Cohort 5 - Day 2 - 4-16-2025

Record on my side. And we've got Luke taking some. Some notes today with AI, which is good.

Yeah, you can kick him out. I. Trying to.

You know what, you can leave it on if you want, because it's quite useful for you later on. And if you're nice, you could share that with the class.

Yeah, of course.

Yeah, Those. Those notes are. Are quite useful, and I should have. I should have that on, and I just don't. But anyway. All right. How's everybody doing this morning? How's everybody. How did everybody do yesterday? Thinking about some of the things they were learning and just kind of letting it process. Let's talk about it a little bit. What stood out either. Either yesterday or, you know, afterwards, things you thought about and processed.

Can I go.

Go ahead, Luke.

I just thought it was so cool that we went from like, creating a little game to standing up an app, you know, and I'm just impressed with that. It's really quick iteration, you know?

Yeah. That process of going from, okay, I can do this to, you know, something simple, and then all of a sudden, wow, we're going way complex. And that jump is significant, but meaningful. It's awesome. Great. What else? Go ahead, ej. Yeah, I'll try, man.

Pardon me, Eric, The.

The chat gbt, in creating, you know, creating some thoughts there and ideas, business plan, whatever, and then running it through GROK with the deep research or whatever it's called, and then. And then, you know, getting. Getting updated on that and getting all that research, then putting it back into chat gbt, then back into GROK for the thinking concept to just deepen, deepen the research and come up with more. More info. I thought that was pretty cool. Didn't think of doing that.

Yeah. Yeah, that's fun. Bouncing back and forth between models is super powerful, and it unlocks just. And again, you can do that process, guys, you know, for three hours, right? And I've done that, right. I've gone back and forth and really dug deep into something and. And I'll ask Grok one thing, and I'll ask another model another thing, and I'll even. Even more than chat GBT and grok, but those are just two easy ones that we use in this course that are very, very powerful. The two of the best in the world. But yeah, I love. I love letting AIs challenge each other, right? Because that, that. That safeguards a bit against hallucination, right? Like, you know, you want. You want to make sure that you don't get bad information. And if. If an AI reads something, it's gonna say, oh, you know what? This actually isn't. This isn't right. You know, he's gonna catch it there. So that's good. Cody.

No, I was just gonna kind of iterate on what Eric was saying. It's just using the two algorithms, right, to speak to each other and refine it and then go back and see where the, you know, see how it really ties in. And then I kind of migrated it to my cell phone, so now I can move pretty instantly. As I mentioned to you, the usage that I have a lot of time is narrative.

Right.

Putting out direct responses. So for me, that was, that was really, really unique.

Yeah, that's cool. Yeah, I really love that. Welcome cameo. Good morning.

Morning.

How are you?

I have something to say.

Yeah, jump in.

Matt.

If I would have known about this probably four years ago, it would have saved me half a million dollars. And I, I was, I was about ready to have a requirements review with one of our sub development groups on this coming Friday on another project. And I, I've already texted him and called him off. I'm going to see how far I can get with this before I engage with them. I still might have to engage with them for some native mobile app stuff. But we still got two days left.

We still got two days left. And yes, the mobile native apps too. Yeah, we've got plenty to talk about with native apps. It's a great topic and definitely one that's important because of how we're a mobile first world. Right. And there's a lot of things that native apps can do that a web app can't do or a progressive web app can't do. But yeah, great. Matt, you're so spot on. Like, when I first started doing this, I was just like, I am never ever spending money on a dev shop ever again, at least not until a model is proven. And I can very easily afford it because I need to really robust, scalable.

Get your MVP to get your MVP out there and just kind of. And really start getting some customer feedback before you really spend some serious money. 100%, you know, and I'm, you know, if we don't cover in this course, because I'm a little bit more technical, but I'd love to follow up, make sure I understand how we integrate with GitHub and create development environments versus live environments and really prep it. So if I did pass this whole thing on to a development shop, that new AWS infrastructure, they could hit the ground running. So.

But we can talk about that.

Later.

But.

So far, I'm like a kid in a candy shop.

