I Asked Fable 5.1 and GPT-6 Astra to Get Me Out of Copy Paste Hell. The Results Surprised Me.

I gave one prompt to two models and I I gave one prompt to two models and I got two completely different apps, both got two completely different apps, both got two completely different apps, both trying to solve the exact same problem. trying to solve the exact same problem. trying to solve the exact same problem. Copy paste. Hell. Fable built ledge, a Copy paste. Hell. Fable built ledge, a Copy paste. Hell. Fable built ledge, a narrow list that slides in from the narrow list that slides in from the narrow list that slides in from the right side of my screen. Astra decided right side of my screen. Astra decided right side of my screen. Astra decided on shelf, a wide bar across the bottom on shelf, a wide bar across the bottom on shelf, a wide bar across the bottom with nice big cards and a preview pane. with nice big cards and a preview pane. with nice big cards and a preview pane. They both got the exact same brief. They They both got the exact same brief. They They both got the exact same brief. They both got the same amount of time and both got the same amount of time and both got the same amount of time and both of them are functional apps. both of them are functional apps. both of them are functional apps. They're good apps. And I'm a little bit They're good apps. And I'm a little bit They're good apps. And I'm a little bit surprised myself. But the thing nobody surprised myself. But the thing nobody surprised myself. But the thing nobody is measuring is how you compare two is measuring is how you compare two is measuring is how you compare two models in the second round. What if you models in the second round. What if you models in the second round. What if you get past that vi coding fiveline prompt? get past that vi coding fiveline prompt? get past that vi coding fiveline prompt? What do you do then? How do you go from What do you do then? How do you go from What do you do then? How do you go from there to done? That's what we're going there to done? That's what we're going there to done? That's what we're going to go through. It's the pass you make to go through. It's the pass you make to go through. It's the pass you make after you've held the thing in your after you've held the thing in your after you've held the thing in your hands. And I think that matters more hands. And I think that matters more hands. And I think that matters more than ever now. You can get both shelf than ever now. You can get both shelf than ever now. You can get both shelf and ledge at the end of this video. I'll and ledge at the end of this video. I'll and ledge at the end of this video. I'll put the link up and it'll be super easy put the link up and it'll be super easy put the link up and it'll be super easy for you. So context, I gave ChatgPT6 for you. So context, I gave ChatgPT6 for you. So context, I gave ChatgPT6 Astra and Claude Fable 5.1 the exact Astra and Claude Fable 5.1 the exact Astra and Claude Fable 5.1 the exact same job. I said, "Please, I struggle same job. I said, "Please, I struggle same job. I said, "Please, I struggle with copy and pasting. I've talked about with copy and pasting. I've talked about with copy and pasting. I've talked about copy pasting hell on this channel.

copy pasting hell on this channel. copy pasting hell on this channel. Please, please, please build me a native Please, please, please build me a native Please, please, please build me a native Mac clipboard manager that feels like Mac clipboard manager that feels like Mac clipboard manager that feels like Apple should have included it. Something Apple should have included it. Something Apple should have included it. Something that remembers the text, the images, the that remembers the text, the images, the that remembers the text, the images, the links I've copied, lets me pull them links I've copied, lets me pull them links I've copied, lets me pull them back with a keyboard shortcut. I don't back with a keyboard shortcut. I don't back with a keyboard shortcut. I don't have to remember which one it was. I have to remember which one it was. I have to remember which one it was. I don't only use the most recent one, don't only use the most recent one, don't only use the most recent one, etc." So, for this app experience, Astra etc." So, for this app experience, Astra etc." So, for this app experience, Astra got me something that I liked that I got me something that I liked that I got me something that I liked that I actually used for copy and pasting. It actually used for copy and pasting. It actually used for copy and pasting. It got me there using fewer tokens. It got me there using fewer tokens. It got me there using fewer tokens. It saved me about an hour and it gave me saved me about an hour and it gave me saved me about an hour and it gave me more time to try it and ask for changes. more time to try it and ask for changes. more time to try it and ask for changes. So, I got through more revolutions of So, I got through more revolutions of So, I got through more revolutions of the app. Fable gave me a design that I the app. Fable gave me a design that I the app. Fable gave me a design that I also liked, but ultimately it wasn't as also liked, but ultimately it wasn't as also liked, but ultimately it wasn't as usable because of tiny little details usable because of tiny little details usable because of tiny little details like hotkeys, and you'll see that as we like hotkeys, and you'll see that as we like hotkeys, and you'll see that as we dive in more. And I really really care dive in more. And I really really care dive in more. And I really really care about what I can actually hand over and about what I can actually hand over and about what I can actually hand over and how I use what comes back. So I'm more how I use what comes back. So I'm more how I use what comes back. So I'm more productive with AI. And in this case, productive with AI. And in this case, productive with AI. And in this case, the job was small enough that I could the job was small enough that I could the job was small enough that I could try it immediately, right? I could copy try it immediately, right? I could copy try it immediately, right? I could copy something. I could open up the panel in something. I could open up the panel in something. I could open up the panel in ledge, open up the panel in shelf. I ledge, open up the panel in shelf. I ledge, open up the panel in shelf. I could find it again. I could get back to could find it again. I could get back to could find it again. I could get back to what I was doing and I just got into the what I was doing and I just got into the what I was doing and I just got into the habit of actually using it. And that's habit of actually using it. And that's habit of actually using it. And that's the ultimate test for software. Now the ultimate test for software. Now the ultimate test for software. Now let's look at some of the different let's look at some of the different let's look at some of the different choices these models made. Ledge is a choices these models made. Ledge is a choices these models made. Ledge is a narrow list view that pops in from the narrow list view that pops in from the narrow list view that pops in from the right of the screen. You can copy a right of the screen. You can copy a right of the screen. You can copy a color. It will show you a whole swatch.

