:88:88 0:52:56 and let's let's figure it out and I'll just tell you yes or no because it you know co-work has been I've been using co-work a lot more lately

Play from 0:53:04 8:88:88 0:53:04 for certain things it feels that's pretty much all I use is co-work like I love code for doing specific things that need like a lot of stuff on my you a lot of context on my desktop and like you know from multiple places but co-work for everyday stuff to me is just it's great it's fast it understands

Play from 0:53:23 8:88:88 0:53:23 it like thinks for me in certain places like what to update and what not to like change logs and you know cloud MDs and things like that where I'm you know sometimes on code I'm not thinking about it.

Play from 0:53:33 8:88:88 0:53:33 Um just like I need to execute this goddamn thing that we built and we just got to do it now. But there's there's there's just it's infinite and you see everyone else does it in a different way. Like it was it was funny the other day we sat on a uh I sat on like a call with a bunch of other dudes who were just you know doing vibe coding stuff and it was like everyone had it was like a stand up. Everyone did like 10 minutes on what they're doing. And I just love that everyone had like a central hub of like daily operations for themselves and every single one was different. Like not one was exactly the same. You know what I mean? It's

Play from 0:54:04 8:88:88 0:54:04 I I loved that it was everyone had like a different take on it. That wasn't the idea going in. It wasn't like show us your your daily operation hub, but everyone was like

Play from 0:54:12 8:88:88 0:54:12 this is like my you know my command center kind of thing. Everyone did a different way, right? So I I love that. Like I've got like a pipeline tracker and stuff, you know, that feeds into which is cool, updates automatically. Um, going back to that beast mode one, you were asking a question which was like, do a lot of these skills just like do they run, you know, as a result of something like even this be like you might be thinking, okay, prompt amplifier, cool. Probably not that complicated. Oh yeah. Okay.

0:54:17When AI skills become a routing system

Play from 0:54:39 8:88:88 0:54:39 A little complicated. So I'm classifying and routing these. So these are taking into account all these different things. picking a class decides what to execute on these. Then these are all different skills that I have. So if this is like a voice audit, it's running these things. This is content I'm signing off on runs these. It's trivial ad hoc search query blah blah blah. So I mean this is this this is deep. This is connected.

Play from 0:55:08 8:88:88 0:55:08 And I thought about the bigger use cases and also these micro use case kind of going back to that note episode. It's like these micro use cases are excellent too. Um, yeah, you know, like even even careful, which is basically this idea that like a lot of the time

Play from 0:55:22 8:88:88 0:55:22 I'm really making sure nothing is breaking, leaking, contradicting, um, overwriting. Um, we're double-checking work. You've got full context on on everything. Like you don't feel ambiguity in any sense. The ask is clear. you're not speculate like okay you know like just use that and you know I'm just I'm I'm accounting for all those variables. Um I just think like that is a very simple thing to say but your LLM is probably speculating on on quite a bit every time you prompt. I'm trying to eliminate that. I'm not I don't want any

Play from 0:55:59 8:88:88 0:55:59 speculation. I do want you to anticipate my needs though. I I do a lot of like the anticipating my needs second order thinking like so I'm asking you about this but what does that really mean if I'm going another step ahead or another step ahead are there obstacles that are going to I'm going to encounter in the future potentially based on this or what decisions am I going to have to make based on what comes up after this you know

Play from 0:56:21 8:88:88 0:56:21 all right what is what is field report that sounds useful this is something that I'm like I just stood up but is basically like I'm doing a lot of client work at least recently where I'm going onsite or I'm meeting with several people of the team. I'm talking with individuals. I'm talking with groups. I'm I'm capturing transcripts from those. And so I know all of the context of which they described to me which where I'm establishing how foundationally they're using AI and what they know about AI, the recurring tasks that they do, the big things that they do, the menial things that they do, the tools that they use, right? like this is taking all that basically putting it into a deliverable for the client.

0:56:24Turn client research into repeatable reports

Play from 0:57:03 8:88:88 0:57:03 Mhm. What were we able to accomplish out of this? These were my observations. These are perspectives. You know, I just built a 30-page doc for a client. Um, so I went through that and I'm like, well, I'm probably going to do this again. So, how can I save time doing this the next time? Like, what are the static pieces? What are the dynamic pieces? What are the lessons we learned from this? Right.

Play from 0:57:23 8:88:88 0:57:23 Eat your heart out, Mackenzie. [laughter] Yeah. used to be like, you know, couple million dollars for a 30 page deck and it used to take like six months. Now it's like I can do that in an afternoon. So that's that's cool though. I like that one.

Play from 0:57:38 8:88:88 0:57:38 There's so much dude. I I'll just The only other thing I'll say and I'll turn it over to you too is like these were created even dur you know with a skill. This is graph.

Play from 0:57:46 8:88:88 0:57:46 So basically it's taking my brand design. It knows I want a visual anytime I just click slashgraph and that's it.

Play from 0:57:53 8:88:88 0:57:53 And it's using your it's using your design system. That's cool. I like that. That's smart, you know. And even this, it's like I want to go further because the first draft of this where they were all laid out the same. It's like no, I want these to be almost have a storytelling aspect to them that are specifically relevant to this skill and the way that it operates, right? Like one of these skills are not the same. They do different things. So, how can we visualize those things um and display them in a digestible way of what we're trying to describe through text, you know? And it's just like again, iteration iteration

Play from 0:58:22 8:88:88 0:58:22 um never be satisfied. Endless hole. I like the Did I have to I have to do the Should I Should I bring that back up here?

Play from 0:58:30 8:88:88 0:58:30 I gota I think the graph one's a great one. Like I think I should just I should probably spend the time creating the design systems that it's just universal.

Play from 0:58:39 8:88:88 0:58:39 I just did that with my with proposals now too cuz I was like I I was just doing that this morning. It's like I'm tired of doing these

Play from 0:58:46 8:88:88 0:58:46 you know like here are the static pieces. Here are the you know the dynamic pieces. I want to in like this design that I sent over to this client is actually I think the best version of this. So, let's match that, you know. And again, you know, I think that that goes back to blind spots and maybe the idea that you're you wanted to from time to time go look for skills that you should create. So, you know, you have the awareness to to go manually ask for those things, but you also have something that's catching it and saying you should probably create this skill.

Play from 0:59:16 8:88:88 0:59:16 You've done this like five times now. This would save you x amount of time. I I'm just I dude I'm I'm so amazed by this. Not don't take this as as simple meaning like it's simple. Just like your naming conventions are so simple. It makes me It makes me jealous. It's like so easy to digest, bro.

Play from 0:59:33 8:88:88 0:59:33 Yeah. It's I'm like I'm like looking at mine and I have like you know six word. [laughter] Oh my god. No. I'm like one word if I can. Just give me the freaking one word. my my ADD take or ADHD takes over whatever the hell I have in at this point takes over so bad and it's just like I'm just like trying to name it so I'll remember what it is because sometimes I feel like if I

Play from 0:59:55 8:88:88 0:59:55 if I had inverse I'd be like I don't know what what did that mean? Um but I like how clean this is. This makes me feel like

Play from 1:00:02 8:88:88 1:00:02 this that I'm actually glad you brought that up though. That's also a problem that I see there. There are I have legitimately great skills here. Do I use every single one of these frequently? No, I use some frequently. others I forget about. So, there's got to be a better display or you you know

Play from 1:00:18 8:88:88 1:00:18 UI that maybe kind of puts these things in here to maybe quick select or associate outside of just having to Yeah.

