Episode Transcript
[00:00:05] Welcome to Prompting Curiosity, a podcast for the AI curious. No coding background required. I'm your host, Dr. Shantae Cofield, also known as the Maestro, and I created this show to explore what these AI tools actually are. Really, though, are the files in the computer, how to use them, and what they might mean for how we think, work, create, and move through life. Whether you're skeptical, intrigued, or already experimenting, you're in the right place. All that I ask is that you stay curious. All right, let's get into it.
[00:00:38] Hello, hello, hello, my curious people. Welcome to episode 36 of Prompting Curiosity. There it is. We got a name, right? We had a name last week. We got a name this week. I am your grateful host, the Maestro, and today we are talking about my AI Wish list. So before we get into the main topic, y', all, I am still hella stoked about the new site.
[00:01:01] Fun fact. At the time that I am recording this, I haven't officially done the hard launch of the new site yet, but by the time you listen to it, it will have already hard launched, and this will be a week later, which is wild. If you've listened to my other podcast, Maestro, on the mic, I say this, like, every other episode, because it is. It is just like a wild thing to me because I record it. I record episodes a week ahead, and actually, for Prompting Curiosity, I record them a week and a day ahead.
[00:01:26] Um, I record on Wednesdays. They go out the following Thursday. So it's like I'm speaking about the future, but I'm in the present, which then becomes the past. And either way, the important thing is that I am stoked about the new site, the brand refresh, all of it. Hopefully, y' all are digging the new cover art. Uh, you should be seeing it, because I.
[00:01:45] I changed everything quite a bit ago. I mean, sometimes. Sometimes things get, you know, bogged down a little bit, but you should be seeing it. Um, and I would love. This is my CTA for today, folks. My call to action.
[00:01:55] I would love it if you took a screenshot of that. Of the screen that you're looking at and shared it to Instagram. All right? I want. I want people to see the new. The new goods, see the. The new thumbnail, the new cover, I would even call it, because I'm proud.
[00:02:10] Take a screenshot, tag me the movement Maestro. And if you want, you're feeling like a bonus tag. Prompting Curiosity podcast. Yes, I have an Instagram account for that. For this podcast. Do I do anything with it? No. Um. Um. And I know asking for the double tag is annoying. As.
[00:02:25] And so if you really want, just tag me the movement maestro.
[00:02:29] That's, um, fine. But I would love it. It would really mean a lot to me. Um, and as per always, I am stoked and grateful for your support. So let's hop on into the main topic for today, which is my AI wish list. So I was talking about, uh, this with Lex the other day, but this is something that I believe is the case with so many things, and that's that growth occurs in spurts, and those spurts tend to be exponential. Like, I feel like I went through like a physical growth spurt when I was like, in first grade. And then I like, never grew again. So if you ever meet me in person, you're probably going to be like, wow, you're smaller than I thought. I'm not like a super short person. I'm 5 6, but I weigh 125 pounds on like a big day.
[00:03:13] And so most people think that I'm just like, would be bigger and like bigger and I'm not.
[00:03:17] But I was like this size in fifth grade, so I like thought I was gonna be big and then like, that. I never kept growing like, this is it. So we tend to see oftentimes those exponential growth spurts and then you're always like, little, little, little, and then growth.
[00:03:30] I believe this happens across all things.
[00:03:33] I experienced this, uh, with beach volleyball. Well, shit will stay the same for months and just like at the same level playing, it's not bad. Just at the same level. And then suddenly you unlock a new skill and you like, practice that and then your progress goes from flat horizontal to, you know, up into the right. And it's a really, really cool thing.
[00:03:52] There was a lot of that, you know, that up into the right for the early years of LLMs, large language models. Right. Claude Chatgpt, namely Chatgpt. And it's wild to say the early years because we're only, you know, three and a half years in. Right? Chad GPT 3.5, that was. Or GPT 3.5, which was chat GPT, like the original that was released in November 2022.
[00:04:16] So we are currently in March of 2026.
[00:04:20] Um, so, you know, overall it's still in the early years. But what I'm talking about is like that first one to two, even into the third year, uh, we saw that up and to the right and that exponential growth.
[00:04:31] Right? But as it stands right now, present day, March 18th is when I'm recording this. Uh, it'll be not that day. When you listen to this, I think it'll be March 26th. When you listen to this, if you listen to the listen on the date it drops. Um, but at present day things aren't being rolled out or you know, quote unquote shipped as they say in, in the tech world as quickly. And I think perhaps if you're a cloud code diehard, uh, you may give me some pushback, right? Which is fine. But for 99% of the people who use LLMs, the changes in capability are no longer exponential. We are in a time right now where it's about putting what we have, the technology that we have to work, and exploring everything that's possible.
