Why Your AI Coding Projects Fail Before You Start

インデックス作成: 2026-05-22

[00:00:03]A lot of people have been asking me how they can use the 12x project launcher that you've seen in my previous videos.

[00:00:09]If you've watched any of those previous videos on the architect builder method, you already know the basic idea what I'm trying to do. You use chat GPT or claude or any other large language model as the architect. Then you use codeex or claude code or cursor as the builder. So the architect thinks through the software.

[00:00:28]The builder builds from the blueprint, right? But here's the part that people underestimate and was very clunky for me as I refine this process. The hardest part isn't the building going back and forth. The hardest part is getting started correctly because before you ever hand anything off to codeex or cloud code, you need to get the project set up. You need the folder structure.

[00:00:47]You need the starter prompt. You need the architect to understand the method.

[00:00:52]You need to capture the project goals, the users workflow data, risk, the requirements, the acceptance criteria.

[00:00:57]at least a basic of that and the builder handoff. You need to kind of have an idea of what you're trying to build. It can be very basic, but that project launcher really helps. If you skip all that kind of setting up, the builder is just going to be guessing and that's where those AI software projects fall apart. You got to have that discipline.

[00:01:17]So, I created this 12x project launcher to remove that starting friction that I I experienced starting these projects.

[00:01:23]It's going to walk you through some basic intake. It's going to help organize the project, create the folder structure for you with some templates and dummy data, and it's going to give you the starter architect prompt so you can begin the process cleanly. That is the most important part. And now you can have access to it. I've turned this into a web-based application that you can use yourself. You can go follow the link and grab it for yourself. And so in this video, I'm going to show you step by step how it works, how to get started in a project, and I'll take the project through completion. So, I'm going to demo a simple project on purpose. We're going to use a basic dashboard project where I'm going to take two disparit pieces of data, an Excel order export and a CSV support ticket export and turn that project into a clean executive dashboard. But don't get hung up on what we're creating here. I'm keeping it simple. You can use this launcher for this simple spreadsheet automation, a reporting workflow, an internal tool, a client portal, or full-blown software product. People ask me, can I use this only for workflows? No, you can create real software products in there. It's not just for workflows. It's for starting real software projects the right way. So the launcher does not replace the architect builder method.

[00:02:32]Just helps you get started, right? And so once you get started, the process then will still be architect builder, architect builder, architect builder, back and forth, updating your file structure, defining the sprint, handing it to the builder, validating what comes back, and then keeping it going. Rinse and repeat. So, let me show you how to start a real project with the 120X project launcher. All right. So, here we are with the project launcher. You'll get the first thing that you want to do and probably most important thing is setting up your architect workspace.

[00:03:03]Okay. And so, the architect workspace you remember is going to be chat GPT or claude the basic large language model.

[00:03:10]Any large language model that you want to use, but in this case in the demo, I'm going to use a chat GPT. And we're going to set up a project. And so let me bring in my chat GPT window in here. And essentially we're going to set up a new project. And so if I set up a new project, let's call this just demo for YouTube. And we're going to create the project. Now, if you've never worked in project before, many of you probably have. You notice you've got a place for chats and you've got a place for sources for files that are going to be part of your project. And so in this case, the first thing that I'm going to put in is the methodology that we're going to use for our project.

[00:03:55]So if you look at the architect workspace right here down here, there's a lot of text, but this is the markdown file of the whole architect builder method. You can read this here, but it's how to use this method. It's the culmination of 15 months of me building projects and coming up with this methodology. It used to be in five or six folders. Now it's in this one markdown folder. And this is important for your chats within your project to understand the architect builder method. So we're going to download this markdown file. And then in chat GPT, we're going to go ahead and we're going to add into our sources that folder that we just downloaded, right?

[00:04:42]That markdown file. It's this file right here. And now it's in there. And then, of course, in this case, I said we were going to use a couple data from data sheets, spreadsheets, and a CSV file. And those files are in I put in demo project on my desktop a folder called demo projects. And in here I got my demo data. And in the demo data I've got this spreadsheet that is going to be uh a whole series of orders with a bunch of messy data. And then the other thing is going to be this demo support tickets CSV file. And imagine this is a simple project, but just imagine that this you've had multiple 15, 17, 20, 50 different pieces of data to help with the dashboard. In this case, we're just going to use the two.