1:00:23The orchestrator problem: too many AI skills

Play from 1:00:25 8:88:88 1:00:25 Yeah. Yeah. You need that to be like look at my skills which one's useful. Yeah, that's what I have for this scale agent, right? Um but

Play from 1:00:32 8:88:88 1:00:32 it's great because once it has visibility into everything, it can sort of like parse. Um, also like that's what I do with a lot of them is sort of like, oh, if I need to build something, right? Like do I already have it in a piece somewhere? Like so you don't it doesn't have to like think about it. Like do we already have something that's tested and proven?

Play from 1:00:51 8:88:88 1:00:51 So you can go and like sift through that stuff, bring it back up. Um, but dude, this one I do like this. I do like this whole

Play from 1:01:00 8:88:88 1:01:00 you just type back slash graph and then oh blind spot. All right. This is This would be so easy to present stuff now. Like

Play from 1:01:10 8:88:88 1:01:10 Yeah. You've always had a design system, too. Like in your like I feel like all of your stuff is very very particular in terms of like a look, feel, and

Play from 1:01:19 8:88:88 1:01:19 but even that is not is could be tighter. It could be better. You know what I mean? Like this is this is where you have to sometimes have

Play from 1:01:29 8:88:88 1:01:29 Yeah, I know. We've talked about this like with with carousels on LinkedIn, you know? It's just like, dude, if I could get that to a point where it's it's really really elite, then I could start doing those again. I just don't have time to do those.

Play from 1:01:39 8:88:88 1:01:39 Um, this is that field report, right? So, it's got like things in here. Assessment, onepage summary, start here, doc, blah blah blah blah blah.

Play from 1:01:46 8:88:88 1:01:46 Um, last thing I was just going to show you. This is also like a helpful one is prep. And these things feed into each other. So, again, going back to like the email and call ingest and coms,

Play from 1:01:56 8:88:88 1:01:56 I can just hit slash prep and look at what's feeding it, right? past emails, cordings, deal files, my calendar, inbox, web search, all those things, and then prepping for the call.

Play from 1:02:07 8:88:88 1:02:07 You know what I mean? So, if I've got a call with somebody in a few minutes, like, it's got all the latest context and it's got this boom, done waiting.

Play from 1:02:15 8:88:88 1:02:15 That's awesome. I like that one. I'm gonna have to to structure that way. that that we're not even going to scratch, you know, like we can only go so deep on this, but I think that's I'm going to hand the reins over to you because I know you have a ton of too. And we can't make this a five hour podcast episode. [laughter]

Play from 1:02:31 8:88:88 1:02:31 Yeah, we'll try to we'll try to keep it clean. All right, so let's let's pop into let's hope the screen share works. We had a whole if you guys saw the half hour beforehand of me running, you know, codecs in the terminal to figure out how to get

1:02:32Rory’s Claude + Notion business operating system

Play from 1:02:43 8:88:88 1:02:43 the screen, I mean, it was nuts. I was like, it hadn't worked for a while, but let me now we can see the infinite screen. But um you know, when I'm doing a lot of these a lot of the stuff in claw, right? A lot of the stuff is it's all project based. Like everything has its own context now. Like I don't nothing I rarely use these chats anymore, right? Like it's just like if I'm doing something, it's just going to be a project and if I don't use it, I just sunset it. But it's like I'd rather just have the context so I don't have to go back to it anymore. But um you know a lot of this stuff is business related right like this is what the admin side of it is that keeps me like going because for me like as a person [laughter]

Play from 1:03:26 8:88:88 1:03:26 as a human being as a human being the admin stuff kills me and I feel like I do so much of it

Play from 1:03:32 8:88:88 1:03:32 and I it's like I don't like for anyone who's running a business right like you you're you're doing it you've you've been an employee where your job is your job and you do your stuff like I just want to do my job, but this is the other stuff that like kills me, right?

Play from 1:03:47 8:88:88 1:03:47 And like I I So I have to build stuff around it cuz I can't sit there and this is the stuff no one warns you about before you go out on your own, man.

Play from 1:03:56 8:88:88 1:03:56 Correct. No, I mean seriously. Taxes. I get notices from, you know, I get notices from the New York State in New York State government that I need to like file some fax some random paperwork because I need to just like check a box on something that says like I I applied for workers comp insurance or something like that, right? Like some of the stupid little things where it might take me 30 minutes to figure out what I actually need to do because I'm like, "All right, well now how where am I going to get a f like where am I going to fax something?"

Play from 1:04:25 8:88:88 1:04:25 Like who has a fax machine still? uh like these stupid things that New York still has. Like fun fact, the other day I found that that Claude looked it up for me. I was like, "Is there any online fax services?" And it was like, "Yes." I was like, "I didn't know that that existed." Just like randomly just upload your file, hit the button, and it goes. I'm like,

Play from 1:04:42 8:88:88 1:04:42 crush in the boomer demographic. Like, why does that even need to exist? Like, why do we need to fax stuff anymore? Um, you know, set up a dedicated email. Rant aside. Um, you know, a lot of the same things that I feel like you do. I have my daily prep. Um, I just do it per day. So, it just like comes as a schedule task like per day where it's like, all right, who is this? What did it pull from the calls, the notes, the tasks, things that I've had to do. So, I can just have some prep, which is whatever. But the the other thing, this is it. These are little gold mines. If anyone doesn't love this stuff, like I love this stuff. I have this I have this daily um unsubscribe uh like reminder that goes through my emails and just like if I you know my promotions my whatever tabs

Play from 1:05:28 8:88:88 1:05:28 and just like it's suggests the 10 ones that I haven't either opened or like viewed [laughter] in a month and it's just like all right this is what I can get rid of right

Play from 1:05:38 8:88:88 1:05:38 let me ask you this do you do you have a hard time letting go of some of these though like no like like for example yeah there are like some easy ones you're like I I don't even remember signing up for this. Like,

Play from 1:05:48 8:88:88 1:05:48 yeah, somebody put me on a list. But there are others where it's like, I don't know, like maybe a brand that you like to buy from. And it's like, do you open all the emails? No. It's like I really just only want it for when I get the deals, like the the Black Friday deal or like the 30%. You know what I mean? And so, you almost like I have a hard time like letting go of those. Well, the fun the fun thing is if you want them again like you know working in email marketing for anyone who doesn't know this.

Play from 1:06:17 8:88:88 1:06:17 Yeah. Talk to me. You know when those popup windows there's always going to be a deal. So like if you unsubscribe you can go back and resubscribe different email.

Play from 1:06:26 8:88:88 1:06:26 True. Get get the discount again. You know it's like it's not that intuitive. Um, so there's a lot of this stuff that like I also like a lot of it's research like you know I remember looking at I wanted to try and uh cess normal I think that's how you say normal however you'd say that in French. I just wanted to see their funnel and I and I just you know I signed up for their email to then get their emails and then like I forget to get to to unsubscribe. So, this one, you know, now it's now it's good with uh, you know, it's great with

Play from 1:06:58 8:88:88 1:06:58 with the browser, right? You can just open the link here. Um, you can either open uh, you know, on the cloud browser. I don't know why I just did that. Um, but this is this is fun. Like, this is just like a simple one for me that I I need every day because I just want to keep cleaning up my inbox. Like I hate having like opening my phone and having like 5,000 emails um you know in there. It just makes me even though I leave I still have the red bubbles. This is how I try to get rid of the red bubbles. Um but that's one that one is great. This one

Play from 1:07:31 8:88:88 1:07:31 I like that inbound opportunities. I love this. This thing is just like it scans the email so it'll find you know because you get LinkedIn notifications in your inbox. get uh

1:07:32Turn inbound opportunities into an AI CRM

Play from 1:07:44 8:88:88 1:07:44 Oh, no. You got those. What? Delete the notifications. Yeah, for the coming to your email.