[00:05:12] Brief side note here, we're going to nerd out for a second.
[00:05:15] Uh, as it relates to OpenAI ChatGPT, we witnessed, we literally witnessed the end of these exponential growth spurts with the release of GPT5. And I covered that, I spoke about that. That was in August of 2025. So I actually started this podcast on the day that GPT5 launched. It is coincidental, but until then, right, with each release, one GPT three GPT four, excuse me, GPT. GPT 3.5, GPT four, I believe there was a 4.5 maybe. And then five came out.
[00:05:45] And until then, right overzealous CEOs like Sam Altman, especially Sam Altman, they were doubling down on the idea that they could continue to scale LLMs by simply making them bigger, right? You make these LLMs bigger, you give them, um. I almost had more prognosticators. That is not the right word.
[00:06:04] That's not the right word at all. More parameters and that they will get, you know, quote unquote, we'll call it smarter.
[00:06:13] Until GPT5 came out, that had actually been the case. Which is contrary to what you uh, know leading like, like machine learning scientists and, and the, the tech nerds, the people that actually know about this stuff, they had said the opposite. They were like, no, you can't just make these LLMs bigger and they will get better.
[00:06:34] And OpenAI was like hold my beer, right? Um, and it was true, going from 3.5 to 4 and like, you know, we don't know what they trained it on, but they models to get bigger. We don't know the ultimate count of the parameters, but the things that get bigger, right Again, LLM, the size of an LLM is based on the number of parameters, not the size of the data set that it is trained on. So just for whatever that's worth to you, um, but my point being for, for a bit there, you could just make the models bigger. And Sam Altman was betting because he's the worst, was just like, we're gonna scale our way to AGI. Artificial General Intelligence. That's what, that's what that stands for.
[00:07:17] And it looked like it was gonna happen, although everyone that actually knew anything about anything was like, no, it's not gonna happen. And then GPT5 came out and everyone was like, weep Womp. This fucking sucks. It's not any better. You lied to us. And everybody was sad. Everybody that used ChatGPT was sad because the uh,04 model, the 4O model, people love the personality of that one. And then he got rid of it. People were like, bring it back.
[00:07:40] I, uh. All that to say, right? It feels like right now the, the growth curve has gone horizontal. Right Again. If you are a cloud code Stan, feel free to push back. I love, I'd love to talk to you if you're a cloud code stand. We're going to get into cloud code a little bit today, but I would love to talk to you if you are a cloud code stand. If you don't know what the fuck I'm talking about, don't worry about it. We're going to get into it a little bit. Um, but in my, in my opinion, the growth curve has gone horizontal and the gains to be made now are in how you are using it and what projects you're having it complete. And I was very intentional with the wording there at the end of that sentence. Projects that you're having it complete or tasks that you're having it complete? Because that is the next level of using AI, right? Uh, which is leaning into what is called agentic AI. I have talked about this in a bunch of previous episodes, but agent, as a reminder, agentic AI is AI that can actually do things. It can execute tasks. It doesn't just answer questions.
[00:08:42] So that in mind, I want to spend the rest of the episode going through my very short wish list for AI, AKA the things that I want to and that I am going to use AI more specifically called Claude Code to build. All right, so in the spirit of brevity, I'm going to refer you Back to episode 29. Um, that'll give you the deep dive into the deep but not like super nerdy dive into Claude code and Claude cowork link those in the show notes for you.
[00:09:10] The tldr, right, is that those who don't know TLDR is too long didn't rest too long, didn't read. In this case, I probably say I don't too long, didn't listen. Tldl um, but the TLDR is that Claude code is Anthropic's agentic tool and Claude cowork is Anthropic's agentic like tool. My cloud cowork is basically Claude code for non coders. Um, and fun fact, Claude code built Claude Cowork. That's fucking wild.
[00:09:40] But both of them can execute tasks. It's that cloud code has way more like, significantly, substantially more functioning personality.
[00:09:48] Both of them use Anthropic's various LLM models, which is Opus Sonnet as their quote unquote brain. I know it's confusing. It's weird. Like, wait, what's happening? Just stick with me because all I want you to understand is this kind of, uh, uh, I have a goal. I have two goals in sharing my AI wish list, right? It's twofold. Number one is going to put me on the hook to actually build these things. Like I'm planning on building them, but hopefully I build them sooner than later because I told you guys about them. Um, which means that I can then learn and then I can teach you.