[00:05:32]And so I'm going to add those files into the project as well. Think I can drag and drop them? I can. And I'm going to put them into here. And so now my project has context, right? the project within chat GPT can be clawed as the same thing. You can do the exact same thing sources. So that way my chats will always have context and reference for these. But most importantly is that builder method that you get directly from the project starter which is right here. You can download or you can copy this and create your own markdown document if you want to. Now, the second thing, if you've ever used projects, is the ability to set custom instructions within your chat GPT or claude project.

[00:06:17]And so, in here, we're going to do the same thing. If you go up here in your settings, you can edit the instructions.

[00:06:24]And in this case, I can have specific custom instructions on how I want the chats within this project to respond.

[00:06:33]And so I thought it was important that I found that we can get some really good information around the architect builder method is if we in here create the custom instructions and this is what these are right here. All of this are the custom instructions that we're going to put into our project. So I'm going to copy these again straight from the project launcher. And then I'm going to go back to chat GPT and I'm going to paste these custom instructions into here and save it. And you can add other things, but as a minimum, you're going to want to have this where it talks about, you know, always act as the architect. You're not going to create code. You've got core rules about the architect first, builder second. And this is a very critical step because you're basically priming your architect to act like an architect. That's all we're doing. So, we're going to save it.

[00:07:24]And essentially our project now is primed with the custom instructions that we talked about and it's primed with any reference documents that we need for our code or our application and most importantly the markdown file that explains the architect builder method.

[00:07:41]So it always knows what to do. So our architect is ready to go, right? It's primed. It's set up. And so once we're done with that, now we want to go ahead and set up the project so that we can get that architect prompt that we can put into our architect. In this case, chat GBT or Claude. So for here, we're going to focus on starting a new project. So I'm going to click on that.

[00:08:07]And now it's asking me to put some inputs in here. And again, there are a few things that are mandatory, but a lot of it's optional because the architect will ask you questions. But the more that you can fill in here, the better it's going to be. And it's common sense.

[00:08:22]You got some helper inputs in here. So in this case, and I've copied this. I'm going to copy and paste for the sake of speed. But in this case, we're going to call this the executive data dashboard demo. And for the client that we're making this for, it's for the Northstar operating group.

[00:08:41]And then the project slug is following the rules, just kind of lowercase. and with dashes, no spaces, no underscore.

[00:08:51]And so we got a slug for there. The project type in this, we're going to say it is an internal tool. And then we're going to say for the text stack, again, if we don't know, if you know what you want to build it in, you can put it in here. Again, it's not mandatory. It's just helpful. In this case, I'm going to say it's TBD during the architecture.

[00:09:14]So, we'll figure that out with the architect ourselves, but you could put things in here like Node.js or React or local file upload or whatever you want to do. And in this case, we're going to have a one sentence description here, which is critical, but I would put something in here. In this case, we're going to turn two external business data exports into a professional executive dashboard that shows revenue, margins, support burden, refunds, and customer health. And then down here, the discovery context. This is where you want to take some time. You can have external conversations before you start here with Chat GPT or Claude to help you brainstorm it if you don't know. But if you don't know the answers, you don't have to necessarily answer them because the architect will kind of extract that out of you. But the more that you can put in here, the better. So I am going to put some goal a business goal in here. And the business goal or the business problem, we're going to say the business problem is the team exports ordered data from one system and support ticket data from another. Someone manually opens both files, copies numbers into a spreadsheet, creating rough summary, sends screenshot to leadership. The process is slow and inconsistent, hard to trust, it makes it difficult, so on and so forth. We could say uh the target outcome in this case I did do some homework on this. We could say here for a target workflow the team exports order from one system. We say the target outcome is that the user uploads an Excel order and a CSV and the app parses both files, matches the records, calculates metrics. The dashboard will help the leadership answer which product lines are generating revenue, which channels have the strongest margin, which regions are producing. So these are the things that I had in my head that I want this dashboard to do. We could say, well, who are the primary users? If we say the users, we could say it's like the CEO, CEO, revenue header, stuff like that.

[00:11:13]What else could we want to put? The current workflow that might be good to put in. So, I'm going to add this in here to the current workflow. Again, if you see rough notes are enough, but again, I'm copy and pasting because I did a little homework for this demo.