color. It will show you a whole swatch. color. It will show you a whole swatch. Code looks like code. Shelf doesn't have Code looks like code. Shelf doesn't have Code looks like code. Shelf doesn't have quite that precision. Shelf is wider. It quite that precision. Shelf is wider. It quite that precision. Shelf is wider. It has bigger cards to look at. It has has bigger cards to look at. It has has bigger cards to look at. It has filters and it has a larger preview. You filters and it has a larger preview. You filters and it has a larger preview. You can see really different ideas about how can see really different ideas about how can see really different ideas about how this tool would fit into your this tool would fit into your this tool would fit into your day-to-day. And that is one of the day-to-day. And that is one of the day-to-day. And that is one of the things I loved is that I can try both things I loved is that I can try both things I loved is that I can try both out functionally and evolve them as I out functionally and evolve them as I out functionally and evolve them as I go. I really like that Fable gave me go. I really like that Fable gave me go. I really like that Fable gave me something meaningfully different when something meaningfully different when something meaningfully different when you ask both models for the same thing. you ask both models for the same thing. you ask both models for the same thing. One of the really useful outcomes is One of the really useful outcomes is One of the really useful outcomes is discovering that you hadn't decided discovering that you hadn't decided discovering that you hadn't decided exactly what you wanted. Nor could you exactly what you wanted. Nor could you exactly what you wanted. Nor could you with a fiveline prompt. Did I want a with a fiveline prompt. Did I want a with a fiveline prompt. Did I want a discrete list? I didn't make that call. discrete list? I didn't make that call. discrete list? I didn't make that call. Did I want more room to inspect an Did I want more room to inspect an Did I want more room to inspect an image? I hadn't decided. Seeing image? I hadn't decided. Seeing image? I hadn't decided. Seeing different design decisions well executed different design decisions well executed different design decisions well executed across two different form factors, those across two different form factors, those across two different form factors, those preferences much easier to express and preferences much easier to express and preferences much easier to express and evolve and understand. Both models evolve and understand. Both models evolve and understand. Both models received exactly the same opening brief, received exactly the same opening brief, received exactly the same opening brief, but they absolutely didn't take but they absolutely didn't take but they absolutely didn't take identical paths forward, and I am so identical paths forward, and I am so identical paths forward, and I am so glad they didn't. Fable decided to ask glad they didn't. Fable decided to ask glad they didn't. Fable decided to ask follow-up questions. It included which follow-up questions. It included which follow-up questions. It included which edge I wanted it on. Astra chose a edge I wanted it on. Astra chose a edge I wanted it on. Astra chose a bottom shelf on its own. I ended up bottom shelf on its own. I ended up bottom shelf on its own. I ended up using Fable through Claude Co-work and using Fable through Claude Co-work and using Fable through Claude Co-work and Astra through Codeex on my Mac. So, the Astra through Codeex on my Mac. So, the Astra through Codeex on my Mac. So, the experience includes the applications and experience includes the applications and experience includes the applications and access around the models as well as the access around the models as well as the access around the models as well as the models themselves. And that's pretty models themselves. And that's pretty models themselves. And that's pretty typical with models and harnesses these typical with models and harnesses these typical with models and harnesses these days. setup friction becomes real days. setup friction becomes real days. setup friction becomes real friction for someone who's trying to get friction for someone who's trying to get friction for someone who's trying to get work done very very quickly. And what's work done very very quickly. And what's work done very very quickly. And what's interesting about both of these models interesting about both of these models interesting about both of these models and the apps that they created is that and the apps that they created is that and the apps that they created is that they illustrate how friction can feel they illustrate how friction can feel they illustrate how friction can feel different to different folks. I wouldn't different to different folks. I wouldn't different to different folks. I wouldn't want to call the distinct differences in

want to call the distinct differences in want to call the distinct differences in these design decisions a difference in these design decisions a difference in these design decisions a difference in intelligence. Instead, I think what intelligence. Instead, I think what intelligence. Instead, I think what we're seeing is a distinct perspective we're seeing is a distinct perspective we're seeing is a distinct perspective that helped me as the app creator that helped me as the app creator that helped me as the app creator understand better what I was looking for understand better what I was looking for understand better what I was looking for out of that solution space. Shelf opens out of that solution space. Shelf opens out of that solution space. Shelf opens with control, shift, spacebar. I love with control, shift, spacebar. I love with control, shift, spacebar. I love that big spacebar key. It was really that big spacebar key. It was really that big spacebar key. It was really easy as a hotkey to just grab it and go. easy as a hotkey to just grab it and go. easy as a hotkey to just grab it and go. And there was something intuitive about And there was something