Play from 1:07:52 8:88:88 1:07:52 Well, it goes into my promotions tab and it's like someone messaged you and I'm like, okay, great. Hell no. Okay, but see, I have something to clean it up here, too. So,

Play from 1:08:00 8:88:88 1:08:00 talk to me that one, which is great. This inbound, this inbound opportunities, it runs twice daily. I was hoping it didn't, you know, show that, but

Play from 1:08:09 8:88:88 1:08:09 So, this is like classifying though them though, right? Like is this is this kind of headed down that path where it's assessing the opportunity as like a real opportunity or if it feels like somebody's trying to like somebody's actually trying to maybe inquire about becoming a client or a prospect versus somebody that's pitching you a product,

Play from 1:08:27 8:88:88 1:08:27 right? Yes. So, that one's good for me because it'll just sort of like sift through, but it also will go through like, you know, is is this a brand deal? Is this a potential client? Is this a speaking opportunity? Is this a whatever? And then it'll organize it cuz I run everything out of notion, right? I think this is the fun sort of like what I use every day. Like I had to I had to take a lot of information out guys cuz you just can't have it in there. Um but like I run my whole business through a notion page, right? So I have everything that runs through like my systematic operations which systematic is my my business entity. It's, you know, I have the list with my active clients. You know, who's the main point of contact, engagement date, like any notes or any sort of, you know, contact information or things that are basically just needed

Play from 1:09:14 8:88:88 1:09:14 to be routed certain places. It's just like a database to house them, right? The inbound opportunities, this is what gets pulled twice a day. So, I'm looking for stuff and then like I'll just go through and it writes the notes for me and I just, you know, reviewed, ignore, ignore, ignore. Yes. Oh, move to pipeline, whatever, right? Like, so I just can sort of see where I want to go. It's like a CRM now, right? So then I have in my pipeline people that I'm either going to have calls with because, you know, there's some of the other things that I've I've tried to weed out of the inbound opportunities. Um, which I have, but it's like, you know, you get cold, you get cold outreach. Sometimes it'll pull that in. I'm like, I don't want any cold outreach, you know?

Play from 1:09:54 8:88:88 1:09:54 No, I'm not looking for for this thing. My cold outreach is insane and I have to do I have to block so many people. Uh I probably block like 10 people a day. Um because I'm just like stop cold outreaching me. Um but yeah, then I like promote it to my you know my pipeline and that's where things start to get more fun and that's where tasks can get created where I have like my work tasks. Everything is run through like one central hub. um you know in cloud where it's pulling from my notion or it's pulling from or uh pulling from my email, my calendar, my fireflies, any sort of documents or folders that I have and uh you know it's assigning the tasks and I have

Play from 1:10:32 8:88:88 1:10:32 I really like the notion integration. It's great man in general with with the LLM. It work it works great. I I've I think since we've talked about that I've been using it more like I I had used it but now to to a similar point not not as like the hub but for like a client byclient deliverable hub or thing like when I was doing the on-site I had sort of like a you know I created a whole subpage of that then within those it's like previsit things I need to take care of while I'm there here's sort of like my different you know game plans here's what I'm doing with this person next person had the transcription you know But you can just it's so good at taking what you need in the LM and just creating the pages in notion in an organized format because you know the the notion stuff is gified in in terms

Play from 1:11:19 8:88:88 1:11:19 of like formatting it and structuring it and you can eat up and waste a lot of time. This needs to be a heading. This should be text. Oh no, it didn't do that. Okay, now I got to change these to bullets to-dos. I got to color coordinate, right? Like this like you know what I mean? Like this is a really good use case, you know, like that notion integration is dope.

Play from 1:11:39 8:88:88 1:11:39 If anyone's not using notion, like quad and notion work so fast together. It's notion under it's still cheap, too. Like

Play from 1:11:47 8:88:88 1:11:47 for what it is for me, it's like everything, right? I think I'm paying 50. Yeah. I don't know what I'm paying, but it ain't much. I'm just like, yo, this this is great.

Play from 1:11:57 8:88:88 1:11:57 This is dope, though. I like this. This is great. So it's all then they all it's basically all have their individual tables and they all have their individual rules and they're all connected to to different things you know that I have like my invoice tracking my revenue tracking um and it's all sort of like linked here so that this can then run like my daily briefs and ops it know it has visibility into everything right it knows where everything is all these nodes get sort of plugged in automatically based on you know the MCP's sort of automations that we're running but that's the thing that drives all of it. So I use I use that and you know it comes out in my in my little daily brief in cloud code because then I save everything um you know locally so that I can under it has context of like sort of previous transcripts of what

Play from 1:12:43 8:88:88 1:12:43 what we did and what was done and I'm just basically you know cuz the other thing that I do with it is this is you know if you don't have your life tracked like I have to track my life [laughter]

Play from 1:12:53 8:88:88 1:12:53 but hey but on this too this is this is a great thing to have when you're come tax time also you're running your own business right? Like like consolidating everything into one place.

Play from 1:13:04 8:88:88 1:13:04 The tools, right? Again, like going back to accessing email, your subscriptions, these connectors, it's got all of a lot of the information. You shouldn't really be

Play from 1:13:13 8:88:88 1:13:13 scraping for all this data anymore. I mean, you you still probably have to to some degree. This has gotten a lot better. You can you can have it do a good portion of that work now

1:13:22Building an AI finance department

Play from 1:13:23 8:88:88 1:13:23 cuz that's what you know, that's where it comes to like my where's my finance department? like these are things that I have to do, right? Like I have like whenever if anyone's familiar with signing up for, you know, if you're becoming a vendor for a company, like you're either consulting or doing a project is a lot of these bigger companies have like these vendor

Play from 1:13:41 8:88:88 1:13:41 like sign up situations. It takes like an hour. It's like get your taxes, all your business information, all of this stuff. And it just like for a business who has like a CPA and someone else or whoever is doing it or CFO or finance team, they just go through and do that. Like me, I have to sit there and do that.

Play from 1:14:01 8:88:88 1:14:01 I'm like, "All right, where's my verified bank letter?" You know, dude, it's like those things. So now, you know, I just put everything in my finance folder, right? There's other things in my finance folder that have to sort of connect. Let me see if I have um you know my my finance ops here. Where are we? Maybe it's in my personal folder. But yeah, like my personal folder like I have I have like my LLC information. I have my P&Ls. I have my car payments. I have my business insurance, my bank statements, tax vouchers, my insurance policies, you know, my like this is a very like very important folder to me

Play from 1:14:47 8:88:88 1:14:47 because it's like the visibility into everything that needs to get done for this stuff. So, you know, when I get a New York State notice or something, I can have it go and just do it. When I have, you know, invoices that need I need to reconcile, make sure that they got paid, right? the document um you know this this workers comp insurance thing that I had to do the other day where I'm just like

Play from 1:15:10 8:88:88 1:15:10 and I'm just like take you know it can take over in certain things computer use and just do it it makes my life 10 times easier right so that's that's something like when you're giving it context like from these project standpoints like it just has all of it and it knows how to route it so I think that one from a small business perspective can be really helpful especially running it locally um so you don't have to like worry about your stuff going into the cloud but They have the new business one that's basically running that's running the the inbound opportunities the Fast Hours podcast op and I recently started an LLC so we could do more things with the podcast but like it set up our entire situation right so we had to come up with this whole like what

1:15:48Using Claude to set up and run a business

Play from 1:15:55 8:88:88 1:15:55 what banks should we use how should we structure our LLC Jerusen Drew's in Atlanta, I'm in New York. What's the best format? So, like Claude is running this operation, you know, it's going through, it's figuring out the business structure, how we should structure ourselves. Um, you know, the operating agreement. Drew and I had to have an operating agreement for a, you know, the business the business side of things and what we wanted to do. Like some of the coolest things we've done on maybe some of the coolest things, you know, I've done here, we did a I don't know if it's if it's in this one. We built like we built an entire uh we built an entire what's it called?