[00:11:27]This is the order that I want things to do. An analyst opens both files, so on and so forth. If we talk about the current pain points, the more the marrier that you can put in here, I've got pain points. And I'm going to leave some of these intentionally blank just to show you what it can do. And then just for the sake of time, I'm going to speed this up and you can watch me paste these things in here. Okay, I'm back. I paused this and I took the time to kind of fill in as much as I could. And again, this is where you're going to spend a lot of time. I highly suggest spending enough time in here to answer as much as you can. Again, a lot of this is optional. You don't have to. I'm going to leave these two intentionally blank so I can show you that the architect will help fill in the unknowns. And I even put a note down there at the bottom. I need help with the unknowns, everything in here, as much information as you can. And so once you get everything filled in to where you want, then you want to generate the virtual files. And so you click that and essentially what happens, it takes you to this executive data dashboard demo page and it asks you to do in order. So download a zip file. This is going to be the file that is going to be your file structure. So I'm going to open up that zip file there and we're going to look at it. And so it created this folder right here in my downloads, executive dashboard demo. And if we look at this file structure right here, you can see it's the architect builder folder structure with all the appropriate documents that you need from architecture, validation, planning, decisions, domain. Again, these are all kind of placeholders. And your architect is going to instruct your builder codeex in this case or claude code to fill these documents in as you go. So what do you do with this file structure now?

[00:13:13]Well, in this case, I have a project folder on my desktop and I got a folder in here called demo builds. I'm going to open up that folder. And then I'm going to go ahead and take this folder that I just downloaded and unzipped in my downloads folder and I'm going to move it into my project folder. And now I'm set. This is the project. When I go into my builder, I'm going to start building from this folder. So we did that first part. We downloaded the zip and we put it into our project folder. And then this is everything we've been working up to. So this is the architect starter prompt.

[00:13:51]All right, here you can see that it has taken all of our input. It's going to say where are you going to build it?

[00:13:59]You're going to act as the architect layer and you're going to this is the project name, the client, this is the current status. This is the methodology requirement where you're going to do anything. It's all very detailed in here. It's all getting set up. And this is the critical prompt that you're going to want to put into. We did all this work so that we could get this and we can load it into our uh architect.

[00:14:25]Before we do there, I just want to show you you don't have to copy any of these, but it does show you all the documents.

[00:14:30]If you want to look at them, a lot of them are empty kind of placeholders because as the project gets built, this will get populated by your builder. But at least if you wanted to explore these, you could, but they're already in the folder structure that I showed you. So now I'm going to go back up here and I'm going to copy this architect starter prompt. And essentially for the moment we're ready to start the conversation with the architect. So I'm going to bring my architect into play because we're actually going to start having the conversation with the architect. So we're back into our project this case demo for YouTube with our sources with our custom instructions that are set up.

[00:15:10]And we're going to start the initial conversation. And I'm just going to paste that starter prompt and we'll let it do its work.

[00:15:18]Okay, so that took about 48 seconds and I want you to look at how powerful this is because we set everything up with the custom instructions, everything. Look how nice this is. It's saying do not generate architect pack 001. And remember that is the goal of the architect is to create packs. Packs are a detailed set of instructions. Okay, it'll make sense here in a second. So the intake is strong enough to understand the business problem. the workflow and everything else. But look what it says here. But it is not complete enough for a builder ready sprint. And this is why I left some of those things blank on purpose because I wanted to show you this that your architect is going to save you because it's set up. It's saying we still don't have a systems tools involved, the data inputs, the MVP for the smallest usable next sprint, the dashboard metric definitions. So there's a lot of decisions that need to be made. He did a quick sanity check. He read those two files that I uploaded, right? My two demo files. And it says a recommended project direction for this demo. I would not overbuild this. The recommended direction is a local upload executive dashboard demo. This fits the business story. And then look at this possible project directions. Option one, executive dashboard demo app. Best choice, a polished dashboard. So on good for YouTube demo. Option two, an analyst validation workbench, more operational.

[00:16:36]And then option three, automated executive report generator recommended minimal viable product boundary sprint to upload validate in scope. And it goes through all of these things. Again, for the sake of this demo, I'm not going to go over it, but look at the detail and as much information that this architect is giving you. This is what you're paying the architect for to give you and help extract that vision and that dream.

[00:16:58]This is what it thinks it's out of scope. Here's what it thinks the key decisions are before it generates that pack generation. So again, you're going to be reading this and you're going to be iterating with your large language model, your architect to get everything right. And so you're taking the time to read all this stuff to make sure it's going to build what you want, what your vision is. If it's not, you tell it and you keep going back and forth with your architect until you get it to the point.