[00:10:18] The second thing is that sharing this wish list with you gives you examples of how this tech can be used, which hopefully sparks some ideas for you and that can lead to more understanding, right? Because right now cloud code is just this thing and it's very backwards. I said this time and time again that AI was rolled out in a very backwards way as it relates to marketing traditionally and, uh, more successfully, oftentimes more successfully. When we market something, bring something to market first we do market research and we find a problem and then we find a solution to that problem. AI is backwards, right? LLMs, they're backwards in that these companies said, hey, here's a solution, go find a problem. And that's really where we're at right now with cloud code and cloud coverage. Like, I see this thing, I know it exists. Like, it seems cool. I hear people are using it, but like, I don't really have a problem for it. So I'm going to share two problems that I have that I am going to use cloud code to, to solve, and maybe that will get the wheel spinning for you. Then you try it and you'll start to understand more about what this thing is. All right, so my very short AI wish list, AKA what I'm going to use cloud code to build. Number one, an automated workflow for this podcast, Workflow that I'm doing right now. And number two, a custom dashboard to manage my one on one messaging service clients.
[00:11:36] It is my humble opinion and the opinion of many reasonable and rational people on threads, that AI is going to revolutionize custom web apps and personal web apps.
[00:11:48] It already has, right? It already is. I don't know if you guys remember, but that custom dashboard, or excuse me, let me just say the custom dashboard that I want to make will be an example of that.
[00:11:59] The pill tracker that I made, I don't know if you remember that, me talking about that. That was another. That is an example of these custom, you know, personal web apps. And that was my first foray into, into vibe coding. And I'm like, this is awesome, right? So the workflow you're going to see why I want to automate this. The workflow that I'm looking to automate, Please note, I am never outsourcing the actual creative portion of things. Right? I want to outsource the other thing so that I can just focus on. I really, really love making this podcast and I really enjoy the companion. I, I don't like the, like, what's called the last mile, which usually typically, which usually involves like copying pasting a million times, but I love, I love recording this podcast. It's actually very, very fun for me. I want to get rid of the other stuff, all right? So I'm never looking to outsource the actual creative portion. I'm looking to outsource largely what is considered to be that last mile and typically involves a shit ton of copy and pasting.
[00:12:55] So my process for this podcast, all right, is I write this out. I write the outline. It's very extensive, usually like four or five pages, single space. I record the podcast like I'm doing right now. I then paste the outline into Claude. Uh, from there I have Claude generate the show notes and the companion.
[00:13:13] The. From there I copy the show notes into a Google Doc. I also copy the companion into a separate Google Doc. I edit both so that it's like my voice. It's, it's, you know, what I want it to be.
[00:13:25] From there, I copy the show notes to an episode post on my website.
[00:13:30] I also upload the episode MP3 file, which is the recording of the episode. I also, uh, upload, uh, the COVID art, which is just the podcast cover art, but I have to set it. And then I schedule the episode to go out a week and now eight days later, from there I copy the companion. Right? I said that was, that was in one of The Google Docs. I copy the companion to an email template in kit to go out as the newsletter, and I schedule it.
[00:13:56] I then copy that companion, same thing, to a blog post on my website.
[00:14:01] I go and generate a featured image on Gemini, which has been fucking sucking, I'm just gonna say. And it has like a little watermark on it now, and I hate it. But either way, I have to create a face featured image. And I use Claude. She would come up with the prompt. And then I put that inside of Gemini and I get that image. Then I upload that image to Drive and I also upload that image to the blog post as a featured image. Then I schedule said blog post to go live.
[00:14:27] Then. Yeah, I'm not even done yet. We're almost done, though. Then I copy the podcast episode link from my website and I copy the blog episode link and I put them on a master Google sheet.
[00:14:38] So you see, it's lost lots of copy and pasting just lots of, like, monotonous little things. And it's annoying as. And I'm like, like, this doesn't need to take as long. Claude Code can do this for me, okay? And like, maybe you're like, what? And I'm like, yes, it can do it for me. I will have to give it access to my website, give it access to my ConvertKit or kit. My. That's my email marketing service. Uh, give it access to Google Drive.
[00:15:05] But it can absolutely do all those things, right? And paste. Cannot do all those copy and paste steps. And then I can just go in and check it and schedule it, which is way better. It can do more than that. It could write the thing for me. It could do the whole thing. It could write the outline for this. It could turn it into a newsletter. Uh, it could paste it. It could write the.
[00:15:26] The blog post about a different blog post, and it could do way more stuff. I just wanted to take the stuff that I have that's already in Google Drive and copy and paste it into the correct places. That's it.