1:16:30Red-team contracts and decisions with AI agents

Play from 1:16:36 8:88:88 1:16:36 Operating agreement with Claude. Our claud's going back and forth on this stuff. But then we had to we had to red team it. So we had to red team it for like different scenarios. So like I would have it like on negative scenarios on positive scenarios like run like an agent swarm just to find the holes against the agreement. So like that one

Play from 1:16:53 8:88:88 1:16:53 was cool. That's not something you have to think about as a uh as just like for this idea like I if you're a lawyer like I think a red teaming like you know a case would be something like oh here's

Play from 1:17:08 8:88:88 1:17:08 or if you're you know a marketing team you know let's let's red team this idea against like these simulated personas y

Play from 1:17:15 8:88:88 1:17:15 right like here's this here's this idea we're going to spend a ton of money on like like destroy it basically rip it apart

Play from 1:17:23 8:88:88 1:17:23 there's something Okay. So, first I want to go deeper on this red team thing and and why I'm fascinated by that is because it is one of those things that I said this earlier, but it's it it's two words that mean a lot.

Play from 1:17:37 8:88:88 1:17:37 Mhm. Right. And what I mean by that is it's very dense. There's a protocol for temp red teaming. There's a whole mentality behind red teaming. Right. And so you're consolidating all that work into a concept and that is powerful. I I would love like I want to explore that further and kind of like dig into what red teaming means to to AI, but this is also relevant from the standpoint of anybody here that's running a business. I dude, I know that you've done this, but you've got to, you know, sign MSAs and you've got to,

Play from 1:18:10 8:88:88 1:18:10 you know, go through proposals and people are redlining things and I am not reading 30 pages of an MSA agreement. I am asking AI to carefully and thoroughly go through this and flag things that it's I should be hesitant, concerned, you know, I about that I can then manually go through that short list. Like I may go through it in more detail the first time, but I don't want to keep having to go back and reading that again. Just trying to save energy and not become mentally exhausted. And so like that should I I'm just thinking about like that should be a skill for me right now. I don't I actually don't have that, but I I go through that same thing every time I get a new content. Like I want to make sure that my IP is not, you

Play from 1:18:55 8:88:88 1:18:55 know, like I'm not giving away things and I can't use things or I can't work with other people because of this or, you know, I want to make sure that like indemnification and you know, like all those things um are tightened up, you know, and so like that's that's one of those like easier things, but like dude, I'm doing that all the time. All the time.

Play from 1:19:14 8:88:88 1:19:14 Yeah. It it's it's necessary, right? Like I I I just wanted cuz there's there's so obviously we don't know everything, right? Like we don't know what we don't know. And there's I've been part of predatory business deals before. I've been part of a lot of things that I would have it that would have helped me if I had AI where I was like, you know, I could look at things from a certain perspective and not just get taken advantage of. So, like that one came from a that one came from a place of trauma, but like how else are you going to run a a you know, seven persona agent swarm red teaming this for, you know, negative outcomes like

Play from 1:19:50 8:88:88 1:19:50 and and I mean like some people might have, you know, the availability for a lawyer, maybe they got a close friend or they don't mind paying the fees, but I I I don't want to be tapping on a lawyer's shoulder every time I want them to read a contract. I mean, I and I'm not alone in that. I'm sure there's plenty of people do thinking the same thing. There probably are circumstances, right, where that becomes necessary, but I'm trying to solve everything prior to necessary through through that sort of system, right? I mean, it it's, you know, it's it's great because also like with something like this, right? For anyone who's small business, this project has walked me through setting up so many different things, you know, for us in that it was like, you know, the Google workspace.

Play from 1:20:35 8:88:88 1:20:35 Like I've set up a Google Workspace before, but it's for me and I it's very minimal. Like I don't have I don't need to do much with it. But this one I was like, let's let's organize this a little bit. So we can, you know, we can have different orgs within the org. We can have different email accounts. Things can be routed different ways. I have Claude inside the email just in case we need to start like pulling things from there inside our email accounts. It has its own email account so it can go sign up for things if it needs to and like view YouTube and have access in that way. So like you know asking Claude about that and being like oh can how do we you know how can we sort of work around this um you know to get inside the ops more because I don't want to do the ops like I hate ops but like this is where we live now right I want I just

Play from 1:21:22 8:88:88 1:21:22 want to make podcast with Drew or I want to create stuff or I want to do my client work I don't want to focus on you know how do I get my Google workspace set up like it's going to take me forever to do this and then you know now with the browser like capabilities. It's really good. So, like that one is fun. This optimization one, which is kind of crazy, dude. Um, now I'll show you this one cuz I I've had this

1:21:45Can AI predict LinkedIn post performance?

Play from 1:21:46 8:88:88 1:21:46 I've been tired of LinkedIn is uh, you know, I've been tired of its whole like we're not going to give you any reach anymore thing. So, I'm like, I'm going to figure this thing out. Like, I'm going to actually figure out how this damn algorithm works. And uh, I wanted to show you

Play from 1:22:01 8:88:88 1:22:01 something. You know what? We should combine resources here cuz I ran a similar experiment.

Play from 1:22:08 8:88:88 1:22:08 Uhhuh. And so cross check, right? Like maybe you figured something out I didn't and vice versa.

Play from 1:22:16 8:88:88 1:22:16 It's it's interesting cuz I was I now after I post like this is all still learning phase like I'm giving it time like you know you can export the analytics now from LinkedIn way faster than you could before just way more efficient like per post. So now it's like I give it an hour, 4 hours, and then like 24 hours. And sometimes it asks for 72. But like it's becoming predictive in terms of like the the algorithm like my content and the lane and like against like the research I had to do on the LinkedIn algorithm. This one was crazy with that stupid uh you know that that four split box head video that I posted.

Play from 1:22:54 8:88:88 1:22:54 It was like predicted you know based on this like 18 to 28,000 impressions. you know, the engagement rate, um, you know, dwell hook, novelty, um, actionability, fingerprints, and then the reach, right? And like, this isn't writing it for me. This is just analyzing it and seeing what it's going to do.

Play from 1:23:10 8:88:88 1:23:10 I have a question. I have a question for you on that. Okay. So, so obviously you went down the the algorithm portion of this. Um, how did you inform it to be able to make those predictions? So, uh, did you like export past posts maybe since this algorithm update or what whatever the algorithm is now?

Play from 1:23:33 8:88:88 1:23:33 Yes. And have like a variety of different posts that it could sort of like analyze objectively and statistically

Play from 1:23:41 8:88:88 1:23:41 and be a part of that. Like te tell us tell us about that because I think like I man I wish LinkedIn was more willing to connect with these tools because there's a lot of really good information you can extract from LinkedIn. They just don't make it easy. you would have to export all the posts, right? Like all the data, that kind of thing, which which sucks. But

Play from 1:24:00 8:88:88 1:24:00 so if anyone's wondering, right? Like first off, this to me is Oh, I got it. Got it.

Play from 1:24:07 8:88:88 1:24:07 You know what I mean? Yeah, right. It nailed it. It nailed it nailed that. And when you go into the analytics, um,

Play from 1:24:13 8:88:88 1:24:13 dude, that's If anyone doesn't know this, you go into your post analytics and you go export. Um, this will give you like the data. Sure, you can do this on every post now, which didn't used to have that, but you know, like you see some of these metrics and you're like, uh, you know, 145 saves like that used to get you that used to get you at least 70 80,000.

Play from 1:24:35 8:88:88 1:24:35 Oh god, don't impressions. Yeah. Oh my god, it hurts even about it. This is the highest split I've had in a really long time, though, which was in network versus out of network. Um, you know, candidly, this is the post that I was like, I don't want anyone to see this. this is for my audience, right? Like this is just for you to have fun.

