Introduction to REW AI Agency

Today is a bit of a continuation for anyone who was on last time. We covered, like, Cloud Design and how to hack your real estate webmaster's website and be able to do your own custom design elements without having to use our agency. We're getting ready. I have a big announcement to make here.

It's July fourth for our American friends, and our Canadian friends get it early because our birthday is on July first. But our next release is Independence Day. Right? It's independence from having to know how to be a server administrator, having to know how to be a designer or a programmer to be able to work with your own website, do your own marketing, things of that nature.

It's the official launch of REW's AI agency.

And so my goal with these coaching sessions is to help people figure out where exactly they fit in terms of their relationship with AI. Right? Because, fundamentally, anyone who fits into REW's, you know, ideal customer profile, whether they're an innovator, whether they're an optimizer, or whether they're, you know, an attractor, they're going to wanna use AI. And some people on this call, some people not on this call who are watching the replay, you will be able to do all of this, everything I'm showing you. All you need is some coaching, some guidance, especially as it relates to how to interact with real estate webmasters because that's information you need to get from us. But you'll be able to create your own servers. You'll be able to build your own applications.

And for those people, I'm excited to build things with you and to learn from you because the cool thing about AI is it's impossible to be an expert right now. Because if you if I'm an expert today, I won't be by next week because everything will have changed again.

That's the exciting thing for people who are here to learn how to do it all. But then there's gonna be gradients. There's gonna be levels of people who can do some of this stuff, but they need a little help. Maybe once you've got Cloud Code set up, maybe once you've got cloud design figured out, you'll be able to use these tools.

But maybe some of the technical things like setting up your own Google Cloud environment and managing the security and stuff like that, maybe that's beyond your comfort level or your technical capabilities or just you don't have the capacity for it. And then there are some people who know they need to use AI, but they don't wanna spend the time to figure it out, and they would just rather pay somebody to do it for them. And so that's the big announcement with the AI agency is that we now do everything we used to do, except we have access to AI to do it. And so that means implementation people.

If you're one of those people who, you don't wanna learn any of this stuff, but you certainly want more pages faster for cheaper at higher quality, Great. You can just hire the implementation people, and they'll just go ahead and they can do twenty pages in the time it used to take one. So there's just massive advantage, and it's gonna be high quality. Right?

For those that want custom solutions built they wanna participate in the process, but they can't lead the process and don't wanna build truly custom software, they can hire the agency. They can work with their favorite programmer of yesterday, tomorrow. And the only difference is that programmer now is fully enabled and educated to have AI tools. Right?

So, yeah, I'm just I'm really excited for the Independence Day thing. Again, if you wanna learn, I'm coding every single night. And if you if you're on my Facebook, if you're on the REW forums and you just you're in there and you're a mad scientist, I would love to learn from you, and I would love to teach you. And then, again, all the way to anyone who just wants to take advantage of this stuff.

Introduction to Claude Chat, Cowork, Code & Agents

So yeah. So last time we talked about Cloud Design, and I wanted to get to Cloud Code because that is probably the most powerful and most used tool outside of the Cloud Chat environment.

But I also get a lot of questions about agents, like, not real estate agents, but things like OpenCloud and things of that nature. And so I wanted to talk a little bit about the difference between what an agent is versus, like, you know, a chat versus Cloud Code. And after we go through some use cases with the agent, which is gonna be Cloud Cowork in this case, I'm actually gonna show you in real time a software I started building last night, and I'm gonna show you how I build software.

And I'm gonna show you how I interact between Google Cloud Shell, Cloud Code, as well as Cloud Chat. And so we're actually gonna try to build a feature today in real time so you can kinda see what's possible and kinda see what's going on. So So can you guys see, a Claude chat right now, something about building EOS tracking something something? Yes.

Okay. So, fundamentally, we have three tabs up here. Right? We have chat, co work, and code.

Now most of us know what chat is for. Right? That's how you engage with an LLM. It is a question and answer type conversation.