intuitive about And there was something intuitive about having the space bar at the bottom of my having the space bar at the bottom of my having the space bar at the bottom of my keyboard and the shelf pop up from the keyboard and the shelf pop up from the keyboard and the shelf pop up from the bottom. Ledge uses command shiftV, which bottom. Ledge uses command shiftV, which bottom. Ledge uses command shiftV, which is fine, but I found shelf shortcut is fine, but I found shelf shortcut is fine, but I found shelf shortcut easier to remember. That sounds almost easier to remember. That sounds almost easier to remember. That sounds almost too small to matter in a model too small to matter in a model too small to matter in a model comparison. It's certainly not an an comparison. It's certainly not an an comparison. It's certainly not an an intelligence measure. But from a utility intelligence measure. But from a utility intelligence measure. But from a utility perspective, if you're trying to move perspective, if you're trying to move perspective, if you're trying to move fast with copy paste and you're already fast with copy paste and you're already fast with copy paste and you're already used to control crlv, and you find used to control crlv, and you find used to control crlv, and you find something in that part of the keyboard something in that part of the keyboard something in that part of the keyboard that helps you go and like use hotkeys that helps you go and like use hotkeys that helps you go and like use hotkeys to grab those earlier copy paste off the to grab those earlier copy paste off the to grab those earlier copy paste off the shelf, that felt more intuitive for my shelf, that felt more intuitive for my shelf, that felt more intuitive for my hands. Uh, and so that was something hands. Uh, and so that was something hands. Uh, and so that was something that I factored in because I think a lot that I factored in because I think a lot that I factored in because I think a lot of what we want to do with software is of what we want to do with software is of what we want to do with software is make it so easy it fades into the make it so easy it fades into the make it so easy it fades into the background and it's just transparent. I background and it's just transparent. I background and it's just transparent. I also slightly preferred Astra's visual also slightly preferred Astra's visual also slightly preferred Astra's visual design, particularly in the way it design, particularly in the way it design, particularly in the way it handled cards and preview. That doesn't handled cards and preview. That doesn't handled cards and preview. That doesn't turn my taste into like your taste. You turn my taste into like your taste. You turn my taste into like your taste. You might have a different preference here.

might have a different preference here. might have a different preference here. Someone else might prefer a narrow list Someone else might prefer a narrow list Someone else might prefer a narrow list in particular if they mainly want to in particular if they mainly want to in particular if they mainly want to grab a piece of text and move on. But grab a piece of text and move on. But grab a piece of text and move on. But then I started finding things I wanted then I started finding things I wanted then I started finding things I wanted to change. Some cards weren't lining up to change. Some cards weren't lining up to change. Some cards weren't lining up properly on the left. I wanted to properly on the left. I wanted to properly on the left. I wanted to reposition the panel and be able to drag reposition the panel and be able to drag reposition the panel and be able to drag and drop it. The number shortcut seemed and drop it. The number shortcut seemed and drop it. The number shortcut seemed to trigger other things on my Mac. And to trigger other things on my Mac. And to trigger other things on my Mac. And when I copied something, I actually when I copied something, I actually when I copied something, I actually wanted to know I did it, right? There wanted to know I did it, right? There wanted to know I did it, right? There was no visible message telling me it had was no visible message telling me it had was no visible message telling me it had worked. And the last one is such an worked. And the last one is such an worked. And the last one is such an ordinary software detail. Like you do ordinary software detail. Like you do ordinary software detail. Like you do the thing, you just want to know that it the thing, you just want to know that it the thing, you just want to know that it worked because if you don't have that worked because if you don't have that worked because if you don't have that confirmation, you might end up doing it confirmation, you might end up doing it confirmation, you might end up doing it multiple times or paste somewhere to multiple times or paste somewhere to multiple times or paste somewhere to check or wonder whether you selected the check or wonder whether you selected the check or wonder whether you selected the wrong item. Again, it's that tiny bit of wrong item. Again, it's that tiny bit of wrong item. Again, it's that tiny bit of uncertainty that creates extra mental uncertainty that creates extra mental uncertainty that creates extra mental load every time you use the application. load every time you use the application. load every time you use the application. I could tell Astra which edge didn't I could tell Astra which edge didn't I could tell Astra which edge didn't line up very simply with the screenshot. line up very simply with the screenshot. line up very simply with the screenshot. I could tell where I wanted to move the I could tell where I wanted to move the I could tell where I wanted to move the panel and I could tell Astra which key panel and I could tell Astra which key panel and I could tell Astra which key combination was interfering with another combination was interfering with another combination was interfering with another app. I also could just tell it I want to app. I also could just tell it I want to app. I also could just tell it I want to know