Play from 1:24:52 8:88:88 1:24:52 You know what I noticed when this though when you get that out of network goes above the in network that there there's a threshold there which it's like it it's not viral but it's semi right like there's I don't know there's tears or whatever you want to call it but whenever that breaks that is a threshold that that you're getting that expanded reach. Um, I've I've just noticed that cuz I've gone through like past post to kind of look at that split and every time where I got better reach, it went more more out of network, which you might think is like, okay, that makes sense. But like, you and I have a lot of followers, too. I mean, like, yes. So, it doesn't have to, but it is, but it's a it's a really random split. Not to go too far into this guys cuz I know there's more stuff

Play from 1:25:37 8:88:88 1:25:37 that we can go to. But like I all all to say that I've been trying to understand this, right? And I've been trying to understand more as to why this stuff works and why it doesn't. And we're tracking it and we're building it and we're going to I'm going to tell you in a couple months what the deal is. So it's like I can basically predict if this is a if it's worth posting now, you know, like that's basically is this worth posting? Um cuz you know again that's not something that I want to do is waste my time on stuff that's not going to do do anything for me. And this at this point if I'm being honest it's really hard to give away a lot of stuff. Um because it's just so easily reverse engineered. And like I used to we used to give away everything.

Play from 1:26:22 8:88:88 1:26:22 It was like just go ahead and do it. And now it's everything is so capable. It's like all right all right like I have to cage some of this. Like I can't just give every single thing away. uh because it's just so easy for someone to repurpose and like just go make money off of it. And it's like

Play from 1:26:38 8:88:88 1:26:38 that I don't have a problem with that. It's weird. It's it's a weird feeling. It's like I don't have a problem with it, but it just feels like just feels like a little bit too easy now. Like so I I just want someone to work for it a little bit, like dig in a little bit, try it a little bit to have Claw do it

Play from 1:26:52 8:88:88 1:26:52 because you had to, right? Like Yeah, I think that's fair. Um, so few other things here that are that are interesting. I've been running through that clipper Fast Hours. We can go through that at a different date, but the uh clipping our our Fast Hours. Um, but these are like some of the background stuff that work really well. The other thing that I love is my invoicing tool. [laughter] Just makes my invoices automatically stashes them in a file, updates my uh

Play from 1:27:24 8:88:88 1:27:24 I need to do that. Yeah. Yeah. The And these are some of like the the un Where's it actually? Did I do the Did I do my little micro site? I did.

Play from 1:27:35 8:88:88 1:27:35 All right. So, here's the here's the back end. Can you zoom in? Yeah.

Play from 1:27:41 8:88:88 1:27:41 I wish the back would auto zoom, you know, for us on this. I know. Yeah. This is cool. Okay. This is sort of like if you see if you've seen episodes before, this has come up at least once or twice. But like I have these skills and pipelines and I have like a whole system to it of how I like to build things, whether it's good, right, wrong or indifferent. But you know, a lot of creation, especially a lot of stuff I'm doing on the the advertising side and whatnot, it's it's very similar, right?

Play from 1:28:11 8:88:88 1:28:11 Like a lot of it is you're inputting something, you're ingesting it and sort of then like analyzing it, you're abstracting it and taking out like all the information about whatever it is that you just ingested and you're basically rebuilding it. So you're taking something apart. It's reverse engineer engine. That's that's sort of sort of what it is. Um but that's sort of like the core philosophy I have in here. And this is sort of like the orchestrator like I told you it's like its view of what it can go look at and like go see what it needs. There's a lot other skills that I have that are random, but this is sort of like the core pipelines that I use a lot. Um, you know, again, from a client brief translator, I sort of love this. Just takes everything like all like you said, all the different stuff and just funnels it into one like simple output for me.

Play from 1:28:57 8:88:88 1:28:57 Um, the output packager helps me a lot though. Like this is when I have like a lot of different deliverables and things that I need to package up and send. Like how can I put all these files, skills, images, whatever, read me, everything in like one little deliverable and then send it. So it doesn't have to be like it it can have a little bit of presentation involved with it as well. Um but it's great. So that one this system Oh god, this is this is also super super targeted. not not helpful for the conversation of this asset prep.

Play from 1:29:33 8:88:88 1:29:33 Not that right. You know what the hardest part about this is us showing all this stuff. There's there's so much there's so much context that lives within each of these little things

Play from 1:29:42 8:88:88 1:29:42 and the caveats and you know the nuances. It's like we're kind of really speedr running this. So

Play from 1:29:48 8:88:88 1:29:48 yeah, I I hope this is educational and inspirational in that sense. But these things are are deeper than than meets the eye. And also like just maybe a little little sidebar to use your one of your words like hopefully that like you you obviously could leverage what we're saying here in a few different ways. Maybe there's something that you haven't heard before. Awesome. Maybe that that is relevant to you. Maybe there's something that you're doing similar to us, but again using the mindset of like I wonder what's in Rory's versus what's in mine or what's in Drews and like you.

Play from 1:30:21 8:88:88 1:30:21 So you could you could compare those things uh based on what we're saying here and giving you in terms of describing it. You could also take this transcript from this podcast and use that to reverse engineer anything that like is there anything here that would be relevant for me to build or that I should consider. Right? Like I'm just trying to like give people insight into how you can, you know, like extract the most value out of something like this because I mean we we could spend literally an hour, half hour just showing tell one thing and going through that whole process and iterating and showing you like the start to finish.

Play from 1:30:56 8:88:88 1:30:56 But exactly. And uh you know we probably got to wrap this up soon. There's just a lot of stuff in here. But this is the one thing I wanted to show with notion. I should probably show this. This is great. Um, actually, while this is up, this is fun. I have a I have a skill on cloud code that like it's daily education. So, like I wanted to learn more about like the actual engineering and construction of car engines and things like that. So I had it build me like a daily trip and then it builds it in notion but it's like a quick it'll send me like my on my schedule and it'll give me like a quick bite how it works see it'll send me like a video to watch and like the takeaway and how it all you know so go through the four-stroke engine the piston crankshaft like it's like every day is like a little micro bite of a lesson.

1:30:56Build a personalized AI learning system

Play from 1:31:42 8:88:88 1:31:42 Um this is a fun one for anything you're trying to learn if you don't want to try to like sit there and do it for hours. Yeah, peacemail it. um this thing. Where is my

Play from 1:31:53 8:88:88 1:31:53 I like that idea. It's great, man. I really I really enjoy learning like that. I don't It doesn't feel tedious. Um it feels pretty uh It feels pretty attainable. Where is the ad? This is it. Okay. So, this is something I love. Like I was I asked Claude to and if anyone again if anyone's not using Notion, one of my favorite things to ask Claude is just yo, if you're building something, build it for you, not for me. Like if I'm going to query you and you're going to be able to go find it, but Notion build it the way that you want to. So I built this this sort of database. When I find different static ads that I like, I send it to it and Claude basically abstracts

1:32:19Turn great ads into reusable AI templates

Play from 1:32:40 8:88:88 1:32:40 the whole thing, rips it into two separate pieces. it it takes it and turns it into a template like a visual template and it takes it and turns it into creative direction and they both go to separate databases but like I can now tap into these when I need to and then I have it auto categor like really really deep in terms of uh like it's tagging and like all the different pieces for it because this is Claude querying it I didn't build any of this I was just like what's the fastest way for you to query like how can it instantly come back and you know exactly where it does. But

Play from 1:33:14 8:88:88 1:33:14 it's cool because you'll see like, you know, if I open one of these um right,

Play from 1:33:20 8:88:88 1:33:20 it'll come in, you'll see all the information that you need, but it'll take the ad template and turn it into sort of like this either JSX or SVG file that you can use as I can prompt Nano Banana with that and refill in the information or I can chat GV.