LLM's large language models such as Anthropic, which powers Claude, have a whole knowledge base where they already know everything about everything, but, you know, word of caution, that knowledge base is somewhat out of date. But they also have the capabilities of going and doing research and checking so you can say, hey. Go update your knowledge base. I need today's answer for x y z, and you can have a conversation with chat.

Explaining Claude Cowork & It's Best Use-Cases

We're gonna spend a little bit of time with chat later when we do the Cloud Code, but the next thing is Cloud Cowork. And so what Cloud Cowork is, and for anyone else who uses OpenClaw or any other kind of the difference between a chat and an agent is autonomy. And so with a chat, you ask a question, you get an answer. You have to drive the conversation.

But with Cloud Cowork as an example or any agent, right, you can establish a relationship. Think about it like hiring another employee or a new employee. You establish a relationship. You sit down the guidelines.

You talk about, you know, what it is you like to accomplish, how you like things done, etcetera, but you don't get explicit about how they go about the work, and you focus on outcomes as opposed to instructions. I've gone ahead just to give you an example of how I might use Cloud Cowork. Okay? So I I live in the SEO world.

SEO is one of my passions.

And so let's say I want to do a better job as a marketer talking about SEO or keeping up to date with SEO, etcetera. So I spent some time in my chat first developing what kind of instructions or relationships, what kind of prompt I would wanna have for my agent, for my new employee who's gonna help me stay on top of this. Right? And so I've created some instructions.

You are running Morgan's daily search engine optimization briefing. We wanna apply the project's two thousand twenty six SEO, AEO, GEO rules, something I've already written, a bible about SEO for this year. I've already done all that research. We take all of that in first.

That's the knowledge base at the beginning. I've taught my employee how I think about SEO and all the things. And then I wanted to do this for me every single day at six AM. Right?

Step one, I want you to go to the web. Use tool search. Find articles published recently about traditional SEO as well as answer optimization, all the things. Go out and do the research.

I'm not telling it where to search or even how to search. Just go to the research. I want you to bring that back, and I want you to rank and summarize those. Pick the best six articles.

Rank them by value to me based on what we've talked about in terms of what I care about in SEO and all this sort of stuff. We've already had this conversation, this training session. Right? I want you to write two to three sentence summaries, standout stats, concrete talking points.

This guy is out there, this new employee. He's out there doing my research for me. All of the things that I would do in my morning if I had all the time in the world and I was writing about SEO, he's out there doing those things for me, right, or if I had an assistant to go gather that information for me.

Then I want them to come back, and I want them to pre-do my blog structure for me. I don't want them to write me a blog because that actually violates one of my rules about SEO. Because, by the way, spoiler alert, if you watch one of the other classes from this year, I do not believe in AI doing your SEO. In fact, I believe that is one of the ways that as a real estate webmasters customer or coaching student, I believe you'll learn to stand out in that you use AI as an enablement tool, but you will not use it to replace the quality of your authority and experience.

Right? But he can come back and give me the structure. So he can say, yeah. Here's how I think all the h one tags should be, whatever.

Here's what you need to fill in. Here's some bullet points.

But my intention is to use WhisperFlow, which is what I use to talk to my computer because it's faster and it's better and whatever. I have my bullet points on the left, but I'm not going to use those bullet points. I'm gonna open up my blog on the right, hit Whisper Flow, and I'm gonna use my own words to expand on those bullet points and to share my perspective on it. Right? So I've got my layout. I've got my outline. I've got everything.

And so it's all about creating an easy button where you can lay out all these different things. This is just an example of how you might use an agent. Right? So you establish the relationship once. You establish the goal post, the guidelines, and then you can schedule it.

And so another cool thing about Cowork or about any agent is you can give it disparate tasks that require it to connect to other services, and you can connect those services. So I'll give you a simple example. Let's say you're an iOS user like me, and you want to create an iMessage relationship. Right?

You want to use iMessage, which is the native texting app for iOS. Okay? When I'm driving, I get some of my best ideas. And so I built this relationship with my agent whereby we were connected via text.

And so when I'm driving, I could just click a button, and I could say, I want you to text Rupert, please, and then I would just say whatever I wanted. And that would go through iMessage into my agent. He would interpret it. And if there were instructions, it would be ready on my computer when I got back.