[00:17:22]This is just such great information and so much power because we've set up those prompts appropriately and gave it enough information. Focus discovery questions.

[00:17:31]And it's basically telling me answer these questions. Is this primarily YouTube demo? Is this what is all this stuff? And for the sake of the demo, I'm not going to necessarily answer these.

[00:17:41]I'm going to have him do a lot of the work just so that you can see. But in real reality, I'd be going back and forth and thinking about it and a answering the questions. Here's what he's saying. Unless you override me, these are my proposed defaults. And I'm going to go through these and just for the sake of the demo, I'm going to agree with everything he's saying. But look what it's giving us. The proposed defaults, the project direction, the sprint, the stack, the parsing, the uploads, the recommended style, the output authentication. All of these things are there. Recommended move is basically saying generate the pack with your proposed defaults and then we'll go ahead and do it. For the sake of the demo, I'm okay with this just so that we can get on with it and show it. And I'm just going to basically copy this recommended default because remember the goal is here is to generate and get a pack for the builder a set of detailed plans so it can start executing. So let's see how it does.

[00:18:38]Okay, you can see that took 2 minutes and 42 seconds and essentially what it did is it created the architect pack 001 discovery. Perfect. So, what I'm going to do here is I'm going to download this file and I'm going to download it to a specific place. I'm going to download it to that project that we created, right?

[00:18:57]I'm going to go to my demo projects folder and I'm going to go to my demo builds and I'm going to go to my executive dashboard demo, which is the build, right? This is the project that everything's in. I'm going to open this up and I'm just going to put it right in here into the root directory of that folder. Now, if you want to look at that, there's a couple ways. I just want to show you something here. I want to show you what that architect pack looks like and how powerful this is. If I go to my project folder, and there's the project that we're building, our build, and I open it up, and there is the architect pack 001 discovery. And I can open that up and I can read that right here. But I'm going to show you a better way to look at this. I use a tool called Obsidian. You don't need to use Obsidian. Obsidian is just a way for me to look at this. So it's the same structure, but I'm going to show you how it looks a little more user friendly. I can read the markdown files in here. So here is the exact replica of that folder structure in Obsidian. I call it a vault. And here's my demo builds. And here is my executive dashboard with all the docs in here. And then right here, you can see this is my architect pack one. And now I can read the contents.

[00:20:16]Now look at all the stuff that is inside architect back one. It's basically saying this is what you're going to populate state markdown file with the project state, the current status, the business goals, the inputs, the target MVP, the current known restraints. It's going to populate your decisions. This is that file structure that you hear me talk about. This is what keeps this is the brains in the heart of the project.

[00:20:39]And just look at this detail and organization what's in the domain like what this is about. And again, it's all because of the conversation we had and the inputs that we put in to the project launcher. Again, I would review this just to make sure everything is where I want it to be. These are the risks. And again, you may agree or disagree with them, but this is why I read all these things. questions, markdown, inventory, all this stuff is going to get populated into your file structure over here. Like all these documents are going to get populated by the builder when we tell it to. And so this first architect pack, this is what is inside of it. We'll look at the second architect pack, but again, this is just the sprint, the readiness, everything. This is just awesome. This is why I love this format. This is why your projects can get so organized and be so powerful. But anyway, I just wanted to show you what was the content of that architect one pack. Now, before we start with the builder in codeex or cloud code, I'm going to ask it a couple things. How many sprints do you envision? This just gives me a sense of how big this project is going to be. You don't have to do this, but I like to do it because particularly in this early stage, I like to know. Again, you don't have to be limited to these sprints. You can do a lot more. But you see right here, he's basically saying this is going to be five sprints and our project's going to be done. And the real NBT has three sprints and keep it lean, but this is the path. Discovering architecture, upload, validate, parse, and join, executive, dashboard, UA, demo, and polish, and then an optional sales demo layer. At a minimum, we're going to do three. So that gets me the functional demo. So he's basically saying, I'll do four sprints. Okay, that's cool. I'm gonna ask it one more thing, and I'll just do it right here.