I love it, Matt. I love it. Matt Craig, I'll come to you next, Matt, just to skip ahead a tiny bit, Replit does have a native integration with GitHub, which is amazing. You just tell it, you just, you know, you connect your GitHub account, your project, whatever, and then you just say push to GitHub. You just tell the agent, please push the project to GitHub. It pushes it up. Then there's another app that we won't cover in this course that would probably be for an advanced course for this or some, you know, some, you know, more, more advanced, maybe consulting that I can do. By the way, Cody, I did get your email and I apologize I didn't respond. I was, my brain was dead last night. So I'll, I'll get back to you. But, but there's another app called Augment Code, and Augment Code just finished raising $220 million from, you know, all the big San Francisco guys. And basically what they do is they're, they, they're enterprise level Fortune 500 level AI tools that let you manage code bases with 10 million lines of code and sync it across all these teams. And it syncs with GitHub as well. And so there's a workflow where you can say, create, create, create in REPL.it push to GitHub, pull from GitHub into augment code, harden that code, make sure it's SOC2 compliant, make sure it's got all these pen testing and all the different things that Augment Code can do to really make sure that AI code is really, really solid. And then you could push back to GitHub and push back to Replit, build more stuff and have this workflow of going back and forth and really build enterprise level code all with AI as a single dude in your basement. And of course you can have a team do that too, right? But it's just incredible the sophistication levels we can get to down the road. So, yeah, anyway, good stuff, dad. Go for it.

As someone who understood, understood virtually nothing of what you just said in the last 25 seconds, I just have to say, from the standpoint of someone who is probably the least technically oriented as a commercial real estate developer, I didn't need to learn anything about code or. And it was in a different time. And so as someone who's more of a novice to this, I just have to say that it is, I'm still a kid in a candy store. It is just shocking to see how robust the tools are and how far they can take you, how quickly. And if it's not robust enough for you, just tell it, go deeper, go farther, and it will. And its ability to do that while following the premise that you're giving it in terms of, this is the direction I want to go, this is the style that I want, this is the, the tone that I want. And as you noted yesterday, you can even make it speak to you in a specific tone if that. If that motivates you or serves you better. So it's, it's just astounding. And I recognize, I think we all do, that we're still at the very early stages of what AI is going to grow into and what it can do. So it's, it's fascinating to be sure.

Right now you have to kind of babysit Replit a little bit and, or maybe a lot of it. You know, EJ's swearing at Replit last night, trying to get something to happen. And you know that that happens, right? I mean, you get, you get into a loop sometimes and it gets confused. And, and, and I've got a lot of tips and tricks we're going to go over today about how to better, how to, how to plow through the problems in repl.it right there are. It's not a perfect tool. And so I've discovered a lot of tips and tricks. We're going to go over a fun document today and, and practice that live. And you've seen me do some of the things already. But, but, but along the lines of, you know, we're just at the beginning, you know, Replit will get better, other systems will get better. You know, Google just announced a new tool just a few days ago that some of you may have seen in the news where you can create a, an Android app. You know, exactly in the same way that we're creating things in Replit. Although I've had some smart people get in there and they're like, yeah, it's not ready yet. It's not as far along as things like Replit, but, but it'll get here. I mean, give it three months, let them fix their bugs, and, you know, it'll. It'll be pretty incredible. Um, and so give it three years and you won't really have to babysit it anymore. It's just gonna, it's just gonna do it. Hey, I need this app. I need a backend. I need this. I need it up and running, right? And you know, 45 minutes later, it's, it's ready to rock, right? It's written a few hundred thousand lines of code and you've got a fully functioning system. So with that in mind, what does that, where does that leave us as a society, right? What is, how does that change? I want to have this discussion right now. We're going to start this morning with a little bit of just kind of warming up our brains instead of diving right into repl.it but where does that leave us when development is now as available? Let's take a few other things that used to not. What are some other things that used to be not as available to the average person? One thing that comes to my mind that I did, my parents know and people love people wanting to know. I used to make movies, right? Edit digital videos when I was a teenager and I was like way ahead of my time. Like I was really cool. I had all this cool software that would connect my little camera, those Hi8, you know, whatever things, and I could connect it digitally into the computer and, and upload the whole thing in. And then I would edit in Pinnacle Studio, I think is what it was called. And I could edit these, I could do these cool transitions and stuff. And this is back in, you know, 2001, you know, time frame and, and now today, you know, editing a video is, you know, my, my little four year old can do it, you know, or seven year old, a four year old can do it. But the point is that has changed. What else, what else has changed that you guys can think of? That's, that used to be kind of only the professionals with big bucks could do. And now we can do it. Now anyone can do it.

Music productions in that same exact realm.

Back in the day, studios here and everything. But yeah, I've got a home studio here that I can, I can record things at studio quality level. I don't go to a studio anymore to record my albums.