that copy succeeded and I didn't know that copy succeeded and I didn't know that copy succeeded and I didn't have to specify a behavior to add. Astra have to specify a behavior to add. Astra have to specify a behavior to add. Astra was able to just figure this out. Now, was able to just figure this out. Now, was able to just figure this out. Now, you might wonder why didn't Fable get you might wonder why didn't Fable get you might wonder why didn't Fable get that same feedback. That's part of the that same feedback. That's part of the that same feedback. That's part of the speed test. I was able to get through speed test. I was able to get through speed test. I was able to get through 1.0, 1.1, and 1.2 with Astra in the time 1.0, 1.1, and 1.2 with Astra in the time 1.0, 1.1, and 1.2 with Astra in the time it took for Fable to build 1.0. And so I it took for Fable to build 1.0. And so I it took for Fable to build 1.0. And so I thought that was fair. Like you work thought that was fair. Like you work thought that was fair. Like you work through the speed and you can get more through the speed and you can get more through the speed and you can get more iterations through. And you can actually iterations through. And you can actually iterations through. And you can actually evolve as you go. And Astra was able to evolve as you go. And Astra was able to evolve as you go. And Astra was able to successfully work through those changes.

successfully work through those changes. successfully work through those changes. Shelf now remembers where I put it and Shelf now remembers where I put it and Shelf now remembers where I put it and it can be dragged and dropped. Shelf it can be dragged and dropped. Shelf it can be dragged and dropped. Shelf shows a copied message after a shows a copied message after a shows a copied message after a successful copy. the quick copy keys successful copy. the quick copy keys successful copy. the quick copy keys changed and adjusted and the app now changed and adjusted and the app now changed and adjusted and the app now waits until I release the key before waits until I release the key before waits until I release the key before returning focus to the previous returning focus to the previous returning focus to the previous application. And that last detail application. And that last detail application. And that last detail matters because sending focus back at matters because sending focus back at matters because sending focus back at the wrong moment can let the same key the wrong moment can let the same key the wrong moment can let the same key affect the app underneath and you don't affect the app underneath and you don't affect the app underneath and you don't want to do that. All of those changes want to do that. All of those changes want to do that. All of those changes got cycled into version 1.1 and 1.2 two got cycled into version 1.1 and 1.2 two got cycled into version 1.1 and 1.2 two and the installed build now passes over and the installed build now passes over and the installed build now passes over 65 different checks that Astra came up 65 different checks that Astra came up 65 different checks that Astra came up with covering things like clipboard with covering things like clipboard with covering things like clipboard restoration, saved history, keyboard restoration, saved history, keyboard restoration, saved history, keyboard behavior, and there was also live behavior, and there was also live behavior, and there was also live testing of the interactions. All of it testing of the interactions. All of it testing of the interactions. All of it is in the art of building for that is in the art of building for that is in the art of building for that second and third round of prompting. second and third round of prompting. second and third round of prompting. Now, the highest measure of success for Now, the highest measure of success for Now, the highest measure of success for me is whether someone else wanted to use me is whether someone else wanted to use me is whether someone else wanted to use this. Yes, I like to use it myself, but this. Yes, I like to use it myself, but this. Yes, I like to use it myself, but I showed it to someone who's extremely I showed it to someone who's extremely I showed it to someone who's extremely busy and has a ton going on in her life busy and has a ton going on in her life busy and has a ton going on in her life and is on her computer all day, and she and is on her computer all day, and she and is on her computer all day, and she said, "This app is something I would said, "This app is something I would said, "This app is something I would love to use." And she installed it right love to use." And she installed it right love to use." And she installed it right away and started using it. And so, to away and started using it. And so, to away and started using it. And so, to me, I felt like I was able to use AI to me, I felt like I was able to use AI to me, I felt like I was able to use AI to build something useful. I was able to build something useful. I was able to build something useful. I was able to use Fable and Astra to explore a use Fable and Astra to explore a use Fable and Astra to explore a solution space. I was able to understand solution space. I was able to understand solution space. I was able to understand what I liked and I was ultimately able what I liked and I was ultimately able what I liked and I was ultimately able to proceed with the build. Astra was to proceed with the build. Astra was to proceed with the build. Astra was able to produce a native app installer able to produce a native app installer able to produce a native app installer for Mac called a DMG file and I was for Mac called a DMG file and I was for Mac called a DMG file and I was actually able to get it signed uh and actually able to get it signed uh and actually able to get it signed uh and authorized by app so that I have a authorized by app so that I have a authorized by app so that I have a legitimate Mac installer and it becomes legitimate Mac installer and it becomes legitimate Mac installer and it becomes easy for anyone with a Mac to install.