Right. So that is a wonderful use case for an agent. Now you have to set up permissions. There's work to do because you have to make sure you set up permissions.

You have to set up specific folders, and this is critical, by the way, not just for you OpenClaw users, but for your Cowork users. You should not have a machine where Cowork or Cloud Code has full access to everything. You should have specific folders that it has permissions for, and that's it. And one of the things I learned, not through any negative experience, but just through research, is you also should not do iMessage on your regular iMessage machine because then Anthropic technically has access to all of your texts.

Okay? And so the best practice, if you're going to be doing things like using iMessage, you should actually have a separate machine. A lot of people will use a Mac mini, which is very affordable. You can set up a separate number, a separate everything so that that particular computer, that particular machine is not accessing any of your sensitive financial data or files or anything like that.

So, hopefully, that's a pro tip for you.

But, yeah, I mean, you can sync it to your Google, your calendar. You can sync it to your Gmail, your assistant. It's just like if you hired someone. You would decide.

I trust you this much, so I'm gonna give you access to this, this, this, and this. And once you have access, I'm gonna treat you like a qualified professional who has a great background in x, y, z, and I'm gonna say these are the outcomes that I want. These are my needs and when, and I'm just gonna let you get to work. And so that's a really exciting thing about co work.

The other thing I wanna talk about in co work, and then we're gonna get into code because code is cooler. In my opinion, agents are cool. Agents are fun. But agents can really teach you about agents and help you set all this up.

There's one other element. Has anyone used dispatch, the new beta?

Okay.

So Dispatch is really cool, especially because of that scenario where we're talking about the Mac Mini, where you now all of a sudden have a different computer. And so what Dispatch does is it basically allows you to remote in so you can sync your phone via the Claude app to your regular computer, to your Mac Mini, etcetera. And so let's say, for example, you worked on a spreadsheet locally on your computer and you don't have it on your phone and you don't have iMessage set up, or you want it to be uploaded to your drive instead and the password for your drive is already logged in on your computer, you could be like, I need you to go grab that file I was working on in spreadsheets, and I need you to upload it to my Google Drive, please.

And it can do that. It's like remoting into your computer, which is super cool and super handy. Definitely recommend taking a look at that feature, especially if you don't do everything in the cloud and you do need access to your physical files. That's just yet another really cool example of where to use Cowork.

Exploring OpenClaw and Its Implications

So, yeah, I do wanna talk about OpenClaw. OpenClaw, for anyone who doesn't know, is the most downloaded agent that is not associated with a particular LLM. Right? And it was developed by this guy who just wanted a better way, who believed in open source.

Okay? So in our programming world, open source is a wonderful gift. It means that anyone can contribute, anyone can access the code. It's shared.

It's nonproprietary, And that makes it incredibly powerful in terms of how fast the features develop and how robust it gets and things like that. But at the same time, one of the lessons I learned about OpenClaw is that because it's open source, there are a lot of concerns about continued vulnerabilities. Meaning, you have it downloaded on your local machine, there have been many instances of hacking, many instances of bank stuff, and everything like that.

And, candidly, even though I was having a lot of fun with it, I uninstalled it for two reasons. One was the security concern. I have a lot of passwords. I have a lot of stuff on my computer.

I cannot risk even one percent security breach. But the other one, candidly, is that a lot of nerds have built a lot of cool stuff without necessarily taking into account the consequences. And so the tool being extremely powerful doesn't always consider your budget. I had moments where two hundred dollars US would get spent in an entire hour.

But the reality is I played with it for quite a while, and the output was awesome. But compared to the price I paid to Anthropic directly to use Cloud Code or to use Cowork, it was, like, ten times more expensive.

So my two weeks of OpenClaw have told me that I don't wanna use OpenClaw, but I did do some more digging because I know people really love that autonomy of not being tied to one particular LLM.

And so the recommendation, which I have not done, is, first of all, don't put it on your computer.

Host it as as a hosted application. So you could use AWS or you could use Google Cloud. But because OpenClaw is open source, what you can do is you can point Cloud Code. This is a great use case to get into our Cloud Code.