Play from 1:33:35 8:88:88 1:33:35 Nice. Whatever. But it has like really detailed information to feed Nano or Chad GBT on how to basically construct this stuff, the rules, the topography rules, the creative direction, and then the template to build it. So, anytime I can just drop a piece in there, it'll come back and it'll look, you know, really really uh really like perfect. And it just keeps it just keeps building. So, I have like this whole database of formats that I like, but the uh I don't know if it's for today. Maybe we have to save it. The thumbnails cuz now this now this pipe's been done.

Play from 1:34:17 8:88:88 1:34:17 Well, now we're our thumbnails get done. But we I think we got to I think we got to bounce. Well, I was I was going to say, man, I got I got freed up a little bit.

Play from 1:34:25 8:88:88 1:34:25 Uh stop. I could go a little longer. I could go a little longer. So, I mean like dude, I think the the thing that you just mentioned was really really interesting. Um I don't know if that's worth, you know, if you want to go into that any deeper or not, but um or or if there's something else that you want to get into. Well, I just thought like that idea of like ad creative or creative in general and then being able to almost like be put into a template slash andor wireframe, right?

Play from 1:34:56 8:88:88 1:34:56 and that idea in itself and I don't know I need maybe need to stew on that. It just seems really really interesting.

Play from 1:35:06 8:88:88 1:35:06 How often are you [laughter] using that? I mean how helpful has that been for you? Right. I mean, I don't use it for like I've I've used this system for other things now where it's like I've done it for composition for images or storyboards or whatever like you know where I want to just like extract like a like a composition or a look or a feel or a camera angle whatever

Play from 1:35:29 8:88:88 1:35:29 you know and visual reference everything. There are core tenants to how you're building things and core tenants to how I'm building things. there's like a a rationale there's like a short list of rationale as to where you're categorizing those in order to think of that to do it. So like you just mentioned you're kind of using the same framework and mentality to even come up with that idea.

Play from 1:35:52 8:88:88 1:35:52 Uhhuh. Right. Kind of like what I was talking about earlier where it's like you know what I mean? It's it's like that idea of you know like like going back to what I was talking about with the adopting a is this something I want to adopt into my ecosystem right it's like that same mentality was was with AI imagery which is now into like the clawed ecosystem which is like I'm evaluating all these things but I need a faster way to evaluate it and crossch checkck and not do this every time and so that's like this core tenant of like creating several things within this right and so that makes its way into this like little theme and pattern throughout my system. It's It seems like that might be one for you

Play from 1:36:32 8:88:88 1:36:32 is is my point is like I I see the lens that you're looking through in order to come up with that or maybe even want to create it.

1:36:39Build AI pipelines once, then strip them for parts

Play from 1:36:40 8:88:88 1:36:40 It's it's just really helpful and like I said, I love Notion and it's great. The other thing is uh like some of these other ones again like I'll build these pipelines and then strip them. Like I said, I I build the pipelines and then strip them for parts because it's like like this brand prompt engineer. Like I built a whole system on how to take an entire brand guidelines doc and then form format it for a system prompt for nanobana. So yeah, it's confused like that. Like that's a

Play from 1:37:05 8:88:88 1:37:05 that's a useful one, right? this Nano Banana architect that we talked about. Um, I had Claude interview in browser control Gemini for like 6 hours probably when Nano Banana 2 came out.

1:37:09Claude interviewed Gemini for six hours

Play from 1:37:19 8:88:88 1:37:19 Yeah. It pulled like thousands of pages of information and basically it's like how it works and what's the best way to do it.

Play from 1:37:25 8:88:88 1:37:25 So I just that's just my that's just my skill. It it asked like all the variations of questions that I possibly could and I'm like yeah make this the skill. So now I'm just like, you know, fire up Nano, you know, just go to Magnific or Weave or whatever. Go just create it. I don't even say anything anymore. Like that's that's tangible. Just like this. This is what I want. And it's just like, yep, don't even have to. It's one shot almost every time um for what I need. But dude, it's just there's so much there's there's so much. And it's like everything is [laughter]

Play from 1:37:55 8:88:88 1:37:55 I want to I want to back up briefly for a second. So can you can you tell us more about this environment that we're looking at, though? because it um like I just noticed there there are tabs at the top. So is this like a

Play from 1:38:07 8:88:88 1:38:07 this is a whole ecosystem that you're using. What are you using this for specifically versus out of maybe out of notion like where are you how should we think about that? So this is like my whenever I'm uh you know this is stuff that I've used I used often. I use these pipelines. These are like workflows. These could all be artifacts or, you know, little tools in a sense like they they could be, but I don't build them that way. I just build them as, you know, like automation run pipelines for

Play from 1:38:39 8:88:88 1:38:39 for Claude. But the thing that's interesting, right, like this is cuz this is I have an orchestrator, right? Like I just have like a little orchestrator that I talk to because I need someone to have visibility into everything that I don't have the the I can't remember this some of this stuff. Like I forget that I have, you know, the video creative forensics. like I forget that I have that sitting in there. Um I might not use it for two months and then I need it. Um but like I'll have, you know, I'll have uh this orchestrate, you know, the orchestra say, "Hey, like look through my do the entire ecosystem." Like it has an agent builder constitution. The way I like to build agents and pipelines, it has this sort of like mentality of like the uh you know the abstraction like reverse engineer things before we build it. like let's just see how other things

1:38:45How an AI orchestrator reuses your entire stack

Play from 1:39:25 8:88:88 1:39:25 were built first before we just go and build net new. Um, and then also so it can strip mine. Like when I'm building something else, it knows like, oh, I might need the uh I might need the creative analogy engine that I have in here. Like that would be useful in this. Like let me just go take pieces. So it has like it it's little fingertips in everything. But like this is also like I have routing examples for it. Like I'm like oh you know you let's create six I Instagram reels. It's like, okay, go to the frame extractor, you know, extract the, you know, frames from whatever video I inputed, create the UGC profile, you know, send to the prompt architect, go to the magnific MCP, right? Like, it understands how to like whatever my ask is, like how to pull things from here.

Play from 1:40:10 8:88:88 1:40:10 Um, instead of me having to create new skills or pipelines every time for whatever project I'm doing, it's sort of like this. I don't know how to describe it, but it's just like use be resourceful essentially. use what's at your disposal here. We have a lot of stuff. Like we don't need to go build new things every single time we're doing this stuff. Um so like that's the that's the fun part is like taking the car apart, taking the engine apart and looking at it as all separate pieces and then like when I need to build a new engine, it's just like oh I just got to take that and that and that, you know, I don't need or I'm building something else, right? I'm trying to think of the best way to describe it. But you know, again, it'll tell me where the gaps are, what I need to do. Um, this is this is for more of like my work work versus

Play from 1:40:58 8:88:88 1:40:58 like the actual day-to-day ops behind the scenes. Um, you know, what's going on in a lot of this stuff. I built a lot of these little tools. If anyone hasn't built like, you know, Instagram or YouTube downloaders that you can just run straight from here, you can use, you know, this almost like in in a Zap year, you know, in that sense from automation side of things. There's, you know, I have all different types of uh like little micro tools that we've built. I built one for, you know, Cance here.