easy for anyone with a Mac to install. easy for anyone with a Mac to install. Now, one of the things I want to circle Now, one of the things I want to circle Now, one of the things I want to circle back to is speed. Speed changed my back to is speed. Speed changed my back to is speed. Speed changed my preference and I I should be honest preference and I I should be honest preference and I I should be honest about that, right? When the initial about that, right? When the initial about that, right? When the initial build comes back sooner, I can try it build comes back sooner, I can try it build comes back sooner, I can try it while I have time and attention for the while I have time and attention for the while I have time and attention for the project. I can notice a problem I would project. I can notice a problem I would project. I can notice a problem I would never have thought to put in the first never have thought to put in the first never have thought to put in the first prompt and I have all of that brain prompt and I have all of that brain prompt and I have all of that brain space loaded up to understand what's space loaded up to understand what's space loaded up to understand what's going on and then I can get another going on and then I can get another going on and then I can get another version and see whether that change version and see whether that change version and see whether that change helped. Effectively having a faster helped. Effectively having a faster helped. Effectively having a faster speed of iteration affects the quality speed of iteration affects the quality speed of iteration affects the quality of what I end up being able to produce of what I end up being able to produce of what I end up being able to produce and share with all of you. And the first and share with all of you. And the first and share with all of you. And the first version became something that shaped how version became something that shaped how version became something that shaped how I want to ask for things next. the model I want to ask for things next. the model I want to ask for things next. the model didn't have to guess every preference at didn't have to guess every preference at didn't have to guess every preference at the beginning because it was moving the beginning because it was moving the beginning because it was moving faster and I didn't have to write a faster and I didn't have to write a faster and I didn't have to write a perfect spec before I'd used anything. perfect spec before I'd used anything. perfect spec before I'd used anything. We're now at a point where you can get a We're now at a point where you can get a We're now at a point where you can get a usable app out of a very short prompt usable app out of a very short prompt usable app out of a very short prompt like I shared with you and then you can like I shared with you and then you can like I shared with you and then you can evolve it one or two times, get it evolve it one or two times, get it evolve it one or two times, get it approved by Apple and then you can start approved by Apple and then you can start approved by Apple and then you can start to share it out. And for this build, to share it out. And for this build, to share it out. And for this build, Astra gave me that experience faster and Astra gave me that experience faster and Astra gave me that experience faster and with lower token use. Now, this was one with lower token use. Now, this was one with lower token use. Now, this was one build. Astra gave me this experience build. Astra gave me this experience build. Astra gave me this experience faster. Astra gave me this experience faster. Astra gave me this experience faster. Astra gave me this experience with lower token use. Am I going to tell with lower token use. Am I going to tell with lower token use. Am I going to tell you every single time Astra is going to you every single time Astra is going to you every single time Astra is going to be faster and lower token use or every be faster and lower token use or every be faster and lower token use or every single time I would pick the Astro single time I would pick the Astro single time I would pick the Astro build? I think it's not the case. I know build? I think it's not the case. I know build? I think it's not the case. I know that a lot of you get excited about that a lot of you get excited about that a lot of you get excited about picking the right model, picking the picking the right model, picking the picking the right model, picking the best model. Is this the best model for best model. Is this the best model for best model. Is this the best model for this? How do I go from one model to this? How do I go from one model to this? How do I go from one model to multimodel? Think about it like this. I multimodel? Think about it like this. I multimodel? Think about it like this. I am giving you the job resumeumés. I'm am giving you the job resumeumés. I'm am giving you the job resumeumés. I'm giving you the job fits. I'm giving you giving you the job fits. I'm giving you giving you the job fits. I'm giving you my experience as to how these LLMs did

my experience as to how these LLMs did my experience as to how these LLMs did in those jobs because if I lean too hard in those jobs because if I lean too hard in those jobs because if I lean too hard on one model or not on another model, on one model or not on another model, on one model or not on another model, I'm the one that's missing out on the I'm the one that's missing out on the I'm the one that's missing out on the intelligence that model brings to bear. intelligence that model brings to bear. intelligence that model brings to bear. And so it's my job to actually as the And so it's my job to actually as the And so it's my job to actually as the human figure out where that intelligence human figure out where that intelligence human figure out where that intelligence needs to be applied and put that model needs to be applied and put that model needs to be applied and put that model in the right spot. It's like giving in the right spot. It's like giving in the right spot. It's like giving someone a great job that's a great fit someone a great job that's a great fit someone a great job that's a great fit for them. Put that model in a spot where for them. Put that model in a spot where for them. Put that model in a spot where it can thrive. And that's been my focus it can thrive. And that's been my focus it can thrive. And that's been my focus and that continues to be my focus more and that continues to be my focus more and that