[00:15:37] And it can do that, right? Uh, and I just have to go and do the very end, which is check it, make sure it's correct, schedule it, and that's way better.
[00:15:46] I could honestly have it schedule it as well. And I just go and check. But I'd rather have it post as drafts, and then I go and schedule it. But how exactly that happens is something that I will discuss in a later episode. Because at this moment, all I only know for, like, full certainty is that it can for sure Complete these tasks via, uh, a combination of what's called a Claude MD file, a markdown file that would have like, all the instructions, things like that, um, and what's called MCP server integrations. But don't worry about that for now. And once I build it out and get it running and I understand it, then I'll be sharing what I understand with you. I actually wanted to do it for this episode, but I was chatting with Claude, my French assistant, and Claude said it would take like a few hours to build out. Though. Worth noting, I have read plenty of cases on threads where Claude overshot its time to completion by, like, a lot.
[00:16:36] Um, but it was just like, hey, this is gonna take like three to four hours. Like, do it on a weekend. You said you're really busy with all this stuff, like, just do your regular flow. And I was like, you know what, I appreciate that. I appreciate you being realistic because I have so much to do. So I'm doing this one myself. I'm gonna be copying and pasting all the things, um, and it stays on the wish list for now. But this allowed me to make this episode and then I will be tackling that in the next few weeks.
[00:17:01] Wish list item number two, which is the second, last and final item, is a personalized dashboard that I can use with my one on one messaging service clients. So for those of you who don't know, um, I rolled out a bunch of new messaging services for my main business, the Movement Maestro.
[00:17:16] These services, I help people with their messaging, their, their communication.
[00:17:22] These services have a start date, a stop date, email correspondence, due dates for forms, feedback, end dates, client specific URLs, bits of information.
[00:17:32] It would generally be nice to have all in one area.
[00:17:36] And they also. It also has. These services also have repetitive emails that would be great to be able to generate with just the click of a button. So you might be thinking, uh, maestro, that's a CRM. And you are right. But all the CRMs out there, like, so much tech.
[00:17:53] I should say all the CRMs that are out there, just like so much tech, right? So much of the tech that's out there, it's bloated and expensive and I just need a few features and I want to look how I want it to look and do exactly what I want it to do. So I'm a build it out. I'm going to build out exactly what I need now for this. I could actually put this together with a regular, you know, with regular cloud. I don't need cloud code. I don't need Claude work, cloud, cowork. I could do it with regular Claude and um, I could have it generate the code and I could plug it into Cursor, which is what's called an ide, an uh, integrated development environment.
[00:18:25] And it's, it's just like a, it's like a place where you can type code and. But it's visual and clear. Cursor is nice because it is AI assisted. Um, but I could do that and I could create, you know, a custom Next JS web app, which I, which I have done. Right. That, um, the pill tracker that I was talking about before, um, that's a custom Next JS web app. The. I made a website for my brother. He doesn't use it, but I made one. Um, I have made a really, um, cool map, uh, for Lex's dad, where that you can color in. He does like territories.
[00:19:02] Um, he's a sales rep and so it's. He was looking to, uh, be able to color code the different territories. And this way they could have the different drivers be like, okay, this one, they're kind of like redistributing the territories. And so I built out a map that's. You could, uh, it's interactive and you can color in the areas and zoom in on things, um, from New York state. So it was pretty cool. Um, I could do that for this dashboard.
[00:19:25] Absolutely.
[00:19:27] But what I think I'm going to do is lean into Claude code here and have it build it for me. And that is the power of these agentic tools. Right. You literally type build me this or I'd like this, and it will execute it. Now there is clearly a ton more involved, but like, at the most simplistic, like, uh, thing I want you to understand is like, that's how you're interacting with it. You can use natural language to interact with this thing. We know that the more steps you give it, the more specific you can be, the better. But you don't have to know how to program is what I'm trying to say. I took a different route and started trying to learn some of the, like, what is actually being generated. I could never generate it on my own, just like out of thin air. But I was like, I wanted to understand what is happening and like, you know, the, the, the scaffolding for things and what I like using and how things get deployed and where they live, how to fix things if they break. But this is what's so amazing about this technology is that because it uses an LLM, a large language model as its brain.
[00:20:31] You can just use natural language and it will do things. That's incredible. And they have made cloud code so much more accessible. Right. Uh, so before you could only interact with it via the command line, so the terminal on your computer and then they've made it so that it has a dip as part of the desktop app. You have to be paying in order to use cloud code. You have to be on the least of twenty doll dollar a month plan.