Yeah, yeah.

And that was even before AI that that revolution had happened. Now you don't even need the musicians.

That's, that's another discussion for sure. What else?

Players are a thing of the past.

I know we don't need anymore Robots.

Can do it better.

Cody, one, one thing that as I kind of reverse engineer this, I try to think, well, what is, what industry is AI proof, right? Again, I want to see because I'm scratching my head. I don't know. Service industry, I guess. But again, that's a whole different topic. Again, I always look at the odd lot where it's coming fast and Robust. And yes, I think that, you know, there's going to be some code that's written for, you know, malpractice.

Right.

There's a.

So, so again, I'm just, I'm alert.

Right.

So AI is going to be domineering. I would tell you, I mean my customer base not to go down this road, but my customer base is a salary guy. He's a guy that, not so much an hourly guy. The persons that buy my products and my concerns is that salary swap of industry. Not, I agree not to be bearish, but I'm just being cognizant of the guy that spends money, disposable income stuff.

Right.

So be it. It's a 45 minute conversation that we could have later. Yeah, I'm looking, I'm looking at. Okay, what, what are we doing here? My first question the other day was, do the valuations get, get mitigated? Because I mean, if I create something we talked about it is the value not, not 5x, is it now 1x? Yeah, I, I typed in something. I just, I can rip code off of anything. It's wild.

Yeah. Where's the moat? Yeah, where's the moat that creates that, that, that multiple.

To your point yesterday though, John, I mean, part of what's happening is look how different this group is today from 24 hours ago. And terms of our capability to be productive and to, and to take things further than we ever could have before. And John was discussing this yesterday. Part of his mission, if you will, is to help everyone out there understand that if you're not AI confident, you're going to be left behind as an individual, as an employee. And so all of those salaried people that you're worried about, they just need to become producers of a different kind now to ensure their, their place in the work, in the workforce and in society. We can just, I mean, again, this is a much longer discussion, but what if everybody were trained on AI? Does, does anybody disappear? Or do we just produce in a dramatically different way and accomplish things as a society and as a world that we never dreamt of before? Everybody just has to have the tools.

How hard is it to replace a plumber with AI?

You may not be able to replace him, but you can make him more efficient.

Yeah, maybe, but I think a plumber's maybe like, how hard is it to train a robot to be able to walk into your house, analyze the problem? I bet they can analyze it pretty well. That's not the problem. But creating the dexterity of a robot to be able to reach behind your toilet and unscrew something or switch out a, you know, a garbage disposal with little tiny tools and things like that. That's very, very advanced robotics that is going to be in the home for quite some time. You could definitely do it. It's definitely doable. But getting it down to a price point where it makes sense to replace a human, ironically is going to get replaced is going to take way longer than a high priced lawyer or maybe even a doctor. Right. I mean, you know, you guys far away either.

Honestly, I think it's not that far away. Dexterity.

Yep.

And, and I think there are a subgroup of things that are like plumbers. So I've been working on an idea for a while to replace sheetrock people. I, I had a really bad experience with the mudder and, and I think that hanging sheetrock and putting mud on is a great job for robotics. It's a very discreet set of work. It's very measurable. It doesn't take fine dexterity. Even putting mud on doesn't take fine dexterity. It takes the ability to analyze if something is smooth or not and continue to smooth it until it's smooth. Right. Like there's very discreet things. So plumbers might be further out, but I think a lot of manual labor is totally ready for disruption.

Yeah, I think, I think it's going to be, it's the low hanging fruit that's going to come first. And obviously the digital stuff that's, I mean by definition AI is digital. So anything digital is going to be rapid. And I don't know if you guys ever read Singularity by Ray Kurzweil, but 15 years ago he was saying in 2029 there'd be no distinction between the human brain and the computer brain. Meaning that you're either plugged in or you're one of those people that are plugged in or you're not. And when you talk to somebody, you're either talking about high level concepts with all the background knowledge or you're somebody that I don't know anything about that. And they're not even interesting person to talk to because. And it's like your level of knowledge is so low. It's like talking to a kindergartner about base or something. You want to go past the basic knowledge information into the ideas and concepts that the human brain is trying to invent. Anyway, that's 2029 is only four years away. Whether he's accurate or not, we'll see. But the concept being either you're a have or have not. And when it comes to AI, you're either a have or have not. As I definitely believe that, that you're either proficient and on the leading edge or bye bye. And it's as simple as that. So. But the digital side, certainly that's what we're working on now. Obviously we're not done. We can't be creating sheetrock hangers and plumbers. But y. We might want to focus on being able to create a website first.