You can point Cloud Code at that folder, and you can say, I love these features about OpenClaw, and I wanna replicate them but in a more secure way. Right? And it can actually access and reverse engineer all of that source code, and it can build you your own agent that has all of the cool features you want, but none of the bloat, none of the extras. And so that's the recommendation if you wanna go forward with an open source agent and build an open source agent.

It's not very hard. It's not very expensive to have a small server at AWS or at Google. I recommend Google because there's a lot of built in plug ins. If I was to go that route again, I I gotta be honest, though, I'm super happy with Claude, so I'm kinda sticking with it because it's it's the easy button for me.

But if you do wanna go crazy and build your own version of computer or perplexity or whatever, you can do that by spinning up a basic server. It'll cost you less than a hundred bucks a month, way less, to be honest, unless you're crazy. And then you can just run your own agent away from your computer, away from all the security risks, etcetera, and just build the features you want. Build just the connector to iMessage.

So you got a lot of choice there. Okay. I wanna move on from agents because I wanna get into Cloud Code, and I wanna get into sort of how I build things personally as the visionary here at REW.

Building Production-Level Software with Claude Code

So I'm not gonna lie. Some of the conversations in this bit are gonna be a little bit technical. It's not to scare you or impress you or anything else. It's just to make sure you understand.

To get to this level, there's a lot of work, but the work is totally doable. We're here to teach you how to do it. You don't have to pay us how to do it. But I'm gonna explain to you what the setup is here, and I'm gonna show you the software I'm building right now.

When I decided to start actually building true production level software, meaning not using Replit or Base forty four or Lovable because, by the way, all of those platforms are absolute garbage when it comes to production level. You can't build what they say on the Internet is not true. You cannot build a production ready application that will be reliable, that you will own, that will be you just can't build a business on base forty four. Base forty four is awesome for prototyping. Just because you can spin something up really quickly and it looks great and it seems to match your vision does not mean it will be reliable. I recommend doing that to see whether or not you like doing this kind of stuff. But when you wanna get serious, you need your own servers.

Okay? And so this is the part where, again, the most technical people, you could do this by yourself. I did this by myself, but I've been in technology for twenty two years. And I still had Richard, who's my Google Cloud certified engineer, available and chatting with me even though I was doing it.

Right? So whether it's AWS, which is perfectly fine, or Google Cloud, those are your two options. Microsoft sucks. But I would choose Google Cloud because there's a lot of native plug ins.

Okay? There's a lot of things that Google Cloud just does automatically to be friendly with AI. It's why I chose it. And so my first decision was I need to host my own applications.

I already went through and tested all the platforms, and they all failed when it came to production level software. And so it takes probably six to ten hours to truly set up your environment. Right? The basics are pretty easy, but we're talking Google Cloud servers.

We're talking about SSO. We're talking about identity servers, the connectors that need to happen. You have to set up your DNS. Right?

You have to point your DNS to a domain. You have to set up the DNS for the various subdomains, all the things. There's a bunch of setup. It's about six hours, eight hours depending on how fast you are and what you know.

It's a pain in the ass. I hate it. It's not the kind of work I like doing, but you only have to do it once, which is nice. And then later, when you're building stuff, you only have to do small steps, like setting up another subdomain or whatever.

But let's just say we did all that hard work, and now we're here to where we have what's called Google Cloud Shell. So Shell is where programmers like Richard, they actually put in true command line code into a server. Okay? I don't know any of it.

I am not a server administrator. I couldn't write a single line of Shell code.

That's why I need Cloud Code, which is what we're gonna talk about today. Because Cloud Code is a Google Cloud certified engineer at the top level out of the box. So it saves me from ever having to learn anything about this stuff that I've avoided for the last twenty five years, and that is figuring out how the hell to write all this command line stuff. I don't know how to do it, and I don't want to. I hate it. Right?

But I do wanna be able to build stuff on servers. I wanna build my own applications. I wanna test them. I wanna debug stuff.