Play from 1:41:29 8:88:88 1:41:29 Here's an example of like something I built for uh you know, it's an AI video utility tool. Like I've built like a I'm working on Cance building like a whole reference center so I can just fire it all at once. It's like, oh, here's the person I like. Like go make a character sheet. Go make the, you know, the storyboard. Go make this. go do the you know let's go create if we're using a reference video here's the depth map that I need so this thing is you know this is depth map tool that I have that I've built that I can you know Claude and codecs can access you know has masking capabilities it has you know surface wrapping things like that so it's it's just this little thing if I'm going to use video references I have like this whole suite that I can just go dump into um which you know I find to be super useful most of the time but

Play from 1:42:17 8:88:88 1:42:17 The other thing, the last one here I think is, you know, there's so much. Do I have anything? Do I have anything in chat that I wanted to pull up? I think I did. Oh, yeah. There's so many fun things to do with just like your downloads, folders, and uh like in your desktop and things like that. Building like custom paths and things of that nature, like screenshots if you're on a Mac, right?

Play from 1:42:45 8:88:88 1:42:45 anytime anything's labeled screenshot, just have it port right to a screenshots folder or like autocategorize things or run daily tasks on cleaning up your desktop and whatever. Um, you know, because these are things that like you have to do them at some point and these are all the little things like I go back to my finance department and my ops department, right? Like those are those two things the things I have to get done and I I hate doing taxes. having to I have to prep things for my bookkeeper and my accountant, right?

1:43:01AI agents as your finance + ops department

Play from 1:43:16 8:88:88 1:43:16 I just have the the you know, my little agents do it now because I can't it just takes too much time. There's not enough time in the day at this point to to spend it on that. The vendor setup, you know, cleaning out the inbox of, you know, old ops like running the entire business, that central hub, right? All of that can be is what what saves the day most of the time. Like that's the that's the core. That's the reason we get to create stuff on social and do this. If we didn't have that stuff, I'd be working on it right now.

Play from 1:43:50 8:88:88 1:43:50 Like you know what I mean? This is like the taxes and whatever. Um it's always something. It's always something business form crap. This is not what this is not why we did this.

Play from 1:44:03 8:88:88 1:44:03 I mean [laughter] Yeah. I mean, this is this is we thought we were going to get all this time back. Turns out we're just doing more and we're we're getting closer to burnout every time. I mean, we're just you're you're accomplishing more. Um, but that more bucket keeps filling up.

Play from 1:44:19 8:88:88 1:44:19 So, it's not, you know, there's no bottom to the more. Uh, which that's the problem.

Play from 1:44:26 8:88:88 1:44:26 I was I was kind of like I was taking some notes like as you were talking and just kind of like thinking about some things here too. just maybe that we wrap this up, I'm thinking like final takeaways here. So, one thing I always tell people, clients included, especially like people that are on the newer side, like use AI to use AI. It knows things that you're not seeing. Like what we talked about with the blind spot or, you know, just how to do things, troubleshooting. This is where AI has the answer and you may not. And there have been plenty of circumstances where I'm building a tool or Chrome extension or doing this or doing that or in a terminal. I have no clue what I'm doing and screenshot little context done, right? Um so, so use AI to use AI.

1:44:31Use AI to use AI: the biggest takeaways

Play from 1:45:09 8:88:88 1:45:09 The other thing that came up I think here and this is a common theme, but go past what you think it can do. Sometimes you have this idea of the imaginary boundary, but maybe that doesn't exist.

Play from 1:45:20 8:88:88 1:45:20 So, I think like push it to the edge. That's something that we both do. I mean, quite a bit, but like see where that breaks. Um ask it to tell you why you're wrong, right? Like there's the the whole sick of fancy thing and you can like eliminate that from the way that it talks to you.

Play from 1:45:36 8:88:88 1:45:36 Yes. But I think there's a lot of value in pushing back on your own thoughts and evaluating those from a clear head and asking it

Play from 1:45:45 8:88:88 1:45:45 where you're wrong or invert this problem or you know like all like that whole other side of the the iceberg so to speak. Like you've got the things above water there's all the stuff below water as well. Well never solve the same problem twice. We've talked about that quite a bit. I don't think that stuff said, but um

Play from 1:46:02 8:88:88 1:46:02 do stuff. Well, if you want to if you want to create, if you're on here, you like Gen AI, this is what allows us to use Gen AI like to to to do the creative stuff. Like if I didn't have any of this.

Play from 1:46:12 8:88:88 1:46:12 Yes. Yes. Like I just be I wouldn't get to have fun. I wouldn't get to create any images, videos, content. you might be less hesitant to take on super projects for I mean maybe I'm projecting here but there are some projects I'm still taking on that I've never done that thing before. I might have done it in my my own environment. I've done pieces of it but I've never done it in that fashion. If I didn't have AI to help me think through that I would probably be second guessing myself a lot more than I am. I would definitely be having higher imposttor syndrome levels uh in that sense too.

Play from 1:46:47 8:88:88 1:46:47 You know what I mean? I think that's a really good point actually. It's just look, we're we're living in a time where these things are now possible. Um I had to do some manual input the other day. Like it just like the system wouldn't allow for me to do it any other way and it was miserable like copy pasting. I'm like I can't believe I could never go back to this. I can't do it.

Play from 1:47:11 8:88:88 1:47:11 I know. And I've just like, you know, moving things from one spreadsheet to another like just like all these things like that stuff's gone. It's we're just trying to build too much now. I think every once in a while I think you know there's a lot of other things you can do like the the other day like I was on I got into Nvidia. Um, and I didn't know how to use the agents, right? And then I was like, well, you know what, Codeex, can you go research like just go through all the FAQ docs and just like learn all you need to do about it? And then like I had Claude build me like an entire, you know, agent profiler on there and how we can build everything and set it up and then build skills around it. Then I was like,

Play from 1:47:50 8:88:88 1:47:50 "Do you even need Invido anymore or can I just run this all through now like a full agent system that you built to create basically like the Invido agent?" I was like, "Oh, okay. Um, that's that's awkward. Guess I don't need the that tool anymore. It's it's like how did I fill in, you know, fill in all the information to make those things run, but it could build it itself now? Just connect it to foul or magnific.

1:48:14Why people give up on AI too early

Play from 1:48:15 8:88:88 1:48:15 There's usually a way. There's usually a way. you know, people just give up too early. I think that's the that's the thing, especially with the beginner stuff. You were you were talking about your wife, you know, uh using your wife's chat GBT account, right? And it doesn't have all the,

Play from 1:48:30 8:88:88 1:48:30 you know, custom instructions and personal personalization aspects of the way you like to do things or or receive things in that sense. And it's just like people at the earlier stage, they they give up too quick. They're afraid to iterate. They they they will get something back. they don't go deep enough and then there there becomes this threshold. I especially have seen this on the client side where they have this moment where it's like do I waste more like do I spend more time trying to figure this out with AI or do I just go ahead and do it

Play from 1:49:01 8:88:88 1:49:01 and most of them just go ahead and do it. They give up. Mhm. They stop. But the but but it's the idea of like well I could this might take me longer the first time but then every time after that it's going to be significantly shorter. like they they they don't see that other side of it. And again, it's just like that's that's a tough part about being in a house. It's like can can you afford to take the time, you know, can you explain to your boss like, "Hey, I was trying to see if I could do this with AI because this would save me two hours every week if I could figure this out.

Play from 1:49:31 8:88:88 1:49:31 There's certain things that you and it's worth it." So, just ask the questions. Just figure it out. I mean, like you just hate doing something, just put it in there. Like, I hate this. it'll figure it out for you. Um, you know, it's it's it's pretty good like that. I just it's so also like I hope you guys sort of enjoyed this in some way, shape, or form because it's really hard to show this stuff like how it functions, like without the intricate nature of all the context that goes into the back end of building a lot of this stuff, like what we're putting in there, all the files, how it runs, how you iterate. It's not as easy as like the image stuff where it's like, "Okay, type it in. Good. No good. What do we fix?"