For sure.

E.J. no one's ever replacing you teaching me what to do as I go into a rapid.

That is true.

Well, John, you just mentioned we're gonna.

Get a robot that's gonna hit your hand off my grab handle, grab the back of my boat. You know, the advanced psychology level stuff requires a human brain and not just any human brain, but like. Anyway, go ahead.

It's interesting though, you mentioned, you mentioned Dr. A minute ago. But when you think about it, there's two things going on around. First of all, virtual medicine is, has exploded over the last, you know, five, six years.

Yep.

And then if you look at, we've probably all seen the, the advertisements about AI and the, the virtual models that are out there now modeling clothing, you know, on the digital models that you can't tell unless you really pay attention. You can't tell they're not real people. And as that technology gets even farther.

Right.

Then the ability to, to have. And the fact that also I hope there's no doctors in the family around here, but the fact that PAs have already replaced doctors in the footprints, you know, at the, at the, at the, in the medical practices, we'll have AI doctors doing telemedicine. Virtual medicine at a very high level. I mean, at a very, I think it'll be very common pretty soon.

So I, I shared this with probably a couple people in my family and then we'll go to you, Mike, but, but I'll share it with the group here. So there was a study, and I'll find it if you're really interested in it. But there's a study just, just a few months ago that came out that said, okay, it basically took, it took 50 really, you know, kind of more advanced, you know, symptoms and diagnoses and things for doctors to look at. And there were three different groups that they, that they tested. One was that they wanted to compare against each other. One was just doctors. Right. You use your knowledge. And it was a group of doctors. I think there were 10 groups or 10 people that could work together. So you had 10 doctors. 10 doctors, 10 doctors. The first group was just the doctors. The second group was just the 10 doctors. But they could use AI in any way they wanted, right? They could build a quick tool, they could, you know, research, they could do whatever. And then they each had to come up with their diagnoses. And the third group was just AI. It just took the, all the symptoms and everything, fed it to AI and AI had to, you know, say, all right, this is what we think it is. And first of all, which group do you think won? Only doctors, doctors plus AI. Only AI.

I want to hope it was doctors plus AI.

I think doctors plus AI.

Okay, good, good. So the, the diet, the correct. So it was interesting that just the doctors got 87% of their diagnoses correct. Pretty good, right? This was a smart group of doctors. Doctors plus AI got 86% correct. Only AI, 97% correct.

Okay, John, here's the thing. Coming from someone who's worked in the healthcare space as a therapist and, and diagnostician in the concussion world for 15 years now, you, I, I'm not surprised at all that the AI can do it better. The problem is, and I think you would all agree, is that nobody knows how to use it yet. All of us are like, yeah, we, as you said yesterday, people are using AI right now like an advanced Google search. That's all they're doing. And if I had known, you know, we all learned enough yesterday to get very excited about the opportunity and the possibility of all of these things that we can do. And if everybody can. You know, I can think of a handful of people in my life who are super, super resistant to technology and they don't want to do this. They don't want to get in. Like my, my father in law didn't have a cell phone until like last year. Like he just refused, you know, and like a normal, like even a flip phone, like, anyway, some people are just super resistant to it. And if my, my husband and I were talking yesterday after the class, I was saying, this is amazing. There's so many things it can do. And my husband is actually a very technological, technological person. He knows all this. I'm right behind my dad is number two not knowing anything that I'm doing and, and a little overwhelmed with, with what we're doing, but trying really hard and learning. Fine. But if he took this class, I would imagine he would be along the lines of some of y' all who would understand it so much better. And if you just give it op. Give an Opportunity to, to learn it, to try to understand it. The, the. It appears that the possibilities are endless. And so if those doctors with AI knew how to use AI in a way that made it beneficial to them, I think they could, I do think there's that. That ounce of bedside manner and, and problem solving in a doctor's brain that I don't think AI Well, AI can never replace a bedside manner. It can never replace. That was. Yeah, Baymax is great, but he's also not the same as a human. He's just not. I agree we could argue it, but I think the point is, is that if and when those doctors could have learned how to use A.I. they might say, hey, what about this? And asked A.I. how to. Or to compute and factor in this and things that, you know, people just don't think about. I don't know how many things times I've been like, well, let's look up the literature and see what it says about this. And I could just ask AI to read all the literature for me and come up with a summary and it might be able to do it. Or like, hey, search in the medical journals from 50 years ago and see if you can find a case like this for me and help me diagnose this or whatever. And anyway, we just have to learn.