[00:22:22]after each architect sprint, can you give me a initial starter prompt for the builder? Uh the so basically the dryr run prompt, the apply prompt, and then the execution prompt. That would be really helpful. Essentially, what I'm asking the architect to do is since he's given me all the files to give the builder, give me the note that I can give the builder to execute this. So he's saying yes, this should be part of the standard rhythm. A dry run prompt, an apply prompt, an execution prompt.

[00:22:53]And we do these three things at the same time. The dry run is basically verified the pack can be applied safely. The apply prompt is actually apply the pack to the project folder. When you say apply, it's going to populate your file structure, which is what you want. And then of course the execution prompt where they actually start building and writing the code. Okay, so for this project, here are the three. So he gives me the dry run prompt. I'm going to copy it. And now we're going to go into our builder. In this case, I'm using codeex.

[00:23:23]You can use cloud code. I made this the dark theme just so that you can know the difference. When I'm in the dark theme, I'm in my builder. If in the light theme, I'm in my architect. And so now we need to do a new chat. Okay, we're starting a brand new chat in our coder.

[00:23:40]But we need to make sure we're in the right folder. If you notice here in my new chat, I'm in another project. I'm in this project over here. I want to be in the project that we just created. So, I'm going to add the new project and then I'm going to navigate to where our project was. It was in demo projects. It was in demo builds and it's executive dashboard demo. I'm in the right place.

[00:24:02]And you can see right here, it's kind of faded out, but it's architect pack 001 is what we want. And so, I'm going to open this. And now you can see I am in the project executive dashboard demo.

[00:24:14]I'm going to paste the dry run prompt and we're going to start and go. And now while it's building over here, you can see it's starting a chat. I'm going to rename this and I'm going to rename this sprint 001 discovery and then save it. And then you can see over here just for organization. So now each sprint that I do, and you can look at my old projects over here, how I have all these sprints. Here's 31 sprints for this project that I did right there. So it just keeps it organized. And then I'm going to minimize this over here so you can see more. And you can see it's going through and it's basically reading the plans and it's saying, can I do this?

[00:24:58]Okay, that took about a minute and 20 seconds. So basically was saying, hey, I read this thing. Here what was present, the dry run result. No files written just like we want. We're basically just saying, "Hey, read this and tell me you can do it." It's got no warnings, no conflicts, no missing folders. The folder is not a git repository, which I'll talk about in a second. So, it's basically saying this is good to go. If you're new to this and you want to verify the output, you can copy this.

[00:25:21]Then, you can go back to your architect and you could say, "Hey, here's what he said after the first prompt, you know, and just get an opinion from your architect." At this point, I wouldn't do that. I would have been happy. I would just paste the second prompt.

[00:25:37]And he's basically saying, hey, look, the dry run passed. The only issue is path mismatch and it's not a pack problem. It means the active project folder is actually says next. Give him this. And I'm going to go ahead and give him that. And for the sake of the demo, I'm going to go back into here. I'm going to paste this.

[00:25:55]And essentially now this is the application piece. And essentially what you're asking to do is is to apply the pack. And basically what it's doing is it's writing all the in outputs to the folder structure. I'll be back in a second. Okay. And that took 40 seconds.

[00:26:11]And so from that 40 seconds it's basically saying, hey, I wrote all these things. I updated all these files. The planning layer is now ready. And we can go from there. I'm ready to execute when you are. And so again, if I'm not comfortable, if I'm new to this, I can copy this and ask the architect what he thinks. In this case, for the sake of the demo, I'm going to go ahead and put that third prompt in here or the execution piece right here. The builder execution prompt says, "Use this after the pack has been applied." I'm going to copy it. Then I'm going to go back to my builder and I'm going to paste the prompt again, have them execute, and I'll pause and I'll be right back.

[00:26:52]So that took a few minutes. They executed basically all the instructions for sprint one. So we're going to take this and copy it and we're going to go back to the architect just to get a review say you know here is result of codeex and we should get an approval and we can close out sprint one and then it'll start building sprint 02.