continues to be my focus more and more as these models get smarter and and more as these models get smarter and and more as these models get smarter and smarter and smarter and more and more smarter and smarter and more and more smarter and smarter and more and more act like colleagues. Where does this act like colleagues. Where does this act like colleagues. Where does this colleague thrive? Where do they put them colleague thrive? Where do they put them colleague thrive? Where do they put them in a role where they can actually in a role where they can actually in a role where they can actually succeed? It'll be your task to match it succeed? It'll be your task to match it succeed? It'll be your task to match it up to look in a little bit more detail up to look in a little bit more detail up to look in a little bit more detail at the roles that you have for your at the roles that you have for your at the roles that you have for your models and to use videos like this as a models and to use videos like this as a models and to use videos like this as a guide to how you can help fit your guide to how you can help fit your guide to how you can help fit your models into the roles that you have for models into the roles that you have for models into the roles that you have for them. When I'm doing writing, if I am them. When I'm doing writing, if I am them. When I'm doing writing, if I am looking for a writing partner, this is a looking for a writing partner, this is a looking for a writing partner, this is a situation where if I'm thinking through situation where if I'm thinking through situation where if I'm thinking through the problem, if I'm wrestling with the problem, if I'm wrestling with the problem, if I'm wrestling with understanding the problem at a high understanding the problem at a high understanding the problem at a high level, I am absolutely starting with level, I am absolutely starting with level, I am absolutely starting with Fable and trying to like think it Fable and trying to like think it Fable and trying to like think it through and then I'm moving to Astra through and then I'm moving to Astra through and then I'm moving to Astra because Astra helps me to actually take because Astra helps me to actually take because Astra helps me to actually take a shape that has a little bit of a shape that has a little bit of a shape that has a little bit of structure to it and start to get it into structure to it and start to get it into structure to it and start to get it into something that resembles a writing plan.

something that resembles a writing plan. something that resembles a writing plan. It helps me to start to think through It helps me to start to think through It helps me to start to think through things very quickly from that point. things very quickly from that point. things very quickly from that point. Again, that speed comes in. But when Again, that speed comes in. But when Again, that speed comes in. But when you're initially thinking through a you're initially thinking through a you're initially thinking through a problem, you're not looking to have problem, you're not looking to have problem, you're not looking to have incredible speed to think that through incredible speed to think that through incredible speed to think that through necessarily. You want to get it right. necessarily. You want to get it right. necessarily. You want to get it right. You want to think it through carefully. You want to think it through carefully. You want to think it through carefully. And so that's that moment when I like I And so that's that moment when I like I And so that's that moment when I like I love the way Fable thinks deeply about love the way Fable thinks deeply about love the way Fable thinks deeply about problems. There's something special problems. There's something special problems. There's something special about it. And I go to Fable first in about it. And I go to Fable first in about it. And I go to Fable first in that situation. Then Astra helps me with that situation. Then Astra helps me with that situation. Then Astra helps me with the writing. And I realize that's a big the writing. And I realize that's a big the writing. And I realize that's a big shift. Like if we were talking say 3 shift. Like if we were talking say 3 shift. Like if we were talking say 3 months, four months, five months ago, we months, four months, five months ago, we months, four months, five months ago, we all would have made the assumption you all would have made the assumption you all would have made the assumption you go to Claude for the writing and then go to Claude for the writing and then go to Claude for the writing and then you don't go to chat GPT for the you don't go to chat GPT for the you don't go to chat GPT for the writing. But here I am in this video writing. But here I am in this video writing. But here I am in this video saying Astra gave me a front-end design saying Astra gave me a front-end design saying Astra gave me a front-end design for an app that I really liked. Astra for an app that I really liked. Astra for an app that I really liked. Astra was able to help me like as a writing was able to help me like as a writing was able to help me like as a writing companion and I feel like I can co-write companion and I feel like I can co-write companion and I feel like I can co-write with it really well. And that is a with it really well. And that is a with it really well. And that is a change. Astra is very steerable in change. Astra is very steerable in change. Astra is very steerable in writing. What about spreadsheets? How do writing. What about spreadsheets? How do writing. What about spreadsheets? How do I think about spreadsheets? If I'm I think about spreadsheets? If I'm I think about spreadsheets? If I'm working on a big complicated working on a big complicated working on a big complicated spreadsheet, my inclination right now is spreadsheet, my inclination right now is spreadsheet, my inclination right now is to have Astra do a draft, to have Astra to have Astra do a draft, to have Astra to have Astra do a draft, to have Astra structure out and prove out that all the structure out and prove out that all the structure out and prove out that all the formulas work and do an initial analysis formulas work and do an initial analysis formulas work and do an initial analysis and then go to Fable for how I would and then go to Fable for how I would and then go to Fable for how I would design that because I still think design that because I still think design that because I still think there's something special for both uh there's something special for both uh there's something special for both uh PowerPoint and for Excel in the way PowerPoint and for Excel in the way PowerPoint and for Excel in the way anthropic designs although I know that anthropic designs although I know that anthropic designs although I know that Astro has made huge strides in that Astro has made huge strides in that Astro has made huge strides in that department. And so on the PowerPoint department. And so on the PowerPoint department. And so on the PowerPoint side, I actually look at uh bulleting side, I actually look at uh bulleting side, I actually look at uh bulleting out and trying to plan the story in out and trying to plan the story in out and trying to plan the story in Astra, assuming I know what I'm trying Astra, assuming I know what I'm trying Astra, assuming I know what I'm trying to get at. I have a little bit of that