[00:20:51] But you can access it from your phone now. It's just incredible. Like they have really just made it just so much access, which is smart for them because people use it and then you pay money.
[00:21:01] Um, but to me this is just like, yes, it is absolutely going to revolutionize how people, the people that want to create personal web apps. And that's what I want to do and I'm going to do. That's the second item on my, my wish list. Right.
[00:21:17] I really do believe all that in mind. I really do believe, uh, that we are in a time where it's not about AI getting better. That's not our focus. Right. Even though it is absolutely getting better, we're in a time where it's about leveling up how we use it and what we have it build for us. And that's the seed that I want to plant with this episode.
[00:21:40] So I'm looking at a time this episode is getting longer than I want it to be. So I'm going to cut it there. Um, and yes, I will of course keep you updated on how it goes, everything turns out. I will, you know, link what I can link, uh, and show you. Um, I, I was vlogging for a bit. I gotta get back to that. But I haven't been coding as much either because I've been delivering on those messaging services.
[00:21:59] It's been taking literally all my time. And then I was doing the, the rebrand, which was taking the rest of my time. Um, but I will get back on that. So if you want to see things, um, the vlog is a really nice way because I do like share screens and you can see what I'm typing in the, you know, in the computer and such and, um, what I'm building. So I will link that in the show notes as well. Um, but for what it's worth, it's also just on the website, if you go to prompting curiosity.com forward/vlog V L O G V as in Victor. Um, you'll see I think I have like seven or eight vlog, uh, posts up and I also have it on YouTube, for what it's worth. Um, but yeah, that is that. So last things last. Before I sign off, how I used AI this week, if you're new here, or, you know, even if you're not, each week, uh, or as I said, each episode, I share a quick example of how I used AI that week. This week, as you may have guessed it, I used Claude for a ton of planning. Regarding those things that are those two things that are on the wish list that I just shared with you. I, uh, chatted with Claude about what was possible and worth noting, folks. I did have to prompt it to suggest for it to have it suggest using cloud code, right, to execute that automated workflow, which is really interesting to me. I was just like, can't cloud code do this? And it was like, let me check. And it was like, yes, it can.
[00:23:17] Uh, in true LLM fashion, right? It often goes along with what you're saying.
[00:23:21] It doesn't necessarily have to fully agree. And I think Claude is actually a bit better at being less sycophantic than, than ChatGPT.
[00:23:28] Um, and I'm really liking it. I am fully 100 happy with the Switch, but I, I was chatting with, with Claude and I expressed familiarity with Google Apps script and it just kind of ran with that.
[00:23:40] And yes, that's could be a very viable approach with some API keys. I know I'm kind of talking gibberish to some of you, but either way, there's a different way I could do it that's like more, a little bit more manual, um, and especially for the dashboard. And I was like, yeah, I could do that. And I was like, but I do, I do think I'm ready to lean into cloud code and just like, have it do it. Um, and also for the workflow automation, cloud code is objectively a better, a way better option. So what I'd love for you to take away from that is ask Claude how it can help you.
[00:24:09] Right? Think about any repetitive tasks that you have and ask Claude how it can help you automate them. Right? This is not using AI, Right? It is not at all about replacing creativity. I will stand 10 toes down on that. It's not at all about replacing creativity. Absolutely not.
[00:24:27] Using AI is about offloading the repetitive, redundant tasks that add up. And honestly, because you're offloading them, you create more space for creativity.
[00:24:38] All right, that right there, folks. That is all for today. Hopefully you found this episode helpful. If you did, I would love it. Like I said earlier, if you snap that picture and shared it on the socials. You can tag me and the pod me, the movement maestro and the pod Prompting Curiosity podcast. Or honestly, just me. It's totally fine. Just check. At least tag me. It's the most important one. Um, but I do want to show off that artwork and I am super, super grateful for you listening. Don't forget, folks, uh, I also have that companion newsletter, the Curious Companion, that drops every Thursday. That is basically the podcast episode in text format. It is the blog. All right. This is the podcast. This is the blog. The blog is the podcast in text format. And the companion gets sent to your inbox. So if you prefer to read or you just want written record of all the things, join the newsletter fam. You can head to prompting curiosity.com forward/newsletter or you can check out the link in the show notes.
[00:25:31] Gonna wrap it up there.
[00:25:32] As always, my friends, endlessly, endlessly give it to me one more time. Endlessly appreciative for every single one of you. Until we chat again next Thursday, stay curious.
[00:25:46] Sam M.