[00:27:17]Now, he did come back and say, "Hey, it's approved, but we do have some cleanup to do." So, we weren't quite done. And again, this is the reason why we go back to the architect. He's basically saying, "Hey, this is still sprint one." And from there, we should be be good to go. So, I'm going to copy this. I'm going to go back to the builder, give him the recommendation from the architect, and we'll go from there. Okay, that took a minute and 11 seconds and it's basically saying now that sprint 01 is complete. They did everything they wanted to do. I've reviewed it. I'm going to copy the builder's output. I'm going to go back to the architect and basically tell him he says he's done with sprint one. We'll see what the architect says. And he's saying, "Hey, good. Sprint 01 is closed. Codeex did exactly what we needed. Got everything correct. Next move. Now I should create architect pack 02. And this pack should create actual sprint 02 and planning file. So on and so forth. Sprint 2 should be tightly scoped. Build the first local dashboard. Use this in scope out of scope. so on and so forth. And I've re reviewed this. Looks good. And I say go for it.

[00:28:23]And he'll create architect pack two.

[00:28:28]Now, while the architect pack 2 is getting created, I'm going to go into github.com. I'll do another video on this later, but this is where I'm going to start creating a repo that I can start pushing my code to. So, just real quick, you can Google this. I'm going to create a new repo. We're going to call this demo for CEO. I'm going to make this private. And I'm just going to create this repository.

[00:28:56]And now it's there. And I've got a repo at this location right here. And we'll copy that for later. But I just wanted to go ahead and create that because I'm going to have the builder commit the code there when the time is appropriate.

[00:29:10]So again, I'll be back when architect 2 is done.

[00:29:15]Architect Pack 002 is ready. And so I'm going to go ahead and do the same thing.

[00:29:20]Now I'm going to download this uh Architect Pack 2.

[00:29:25]I'm going to download it to my executive dashboard, my demo folder. I'm in there already. You can see Architect Pack 001.

[00:29:32]I'm going to save Architect Pack 002 there. If I want to look at Architect Pack 002, I can open up my folder structure or I'm going to use in my case Obsidian. And you can see right here I'm in my demo builds and there's architect pack 2. I can read the contents of everything that's going to get updated just to show you that this is what is actually getting created in here. Again, I would review that. And then he's given me this sprint 002 dry run prompt. I'm going to go ahead and copy that. And I'm going to go back to my builder and I'm going to go into my chat if you will in the builder in this case executive dashboard demo. And I'm going to start a new chat. So a fresh chat with new context. I'm in the right folder. I'm going to paste the prompt and then we're going to run this. And you see it created a new chat in codeex. I'm going to rename it and we're going to call it sprint 002. Upload and save. And now you can see I've got the organization right here with this current sprint that I'm in.

[00:30:40]Okay, the sprint dry run for architect pack 2 is done. All looks good. I'm going to go back. I could copy this and ask to validate it, but I know it's good just from based on experience. I like that response. I'm going to do the apply prompt. go back to the builder. So now I'm asking my builder to go ahead and write and update all the files in the file structure with that prompt.

[00:31:07]Okay, he applied everything correctly.

[00:31:10]Everything's been written. These are the documents that have been updated. Sprint 02 is planning. He's ready to execute.

[00:31:17]I'm going to go back to my architect and I'm going to grab my execution prompt.

[00:31:22]And I'm going to go back to my builder and basically tell them to apply. Again, you could copy the output of your builder and ask the architect to verify it, but again, I've done this enough.

[00:31:32]I'm okay with what we see. And now he's going to write the code for uh architect pack 2.

[00:31:40]I misspoke. He's actually not writing the code. He's actually saying this is what I'm going to write. Again, I do one last approval letter. He's basically saying this is his understanding. This is his plan of attack. This is what he's gonna do. This is what he's going to code. And then after that third prompt, I'm going to take the output and we're going to get approval. It looks good to me when I read it, but I will ask the architect, you know, here is the output from codeex and then basically get the approval, see what he thinks just to get a second set of eyes on it. And hopefully it'll say, hey, this is approved. And then I'll go back and I'll actually start writing the code. And it may seem like overkill doing this, but some people will say, why are you doing this? Is because this is a small project and you probably don't have to do it exactly like this, but when you're getting big projects like this, it's important to keep it in that chunk. So basically says codeex plan is solid. Approve it. My call in the open question. Manny upload. Do not add a load sample data button yet. I agree with that. That's fine. And again, I am going through this fast, but normally I would be reading this, but look at all that what he's recommending.

[00:32:49]It looks good to me. This is the right next move. Basically, what he's saying is he likes CodeEx's plan, and I'm basically saying, "Approve it. Go for it." And basically start writing the code. Now, while it's doing that, I wanted to show you what was getting created. I skipped over it, but these sprints are getting created by the builder. And if you go look at my folder structure and you look under planning, you'll see a folder called sprints. And that might be kind of small, but notice right here, if I can make this a little bit wider, these are the two sprints that we've done. And inside each sprint, you're going to have four folders, an acceptance criteria for what sprint one is. You're going to have a blueprint.