And so we'll skip all the highly technical stuff. We've got Cloud Code installed on Google Cloud. And so what I'm building today, I'll tell you, I started last night at about seven PM, and I spent about three hours on it. This is where we wanna get to.

Right? We have an idea for a software. We just wanna build it. And so in my case, we're running the entrepreneurial operating system, EOS.

Is anyone familiar with that?

It's a great framework. I'm taking my leaders through it, etcetera.

And they have all these tools that are great, but their stupid website is broken right now. The first thing they have you do is, like, this organizational checkup. Right? And you can't save it. It's broken.

Like, well, that sucks, and it has missing a few features. I'm like, why don't I just build that? That would be a fun project for my coaching session today. So in less than three hours, I've gotten here already, and we're gonna continue building it.

So what I did was I had my chat with Claude first. So to be clear, there's no coworker here. We don't need an agent for this. We go from chat to Cloud Code.

Okay? So I had a conversation before I started building, and I always have a conversation before I started building with Claude. I said, hey. What do you know about the EOS system?

And he was able to say it's a well known framework. Here's all the things. I was able to correct a few things and say, please go here. I pointed it.

So now my Claude knows all the things it needs to know about EOS because I wanna run it properly. I said, cool. Now I wanna build this tool, but I don't just wanna build this tool. Because if I'm gonna build this thing, I wanna run all of my EOS system on it.

And so we started asking me questions, and we talked about it.

And we started with, let's fix that whole organizational checkup thing. Right? And so we wrote a prompt. We did some iterations.

We went back and forth, and this is where we are. So I have it. It's here. It's a reliable, tested software where I can go and do my own organizational checkup, and I can check on my own results.

I even have another employee checking on results. I can view those results. I've got reports built, everything. I needed a software for this where it didn't just give me the report. It allowed me to save it. It allowed other employees to do it.

And then we got to dig in. I've got copy to clipboard. I've I've built the download PDF things. Right? These are just things that I just had ideas for as I was building it. Okay?

And then through that conversation, it was like, okay. Cool. What about anonymous versus non anonymous? We wanna be able to make employees feel that they can give this feedback without having to raise their hand and say who they are. So but some people wanna say who they are. So, like, we built all these cool tools. Right?

And it's just it literally is that fast.

And then I was like, okay. What do I need to think about next? Well, I need, EUS has scorecards. EUS has issues lists.

EUS has a VTO, a vision traction organizer. So I'm literally gonna build an entire enterprise software for my entire company. Every single employee in every single department will have access to this tool. This is the kind of stuff that you just come up with an idea, and the next day, you can have version one or the first feature or whatever completely built and test it.

Let's go let's go check this out. Let's go build it.

So I wanna start here.

So, do you see the user screen? Yes. We can. Okay. Eliana, you know me pretty well. What do you think I'm gonna feel how do you think I feel about the screen?

It's very boxy. It it doesn't line up. It's not, You don't like it. It looks like ****.

Yeah. Right? So it understood most of the functional requirements, but it made it look crappy. Now I am not a designer.

We all know that. Like, I can do basic CSS and HTML, but the amount of work it would take me to fix this, even just lining up these stupid boxes and I'm sure a lot of us are like that. Right? It's just frustrating.

We'd never wanna do that in the old world, but this is the point about Independence Day. Right? We now get to be a designer in one line.

So I'm gonna go back to my chat. I'm gonna say, let's fix the UI issues on user creation.

K? Still can't see it. No. I know. I'm just I'm just talking you through it.

So what he's gonna give me back is he says the user page needs UI polish, the form fields, and I we we just recorded that. So the user fields are too small and unstyled. The labels are throwing off the alignment, and the overall form looks inconsistent with the rest of the app. Please fix the styling and make it look clean.

Now I talk to my chat. I don't use Claude code as a chat because I don't like the output. It's way too technical. I don't like the way the UI looks.

So when you're in Claude code, it looks like this, and it's it's terrible. Right? I hate it. But I love the way my chat looks, so I actually have the conversation with the chat, and the chat tells me what to say to Cloud Code.

First, find the, eos dot rupert dot com app. That comment applied.