Yeah, I, that was the point I was going to make. I think that where there's real and Cody, I can't remember what the term you used, the, the. The space of not being where AI can't go, really, it's that it's a real experience. Right. Versus an artificial experience. And so I like what Jessica's talking about where you can. When the doctors maybe get really connected to AI maybe that number will change. I don't know. Maybe, maybe not. It's interesting to see there, but I think in the coaching world where I am with folks, and I loved how AI coaches us going through the questions. That's where I was last night, having it ask me more questions, which was great. But I think there's that element of a real experience, a human experience that actually is going to become more valuable in the future, those types of experiences.

I agree 100% with what you're saying, Mike, that, that realism, not just realism, but real reality is going to become more valuable. Right?

Yeah.

Actually can be approximated by AI. Right. And I'm going to show you guys today how you can, how you can hook up a voice, an AI voice that sounds exactly like you or exactly like whoever the heck you want that can replace you. Know a human function or you can create a voice, you know, interface with, with something and, and how all of you are, you know, within 20 minutes of training for me, plus what you've already done will be capable of setting up an AI powered, you know, voice assistant that sounds like whoever the heck you want. You want it to sound like Arnold Schwarzenegger. You can get his permission or don't get his permission, but, but you can do it because you can upload a little bit of his audio and it sounds like him. Or heck, you could make Jarvis because we've got a lot of recordings of him. You can make your own version that, that answers to you and, and helps you solve humankind's problems or humanity's problems. So, yeah, we're going to get into a place where real people are going to carry a premium and that's cool, right? And guys like, this is not the first time that this type of discussion has happened, right? Think back to the Industrial revolution, right? Everyone thought that humans were never going to be needed ever again. Now that factories and, and these robots and all these things can, can produce things at rates far superior to what humans can do. And we did just fine right now. It was painful. There was pain. There was some, there was some pain in there for sure. And you had some, you know, great depressions and things like that that happened for various reasons. But we're gonna, it's gonna be a, it's gonna be a very, very fast and hard change. We'll talk a little bit more about this later in a, in a discussion about surfing versus swimming that I think maybe only one of you, maybe EJ and I have had that discussion a little bit. We have, and my dad and I certainly have had it extensively. But the rest of you, the surfing versus swimming discussion will be for you to consider whether or not what you are doing today, the things that you're making today, the businesses that you're in already or what you're doing are AI proof and ready, or if this AI revolution is going to come in and just smash you against the rocks and, and say, okay, are you, are you ready for this AI tsunami that's coming in? So we'll talk about what that really looks like and how you can potentially build and avoid that, build against and avoid that type of, of a negative outcome.

Hey, John, you know the, the counter to the Industrial revolution comment, which I think is very appropriate. Yeah, those I'll just take like my world, right? The payments industry, if you look 25 years ago, there were a handful it was a billion, it was a billion dollar barrier to entry. And there were a handful of very wealthy people that, that ran that, that business. Right. Mainly Citibank, Wells Fargo, American Express.

Yeah.

So forth. And take the, the ability now to, to create, you know, to your point, create apps through AI or even before AI became popular, just to be able to, to very, to dig into some of the new technologies with Android and so forth. All of a sudden, you know, 20 years later you have thousands of multi millionaires in that same industry because of the simplification of the technology. Right. And the barrier of entry going from a billion dollars to thousands of dollars for people to create tools and so forth.

So yeah, you know what I love about this example is because I love payments, I'm kind of a payments nerd. And what happened with payments wasn't just. It became more democratic and a lot of people got into it, but the people who started to come in rethought what it meant to pay at all. You know, like the, the switch to kind of the mobile phones where Square came in and they, they didn't just start rethinking payments. That also impacted inventory, it impacted the data footprint that's left behind from payments that over the many, many years that industry hadn't really even that they just didn't even see the potential that was always there that Square has managed to leverage and other players in the space, even, even companies like Toast have, have changed what it means for a payments company to interact with, with the businesses.

Yeah, it's clear, guys, that we're in the middle of a revolution for sure. As one of my good friends, Jeremy Whit would say, it's a tsunami and, and tsunamis can't be ignored. Right. So we'll talk about that more later. So we've gone about, about 30 minutes here this morning already and I want to dive back into some, some more hands on things. Let me, before we do that, does anybody have any, you know, I'm sure some of you at least dove into your, your apps a little bit further in replit last night. Is there anything you ran into? You were just like, help. I need to know what's going on with my project and what, you know, how do I, how do I get past, you know, X, Y or Z? Jesse, go ahead and then we'll go to ej.