[00:33:32]This is the actual plans that your builder is using. It's actual blueprint.

[00:33:38]This is the handoff prompt that we actually used and gave to the builder.

[00:33:43]And here are the requirements. And so this is what makes the code when he actually writes the code makes it so powerful and so clear. There's no ambiguity about what he's supposed to be doing. And so each one of these sprints gets these same set of documents. And this is where again why we do the back and forth. Basically when I do the dry run, I'm like, "Hey, this is what you're going to build. Do you agree?" And then he agrees. And then we say, "Okay, write it and create these documents." He does.

[00:34:11]And then we basically say, "Now go read those and execute off of that." And it may seem like overkill, but trust me, I don't get near the bugs. I don't get near the logic flaws because of doing it this way. And this is about the only way that you can handle really complex projects. I know this is a simple one, but this is what makes this so powerful.

[00:34:31]Okay. So, how did we do? He said approved. I confirmed he's still working right now. So, I'll come back when he's done.

[00:34:40]So, this was an interesting result. This was actually taking a long time. It took about 8 and a half minutes and something was wrong. And I actually stopped the coding and I said, "Hey, what is going on? Something is wrong." And he said, "You know, you're absolutely right. This did take me too long." I started going down a rabbit hole because I was trying to open up a browser to look at what I've created and then I started debugging and I didn't tell you. And so this is interesting. I'm kind of glad that this happened. So I said, "Hey, give me a status of where you're at."

[00:35:12]And it gave this, "You're right. I spent too long on verification. Here's the clean status." But he's got a critical error here on the severity and the spreadsheet, the vulnerability with no fix available. So, this is a perfect thing to take back to the architect so he doesn't go down a rabbit hole. We've got a problem. Let's go back to the architect and say trouble.

[00:35:38]Here is his update.

[00:35:41]And I'll paste it. And this is good.

[00:35:44]Again, this is why you kind of got to watch what's going on because there's some issue there with the file that we gave him, the demo file. So let's see what our architect says and what our next plan of action should be. And he basically said, "Good. This is not a disaster. Sprint 2 is partially implemented but not accepted yet. The next move is not doc's closeout and not sprint three. The next move is a narrow sprint 2 debugging verification path."

[00:36:05]And we've come up with a prompt to give him. So he says his read is the two warnings are not blockers. It's a large JS chunk expected because that spreadsheet was heavy and an XLX audit warning worth documenting but not worth derailing sprint 2. So the blocker is only this the browser upload verification because it showed a blank page. Okay, that's good. Fix that and then we can close sprint too. So I'm glad that that happened because these are the type of things that can happen.

[00:36:30]You think, oh my god, this is impossible. That's why we use this kind of second set of eyes so that we can keep the builder focused. And I had to re rein the builder in because it started going down rabbit holes. So let's paste the kind of the architect kind of recommendation and let's see how he does. Okay, coming back. found the root cause without importing. So that helped going back to the architect. The npm run build succeeds working now manual upload so on and so forth. But we still have some known limitations. Still reports that bundle size warning and no closeout docs are updated. But the sprint 2 acceptance criteria are now met from the runtime browser. So it was good that we went to that. So I'm going to take this output again. It shows you how things aren't always perfect, but going back to that architect helps solve the problems. I'll see what it says about its final output. And he says, "Hey, it's good. It's functionally accepted, but we're not quite there. We got to get the closeout documentation prompt." So, no new features. So, it came up with one other thing to try to fix what he was having the problem with. And again, this goes to the shows you the power of this architect builder method. So, I'm going to copy what he has right there and go back to the builder and see if we can finally get this closed out. Okay. So, it closed out sprint 2 recommended it. I copied his response. I went back to the architect and he said, "Good. Sprint 2 is officially closed. The app now has a working local VI react dashboard." And so maybe I'll take a peek at it here.

[00:37:57]But then we can go ahead and start sprint three. I'm going to go and do a couple things here and then I'm going to pause this and I'm going to complete this and show you the end result here in a second. One thing I would do is we we've usually after every sprint I commit to GitHub. Remember I showed you the GitHub? Since we're closed out, I'm going to end this sprint two, I'm going to say commit and push to get. And just for the sake of it, I'm going to go ahead and give him my GitHub repository.