So I forgot a step. I'm loading up this Google Cloud Shell for the first time in this session. I gotta go find stuff because I got a million softwares going on. Right? But now from that, Cloud Code can go search the whole server. He's got all the access. Hopefully, he goes find all the things, and then I can paste that instruction.

So I don't have to remember what folder EOS is in, any of that stuff. He's just going to work. He's using my credits. He's tokens, whatever.

He's boogieing, which is ridiculous. I do find the Cloud Code UI a little bit ridiculous, but there is a more friendly version if you download the Cloud app and you just tab over to Cloud Code. There's a more, like, visually impressive kinda cleaner UI, from that perspective. But, yeah, he's just gonna chunk away.

He's gonna go and in fact, actually, he's re referencing even though I told I stopped him and told him to go do something else. It looks like he's re referencing the comment I made first. Root cause is clear. Styles has no styling at all.

Great. And so here's where we have some options. Right? Cloud Code would usually, in my case, recommend something and then ask me. This is how you should work with your Cloud Code too. Even if you have auto mode turned on, which I have, I have rules set around my auto mode so that he doesn't take any major actions without asking for permission.

Now I can read this. You know? And he says, hey. Blah blah blah. Add real form controls, whatever.

But I don't want to because I don't like the way this looks, and I don't like reading the white text, and he's too verbose and whatever. So I have my chat. My chat knows me better. And you won't see this, but I'm just gonna tell you.

I'm pasting what he said over here, and my chat's gonna tell me what to say back. Right? My chat said, yes. Go ahead with both ad edits.

I trust my chat at this point because we've been building a relationship and doing stuff. Right? So now we're gonna go ahead with both edits, and he is writing truly to the server now. Right?

He's gonna he he's properly diagnosed that there was no styling at all on that page. And I didn't do that on purpose. We just never built the styling. He's going and doing a bunch of stuff that I could study if I want to.

I can see everything that's being done.

And are you guys seeing me scroll on the code here?

So if I wanted to learn, if I wanted to be a student of design and I wanna figure out what he put in here and how he put in, Right? I could read this, or I could take this, and I could say to my chat, hey. Explain what he did here so I can learn how to do this next time. It's a great tool. We're learning how to do stuff. But for me, we're just doing some visual edits here.

I don't care. I'm just gonna let him go do it. And because we've been building software now for a few months together, chances are I am probably build is clean. Both edits are in styles, blah blah blah.

Eyeball to eyeball it live, run blah blah blah.

Yes, please. Run it. Not on dev. Push it live.

So sometimes I need to ask my chat, or should I go ahead? And sometimes I can just read the last lines of Cloud Code, and this is literally how I build software. Right? I talk to it about what I want.

It helps me build the initial instructions. I push it live. I go test it, and I say, okay. Here's the things I would change about it.

But while this is still going, I do wanna revisit the point. The reason it's so easy for me to do it now is because of all of that work that we did in the beginning. There's a lot of security checks in place. There's a lot of rules and relationships that have been built with Claude over time so that he understands.

Make sure you don't take any go live actions without asking for permission. Make sure you flag anything. Like, there's stuff that you if you're going to be, you know, a prompt engineer or you're going to be building software using Cloud Code, you need to understand that it takes time to build the relationship. Like, you start just natively knowing when to say yes right away and just approve things versus asking questions, and it's great.

Sometimes it takes some time, though, especially when you're actually writing the files.

What shipped? Blah blah blah. Let's go see what he did.

So that was for EOS, which looked like crap before.

And what we were doing, we're adding a new user. How does that look?

Yeah. Did that not just fix every single UI issue we were just complaining about?

This is the world of building software in two thousand twenty six using Cloud Code, guys.

Using Claude Code to Develop an EOS Scorecard Tool

And now this is just UI, but I have another prompt ready. And that prompt is way more complicated, and it's gonna take some time. But I'll explain what the prompt is.

There's a whole like, I mentioned, EOS is quite a complicated system. And so while I chose to build this whole checkup thing first, there are scorecards, there are issues lists, there's vision traction organizers, etcetera. I wanna build my scorecard next. A whole another piece of software.