Sorry, I raised my hand first. Ej. I.

Oh, you muted. Yeah, you bumped that.

My bad. Okay, so you were talking about the files, the MD files and I don't think that I, I put the prompt in just like you asked, and I don't think it did it. And I'm not even sure how to tell if they're there. It says that they've been great. The PostgreSQL database has been successfully provisioned, is ready to use. But then. So the database maybe is done. But I'm not sure about those documents, and I think you saw them in your attached assets.

Yeah. Share your screen.

But I don't see them.

Yeah, share your screen. Let's look at it. Okay, everybody pay attention. When we're. When we're doing these, when we're debugging other people's projects, guys, can't tell you how beneficial it is for you guys to watch and see what's going on here. Okay? So first of all, I immediately see your files. See them over on the left. Progress.

Yeah, but those are the pasted things that I put there, right?

Yeah. Scroll down. Or just look down there. Right there. Progress.

And read me.

Yep, you just scrolled right past them right here. Big bold letters.

Okay, tell me again what those are.

Click on it. Click on one of them. All right, there it is. There's your progress. MD here's your current phase. Everything that's been completed, everything you're working on. Oh, cool, huh? Yeah. If you don't have them, we'll make them. Yeah, you got to make them. So. So that's fine. Mike, I'm going to have you share your screen in a second. I want to see if they're there.

Okay. What about the readme one? What is this?

There you go. Yeah. This is your business plan, right? Or it should be, at least.

How is it different? From what? From these other.

Yeah, scroll down. Scroll down a little bit more in your progress. Sorry, you're reading it. Yeah, it should be. It should. It should match what your business plan was that you pasted earlier. It should be pretty close, but go ahead and keep scrolling down the business model. Yeah. So really what it's trying to do, the purpose of these documents, guys, is, is to provide guardrails and alignment to your agent so that it doesn't just, like, go off the rails. Right. To stay on that. To stay on that analogy. When a lot of people using replit, if you get into the Reddit discussions and see people struggling with it, they just say, I can't get it. It does. Never knows what I'm doing. Every time I start a new agent chat, you know, I have to, like, tell it everything, and it gets lost and doesn't know where files are. And it, it doesn't understand why. And, you know, in reading, you know, in struggling through similar issues with my first projects, you know, many months ago, I was starting to work on all of this stuff and learn, you know, learning it all. I would get lost, too. Replit would get lost too. Now replit has gotten smarter. I would say replit has gotten probably 10x smarter since I first started using it. But this, this little trick of putting a progress MD that says, okay, this is what we built so far and this is what we still need to build. And this readme MD really is supposed to be focusing on. See, your, your readme MD is kind of gone off the rails a little bit. What I would say, Jesse, go click on the agent. And it's not that, it's not that bad, right? But click on the agent. And I want you to say to the agent, this, I want you to say, I want you to use the attached or I want you to use the business plan. Sorry. Yeah, say the business plan in attached underscore assets. Perfect.

Okay.

And, and make sure the readme MD file matches it. Enter. Okay, so we're going to kind of realign your readme MD file. It should, it should be, you know, so it should be able to see attached underscore assets. It sees that there's two things there. One is the, you know, the plan. One is the business plan, right? So one's development plan, one's the business plan. But, but really, guys, when, when your AI understands better context, it will make better decisions for how something should be architected. Right? For how the entire thing should be put together. Okay, this is the end goal. This is what we're trying to do. This is the business model. You know, all of that stuff is important so that it understands why you're trying to do that. So anyway, very good. That should, that shouldn't take very long. But now you're, now you're more aligned again and we're going to do a little bit more training with that. So we'll go ahead and stop your screen sharing here and thanks, John. E.J. go ahead. And we'll come back to you, Mike.

So, yeah, so if you all of a sudden had. Oh, man, I didn't think about that. So kind of an adjustment. Like, for example, on mine, I have a certification test and I have an onboarding for this AI assistant which both had a set of questions. I'm like, oh, we should combine the two. That way the user is only doing one set of onboarding questions. Do you need to go back, open the readme and change the readme file to reflect that because it wasn't in there before. And obviously the replit didn't create that situation before.

In a perfect world, yes.