[00:38:29]And remember, we created this in here again. So, I'm pushing the code up to the cloud into this Git repository. So, I'm going to copy this and then I'm going to give him that link.

[00:38:43]And then he should go ahead and commit that code. Okay, the git has been pushed. Just to show you again, if I look at the code, if I look at the repo, I've got one commit here and it's got the complete sprint O2 dashboard demo.

[00:38:59]So basically everything that we've done up to this point, this is a good starting point. It's like saving. And so if anything happens and we get kind of cattywampus or sideways, every time we get a sprint done or major milestone, I commit. And so we can have a save. We can always fall back to a known entry point. So again, more videos on that later, but I just wanted to show you that I committed there. And then back at the architect, it looks like, hey, I'm ready to generate pack three. Just say when you're ready. And then I'm going to go ahead and pause this and I'll complete all the remaining sprints and I'll show you the final output here in a moment. Okay, just real quick, I just wanted to show just to remind you case we got Architect Pack 3 ready. Just to remind you of the process, I'm going to go ahead and download the architect pack 3 into my project folder. We've already got architect pack one and two. I'm going to save it there. I'm going to copy the dry run sprint prompt for sprint 3. I'm going to start a new chat in my executive dashboard demo by clicking on this. I got a new chat. I paste my prompt and then I'm going to copy this name just for my organization and we're going to run this and I'm going to rename this and we'll call it sprint and then paste 003 demo polish and readiness and away we go. Okay, I'll now I'll pause here and then I'll come back. I'll have everything complete when all the sprints are done.

[00:40:35]Okay, we can pretty much call it complete. We actually stopped at sprint three. We could have done sprint four. I was talking to the architect, but we're pretty much at a spot where we can look at this and demo it. I want to open it up in an external browser here. And I will bring it in.

[00:40:52]And this is the final output just after three sprints and about an hour's worth of work. Again, this is simple. It's asking us for an input to choose our order export and we will upload that there. And then our CSV file. Again, this is I don't know how practical this is, but it just shows you. And then to analyze the dashboard and we've got some blank things reviewed, but we can still look at it. And then here's the validation, the net revenue, the refund rate. So again, we got a nice little simple dashboard and just after three sprints and about an hour's worth of work. And then also we committed to GitHub which if we look right here I can refresh and I've updated until I got another commit in there. And so again we would just keep building. So again we only did three sprints and we can look at our folder structure again. And you can see we've got the three sprints there. One, two, and three. And each one of those has and then I don't know if you can see that very well. Each one of those has their acceptance blueprint, handoff, requirements, everything. So again, just a nice way to stay organized. Again, if you look at there's all our architect packs that we did for this. And if you look at the state of where we're at, again, it's updated. Where we're at, the completed sprints, we did two. The current sprint is three. When I close out the sprint, this will get updated.

[00:42:17]So again, just a nice way to stay organized. And hopefully you can see from the example by using the project starter we were able to do this and have this level of organization. And so again we started with the project launcher and using the project launcher allows us to basically prime the architect with the right prompt and with the right knowledge base. If we set up the workspace, remember we put the architect builder method, the markdown file in the reference folder for the project and then we also set up the custom instructions down here so that it responds understands how to act as the architect. When we primed the project with that, we also put our reference styles within the project. And then when we started the project, we made sure that we put all the input in that we needed and gave it the context. And then we downloaded a zip and then we put that zip folder into the project folder that we have on our desktop. That gave us our file structure. And then we got this architect starter prompt. That architect starter prompt allowed us to start the initial conversation with the architect.

[00:43:24]And away we go when we build the packs and start our project. So you can start using this project launcher if you want.

[00:43:30]Again, for small projects, big projects, you can get it now with the link down there. I got an introductory price for the first 50 lifetime access. It's 149 bucks. After those 50 sell out, it's going to go up to another 25. At 249, and then after those get up to 50, I'm going to do $49 a month. So, you get lifetime access, no monthly subscription. You can get it forever.

[00:43:51]I'm going to be updating this as we go.

[00:43:53]And so, let me know if you got any questions, and I hope to see you on the inside. Use this architect project launcher. And then you can also get more information about my builder cohort where I can work with you one-on-one to help get you started. And you get this tool as part of that as well.