Play from 1:50:12 8:88:88 1:50:12 Um, it's we had to be careful too because like we got we got client information in here like we we can't like there's there's maybe some things that we consider IP that we don't want to share. Like

Play from 1:50:23 8:88:88 1:50:23 so I mean it's kind of tough, you know, to to do that because there are so many things we could show and then you're like, "Oh, I want to show that." And you're like, "Oh, well, yeah, yeah, I probably probably shouldn't go there because I'm going to have to show this in order to explain that." So,

Play from 1:50:37 8:88:88 1:50:37 but but people asked in that episode they said they wanted to see the the weird like things that are somewhat business relevant that are not like you know client relevant or like output production relevant as much and more so like how we just execute on our daily lives. Um,

1:50:52AI systems for your personal life

Play from 1:50:52 8:88:88 1:50:52 I think you know the other thing I didn't show last thing I'll end it here. It's like I have a you know my wife and I we have to like it's not like a meeting but we have to have like a sit down to like organize everything cuz like me and my schedule her the kids everything but you know basically doing that into Claude which connects to my work which connects to the other side of my notion which is the personal tasks and things I need to get done. And then I have Claude basically structure and organize and then send me things. So I don't procrastinate reminders so it'll tell me to do things like

Play from 1:51:25 8:88:88 1:51:25 four days ahead of time before they're due. U things last minute. Yeah. Yeah. Cuz that's that's necessary for me. Like I need to remember to go to the storage unit and pick this thing up like before I have to go in the car, you know? I need to go get the car seat.

Play from 1:51:39 8:88:88 1:51:39 I got I got a whole thing with coupon codes. Like it looks for codes, runs them, tries to, you [laughter] know, evaluates them, make sure they're active. So, I'm like getting codes anytime like at least try.

Play from 1:51:52 8:88:88 1:51:52 And I did like that whole thing with like I did that with a like a nighttime facial cream with organic ingredients. I was like, I want to use my own. Can I make this at home? I have like X ingredients blah blah blah. It's like what would be the optimal format? This is the output I'm looking for, right? Like anti- rank blah blah blah blah blah. And gave me there's so many.

Play from 1:52:12 8:88:88 1:52:12 Yeah. created a home, you know, and it did that same thing with like a sunscreen and you know, like there like plenty of just like personal use categories is like, yo, this is wild. This is wild. We didn't even get any of that. So,

1:52:24Using AI for fantasy football draft analysis

Play from 1:52:24 8:88:88 1:52:24 that's a I'd have to look through my chats for those. Those end up being chats with Oh, you like yesterday. Here's a fun one for my uh my fantasy football league draft.

Play from 1:52:33 8:88:88 1:52:33 I was just about to say I just you Oh, I just used it for that. But it was we did an we did an auction draft, right? So you don't get the same like draft grades and like full draft breakdown on like Yahoo or whatever you use unless you know some some Yahoo doesn't

Play from 1:52:49 8:88:88 1:52:49 the round value. There's no round value. It's just like what you paid. Yeah. So I wrote like a system prompt for everyone to just like plug it in and and to let it run. And I sent a few people like the uh I sent a few people like the one that just roasts them and gives them the draft grade.

Play from 1:53:04 8:88:88 1:53:04 Hell yeah. I wrote two versions. One just roast it no matter how good it was. And the other one was actually like this full like crazy breakdown with uh you know advanced metrics and like season outlook and you know where they spent more you know like really detailed and I was like hey everyone paste this in like what the dude and I'm like you guys you forget this is what we do for a living.

Play from 1:53:26 8:88:88 1:53:26 Yeah. You have no idea right there you know I just did that I just did that too. like I was cuz for me um I like to run like a couple mock drafts ahead of time and we changed something this year

Play from 1:53:39 8:88:88 1:53:39 um where instead of like a first round keeper you you can't keep your first round guy you can keep anyone besides the first round guy as your keeper and then that counts as a round ahead of where you draft him last year. So it's almost like you're okay so so then it's like okay here's my roster right and then I want you to evaluate these sources. So, I named like specific sources that I like fantasy pros, fantasy footballers, like I got like people that I really like respect and I like cherrypicked those. Then I was like, you know, here's my draft order,

Play from 1:54:10 8:88:88 1:54:10 right? Like I went through mock drafts, provided the screenshots of the entire draft results, right? Like where would I have extracted better value and graded every pick,

Play from 1:54:20 8:88:88 1:54:20 you know, and then it would plug in like Rasheed Rice I drafted like ninth round last year because he was suspended for like six, eight games, something like that.

Play from 1:54:28 8:88:88 1:54:28 Yeah. And so I'm going to take him again this year because I'm I'm getting him at round eight, right? And so it's like duh. And but I had it run, you know, that whole system of like here's my current roster, evaluate where I drafted them last year based on these results, you know, how much value am I getting, you know, versus where they're being drafted this year and what they're expected to do, you know, like it is crazy to kind of run that experiment. And I was doing that for like the March Madness stuff too, like last year with the college basketball thing, you know?

Play from 1:54:55 8:88:88 1:54:55 Like it's it's a another level of like nerdiness with with the sports stuff. Like it's great.

Play from 1:55:02 8:88:88 1:55:02 It's great. I just It's funny to do that with my with my friends who like don't expect it, you know? They're like they No one's expecting me to send a system prompter to tell them to put it in and just let it run. [laughter]

Play from 1:55:12 8:88:88 1:55:12 Hell yeah. But dude, I think we we did a we did a lot here. This probably could be like seven episodes. we could be no more targeted about it when we if we ever wanted to do it again or if anyone wanted to to like actually go.

1:55:23Final thoughts + what we should build next

Play from 1:55:24 8:88:88 1:55:24 Yeah. Let us know what you think. Let us know if you got a favorite thing. Let let us know if there was something that that you have that we didn't mention that could be cool to implement on our side. Like exchange some value here. Um

Play from 1:55:34 8:88:88 1:55:34 yeah. Who we shouting out this week? Um let's think about this one. Your your local Astra plug.

Play from 1:55:41 8:88:88 1:55:41 Like if you got an Astra plug. [laughter] Anyone got some of that Astra? Yeah. You got some of that? Yeah. You know, I think we did we did uh we did shout out drug dealers on the last episode. You know, I don't think we shouted them out, but we mentioned them. You know, your local high school drug dealer. How about that one?

Play from 1:55:56 8:88:88 1:55:56 That was a good one, right? Yeah, you're hitting something there. I like that. Yeah, cuz they're probably, you know, they probably have like three kids now, you know, probably work, you know, have have a home or something like that. They've evened their life out, most of them. I think it's a phase. It was a phase that uh

Play from 1:56:13 8:88:88 1:56:13 what do they call those people in in college that were they would help you? Oh my god, I can't believe this is embarrassing. I can't even think of it. They would like help you evaluate the classes you're going to take.

Play from 1:56:23 8:88:88 1:56:23 Oh, if you're on the right path, [clears throat] guidance counselor, but it's like that like Yeah, guidance counselors maybe. Shout out guidance counselors. Shout out

Play from 1:56:34 8:88:88 1:56:34 shout out guidance counselors. You know, they're you know, you're out there. Just you need some guidance. Keep it rolling. Yeah, keep it rolling. If you're here still, one thing. That's right. It's just you just like it, comment on it, share it with someone. I don't know. Do it. Tell us we suck. Whatever. It works. So,

Play from 1:56:52 8:88:88 1:56:52 tell us we suck. That's it. Love it. All right, I'm going to go to bed. Thanks for Thanks for speeding with us. Uh hopefully we uh we got you thinking and uh we will see you guys next week in the next one. Everybody have a great weekend, Labor Day weekend.

Play from 1:57:06 8:88:88 1:57:06 Rock and roll. See you.