I want graphs. I I want everything. Right? And so I spent a lot of time talking to Claude this morning, just in my chat, to build a prompt to help me start that process, but I intentionally did not start the process because I wanted to do it on this call.

And so I'm gonna go back to my chat, and you won't see this. I'll just read it to you. Okay.

What is the prompt to start building a report card?

So, again, we talk a lot because there's a lot of questions and answers. So I sat outside with my bulldog in the yard just talking about what my current scorecard does and what EOS says about scorecards and all that sort of thing. And so I've got this prompt ready, and it says, type this. We're gonna build phase two, the EOS scorecard.

Before writing any code, I need you to review the existing code base, then I'll describe exactly what I need you to build. Please start scanning the projects so you understand what's already there. Right? So it's suggesting to Cloud Code that it does a full review of what's in the software right now.

So first, I'm gonna commit this, which is fine, and then I am going to say this. So you want to do this. And, again, as you develop this, if if we were spending time together, like, truly learning how to build software and whatnot, there'd be some other pieces I'd be talking to you about, right, such as using GitHub so that you could store versions of your software. Because one of the things you need to know is sometimes you will go in the wrong direction and have to throw out all of your work, and there's no way to come back.

And that's just software engineering in general. So, you know, for those of you who come along on a journey and learn how to build software, you'll start building those practices over time. Right? I need to have a repository of my code so that I can always roll things back.

Right? I need to have fallbacks. I need to program backups. There is a lot to know if you truly wanna build enterprise level software, and you wanna be protected from losing everything by accidentally hitting delete or writing the wrong thing.

That's way beyond that whole list of everything to do is way beyond this this call. This call is meant to show you just how powerful Cloud Code is in building software once you have it set up correctly.

Because any person on this call can be a software engineer now. Any person on this call can be a designer now. K. He's ready to go. Awesome. Okay.

He's ready. Give me the prompt.

This is what we came up with, and it's not extremely verbose. Here are the requirements. The scorecard is a spreadsheet style view with these columns, department, measurables, goals, owner, weekly date range.

Each weekly call contains an actual result entered by the owner. It meets or exceeds the goal, and it's green. Or if it's below the goal, it's red. Values are either a yes, no, or a number.

The rules are only the assigned owner can enter their own rows numbers. Admins and super admins can edit anything. Individual contributors can only see their own scorecards by default, but a checkbox permission can grant them department wide or company wide view. Admins can see everything by default.

Quarterly design.

Goals can change each quarter, but older data is never deleted. Metrics can be active or inactive per quarter. Metrics that continue across the quarter should support long trending. In other features, the columns need to be sortable, alphabetical for text, high to low, and low to high for numbers, date ranges, on demand graphing, trending using popular great looking chart libraries.

The whole thing should look like a spreadsheet. Do not write any code yet. Tell me what database changes, API changes, and UI changes are needed, and flag any design decisions that I need to make before you start. That took some time.

Right? And it also takes experience to know to ask it or tell it that you want the column sortable. You're gonna build three spreadsheet style looking things before you realize every single time you realize you wanna sort that column. Right?

You develop these things over time. And so I'm gonna go over this is probably gonna take a lot longer. Bet you it takes thirty minutes to build this because this is now a whole new like, Cloud Code is gonna chunk along, and it's gonna have fifty questions for me. And I have to answer them, and I won't have the answer for all of them.

And so what I'll do as part of my process is I'm gonna have to take any question I'm not comfortable answering. I take it over to my Clog chat. We have a conversation about it there. He knows me best.

And then once we come up with an answer, he gives me exactly the commands to put into the command line to keep the process going. Okay. Hopefully, you're excited about this direction. I certainly am.

I'm I'm really loving this new technology, and I'm loving AI, and it's it's gonna change the world. And we wanna be at the very front of this, and we wanna be with customers who are just like yourselves who are already building stuff and really wanna lean in. And I know there's a lot I can learn from you guys, so I'm excited for that. Thanks, Morgan.

Great. Thanks, Morgan. Thanks, Morgan. Thank you. Thanks, everyone.