to get at. I have a little bit of that to get at. I have a little bit of that structure. And then I move to Fable. And structure. And then I move to Fable. And structure. And then I move to Fable. And Fable helps to flush it out. But Astra Fable helps to flush it out. But Astra Fable helps to flush it out. But Astra can be like a really powerful research can be like a really powerful research can be like a really powerful research companion to get there. When it comes to companion to get there. When it comes to companion to get there. When it comes to how I use computer use, how I integrate how I use computer use, how I integrate how I use computer use, how I integrate with other tools on on my Mac, I'm going with other tools on on my Mac, I'm going with other tools on on my Mac, I'm going to Astra in Codeex or Co-work for that to Astra in Codeex or Co-work for that to Astra in Codeex or Co-work for that because the speed at which that model because the speed at which that model because the speed at which that model can use the computer, it's night and day can use the computer, it's night and day can use the computer, it's night and day difference. It's like it helps me to be difference. It's like it helps me to be difference. It's like it helps me to be much more effective. And I don't think much more effective. And I don't think much more effective. And I don't think in that case it's a particularly close in that case it's a particularly close in that case it's a particularly close comparison between where computer use is comparison between where computer use is comparison between where computer use is for codecs and co-work and where for codecs and co-work and where for codecs and co-work and where computer use is for the claude computer use is for the claude computer use is for the claude application. Frankly, cloud needs to application. Frankly, cloud needs to application. Frankly, cloud needs to catch up on computer use and go faster catch up on computer use and go faster catch up on computer use and go faster on computer use to be more effective. on computer use to be more effective. on computer use to be more effective. Uh, and in this case, speed helps. The Uh, and in this case, speed helps. The Uh, and in this case, speed helps. The last thing I want to talk about is how last thing I want to talk about is how last thing I want to talk about is how we think about multi- aent systems we think about multi- aent systems we think about multi- aent systems because I know people get real excited, because I know people get real excited, because I know people get real excited, Nate, what do you think about Nate, what do you think about Nate, what do you think about multi-agent systems? How do you think multi-agent systems? How do you think multi-agent systems? How do you think about handling them here? Is Astra about handling them here? Is Astra about handling them here? Is Astra expensive in a multi-agent system? What expensive in a multi-agent system? What expensive in a multi-agent system? What about Fable? What role does it play? You about Fable? What role does it play? You about Fable? What role does it play? You may be wondering, did I use a multi- may be wondering, did I use a multi- may be wondering, did I use a multi- aent setup to get this little app that I aent setup to get this little app that I aent setup to get this little app that I built, this shelf app, this Ledge app built, this shelf app, this Ledge app built, this shelf app, this Ledge app done? No, I did not. I didn't think it done? No, I did not. I didn't think it done? No, I did not. I didn't think it needed it. And I think that underscores needed it. And I think that underscores needed it. And I think that underscores how much you can get done without how much you can get done without how much you can get done without jumping into multi- aent. I got real jumping into multi- aent. I got real jumping into multi- aent. I got real software built with just one agent, one software built with just one agent, one software built with just one agent, one prompt. It was like five lines. I showed prompt. It was like five lines. I showed prompt. It was like five lines. I showed it to you and the software was just it to you and the software was just it to you and the software was just there. I will tell you, I am a big fan

there. I will tell you, I am a big fan there. I will tell you, I am a big fan of open multi- aent systems. Multi-agent of open multi- aent systems. Multi-agent of open multi- aent systems. Multi-agent systems that let you evolve them as you systems that let you evolve them as you systems that let you evolve them as you go. So, for example, having an open go. So, for example, having an open go. So, for example, having an open collaboration point. Uh I launched collaboration point. Uh I launched collaboration point. Uh I launched Ringer for that a few weeks ago, but Ringer for that a few weeks ago, but Ringer for that a few weeks ago, but there's others. All it requires is there's others. All it requires is there's others. All it requires is making sure both of the coordinating making sure both of the coordinating making sure both of the coordinating entities in this case claude and codeex entities in this case claude and codeex entities in this case claude and codeex or co-work have access to a repo have or co-work have access to a repo have or co-work have access to a repo have access to a message board. And you're access to a message board. And you're access to a message board. And you're able to specify in each case what the able to specify in each case what the able to specify in each case what the roles of these agents are. And so you roles of these agents are. And so you roles of these agents are. And so you may specify and say Fable I want you to may specify and say Fable I want you to may specify and say Fable I want you to coordinate and design. Uh which is a coordinate and design. Uh which is a coordinate and design. Uh which is a good role for Fable. I want you to think good role for Fable. I want you to think good role for Fable. I want you to think it through. I want you to design what it through. I want you to design what it through. I want you to design what this app should do and I want you to this app should do and I want you to this app should do and I want you to sketch it out and build it out. Astra, I sketch it out and build it out. Astra, I sketch it out and build it out. Astra, I want you to make sure that you have want you to make sure that you have want you to make sure that you have checked what Fable is doing and then checked what Fable is doing and then checked what Fable is doing and then farm that out to a much cheaper OpenAI farm that out to a much cheaper OpenAI farm that out to a much cheaper OpenAI model to execute. Luna comes to mind. model to execute. Luna comes to mind. model to execute. Luna comes to mind. Luna is really cheap and very good at Luna is really cheap and very good at Luna is really cheap and very good at execution. Would do it really, really execution. Would do it really, really execution. Would do it really, really quickly. So, that's an example of how to quickly. So, that's an example of how to quickly. So, that's an example of how to do a multi- aent model. I think one of do a multi- aent model. I think one of do a multi- aent model. I think one of the things I would call out as I reflect the things I would call out as I reflect the things I would call out as I reflect on this build is that I actually want to on this build is that I actually want to on this build is that I actually want to explore with Fable even more now that explore with Fable even more now that explore with Fable even more now that I've seen what it can do and the I've seen what it can do and the I've seen what it can do and the attention to detail it puts into the attention to detail it puts into the attention to detail it puts into the build and building this again or getting build and building this again or getting build and building this again or getting to a 1.3 1.4 with Astra, I would to a 1.3 1.4 with Astra, I would to a 1.3 1.4 with Astra, I would probably pull lessons learned from the probably pull lessons learned from the probably pull lessons learned from the fable design exploration and pull them fable design exploration and pull them fable design exploration and pull them into into this build with Astra so that into into this build with Astra so that into into this build with Astra so that I can start to take the best of both I can start to take the best of both I can start to take the best of both worlds. We guys, we are spoiled for worlds. We guys, we are spoiled for worlds. We guys, we are spoiled for riches. We live in a world where we have riches. We live in a world where we have riches. We live in a world where we have two different competing AI companies

two different competing AI companies two different competing AI companies that both in under two hours can deliver that both in under two hours can deliver that both in under two hours can deliver incredible builds for pieces of custom incredible builds for pieces of custom incredible builds for pieces of custom software. One of them happens to go a software. One of them happens to go a software. One of them happens to go a little bit faster than the other. And little bit faster than the other. And little bit faster than the other. And you're going to have to pick which one you're going to have to pick which one you're going to have to pick which one you like. My encouragement to you is you like. My encouragement to you is you like. My encouragement to you is because they're capable of showing you because they're capable of showing you because they're capable of showing you really different perspectives on really different perspectives on really different perspectives on solution spaces. You will be smarter if solution spaces. You will be smarter if solution spaces. You will be smarter if you use both. It's like Godzilla versus you use both. It's like Godzilla versus you use both. It's like Godzilla versus Megalodon. The audience wins. It's a Megalodon. The audience wins. It's a Megalodon. The audience wins. It's a great movie. and I'll look forward to great movie. and I'll look forward to great movie. and I'll look forward to sharing the next uh anthropic versus sharing the next uh anthropic versus sharing the next uh anthropic versus open AI comparison with you soon. I'm open AI comparison with you soon. I'm open AI comparison with you soon. I'm giving you those tidbits if you want to giving you those tidbits if you want to giving you those tidbits if you want to go farther, if you want to say, well, go farther, if you want to say, well, go farther, if you want to say, well, what does Nate think about this specific what does Nate think about this specific what does Nate think about this specific scenario? That's where the Substack scenario? That's where the Substack scenario? That's where the Substack guide is going to be helpful because guide is going to be helpful because guide is going to be helpful because it's it's much much richer than what I it's it's much much richer than what I it's it's much much richer than what I can put here in the video. For now, I can put here in the video. For now, I can put here in the video. For now, I hope you enjoy shelf. Go and grab it and hope you enjoy shelf. Go and grab it and hope you enjoy shelf. Go and grab it and uh let me know what you think. I uh let me know what you think. I uh let me know what you think. I certainly had fun building it as as I certainly had fun building it as as I certainly had fun building it as as I see how easy it is to put little tools see how easy it is to put little tools see how easy it is to put little tools like this out. I'll definitely be using like this out. I'll definitely be using like this out. I'll definitely be using them more in my model comparison and them more in my model comparison and them more in my model comparison and I'll look forward to sharing the next uh I'll look forward to sharing the next uh I'll look forward to sharing the next uh anthropic versus open AI comparison with anthropic versus open AI comparison with anthropic versus open AI comparison with you soon.