Transcription
My name is David Andre, and here is how to build anything with Codex. In just two hours, I was able to complete dozens of GitHub issues, which would otherwise take me days to do. Codex really is the most revolutionary coding agent ever created. And in this video, I'll show you how to set it up for maximum productivity.
All right, so to access Codex, just go to chatgpt.com/codex. And pro tip, you don't need to pay $200 a month. A much more affordable way is to get the team plan. For that, you just need two users at $30 a month, which is the most powerful coding agent out there for just $30 a month. And trust me, it is worth a lot more than that.
Now, here's what you need to understand. Codex runs on a new AI model called Codex 1, which is the most powerful coding model OpenAI has ever created. Basically, they took GPT-3 and fine-tuned it on actual senior-level coding practices, aka how you would actually work in a production-level codebase. The reinforcement learning process for Codex 1 included writing unit tests, writing lots of comments, splitting into smaller files, grasping the codebase, and all the other things you would expect from an actual software engineer.
So inside of Codex, you'll see that you can describe a task. This is basically the prompt—what you want it to do, right? Pretty simple. Then you have these two buttons which let you choose the repository and the branch. If you don't see a repository here, you need to click on manage environments. Here you need to create a new environment for your GitHub repo. So as you can see, I have two different environments here. One for testing and one for Vectal, which is a production-level codebase used by over 50,000 users. So if you don't have anything, just click on create environment. You need to connect your GitHub account. That takes like 30 seconds. Once you connect your GitHub organization, just search for the repo you want to connect. Give it a name. You can also give it a description. And later in the video, we'll even dive into the advanced environment setup.
Now once you have your repository connected, you need to choose the branch. I would highly recommend you not use your production branch, right? So for me, main is where prod runs. So I would either use dev, which is my staging environment, or I would use my personal branch, David. Either is fine. This depends on whether you're doing like a quick fix or a very experimental feature. But do not run Codex on your production branch.
All right, so let me show you how Codex actually works. Now, instead of giving it some made-up task, you know, like "VIP code, some front end of website," like everyone else does, we're going to choose an actual GitHub issue from Vectal, which is, you know, my AI startup. And as you can see, we have, well, like hundreds of things to do here. So, we're going to choose something that I actually need to get done. Okay, this is a good one: "Make sure we save the last selected model and the last selected chat agent mode." So, inside of Vectal, we obviously have the model picker, right, where you can select different AI models, and we have chat and agent mode. Right now, it just defaults to the default values, but it would be cool if it saved what the user has last selected and then when it logs into Vectal again, that is selected automatically. So let's give Codex this task.
Okay. So I'm going to copy the name of this GitHub draft. Okay. So this is important. You need to write prompts. You cannot just give it stuff like this, right? Because maybe it will do it, maybe it won't. But you need to be specific. You need to be precise. I'm going to have this as an inspiration, but I'm going to write a proper prompt. I'm going to say: "In our front end, I would like to save the last AI model that the user has selected in the model picker as well as the last state of the chat agent toggle into the browser's local storage. Execute this in the simplest and cleanest way possible. The fewer lines of code changed, the better." So notice how this is a much better prompt than just this quick idea which I wrote down. Right? Sometimes when you get an idea how to improve your app or what you want to do, you want to quickly jot it down. You don't want to overthink, right? That's why inside of Vectal we have this ideas inbox where you can quickly write down ideas. However, when you write a prompt to Codex, you want to be precise. You want to give it info about what to change, what not to change, why this change is even being made, and stuff like that. So this I would say is a pretty good prompt.
And then when you are happy with the prompt, you have two options, right? You can either ask or code. Obviously, since this is a change in our codebase, we want to press code. However, a very underrated way of using Codex is just asking questions about the codebase. Say you're working on a team, right? And there's multiple developers, and someone works on a feature, and then your job is to keep, you know, continue expanding on that feature, but you don't know anything about it. Well, you would have to open the files, read through them, do some testing. With Codex, you can ask it questions, and it—it's by far the fastest way to learn and understand a codebase. So do not just dismiss this button. It is very, very OP, and it will allow you to learn programming and learn the codebase faster than any other way.
So in this case, let's click on code, and this will start another task. Codex is asynchronous, meaning you can have many dozens, even hundreds of tasks running at once while you yourself do some other work. This is the power of autonomous AI agents, and few people are realizing just how revolutionary Codex really is.
See, I get asked this question all the time: "David, which AI tool should I learn?" And usually my answer is try them all and pick something, right? Maybe you like this, maybe you like that. But what matters the most is if you pick something and stick with it. If you are someone who does programming on a daily basis, or let's say you want to build an AI startup, or you just want to do something that involves code, all of you should learn Codex. This is not like other tools. This is something where OpenAI has created a unique product that no other company offers. So whereas my usual answer is it doesn't really matter if you use Windsurf or Cursor, you know, they're mostly the same. As long as you know the tool well, that's what matters. With Codex, my answer is if you're not using Codex and you're doing something with code, you're going to fall behind. You're going to fall behind so hard that it's going to be actually painful.
As I showed you earlier, I was able to complete dozens of GitHub issues from my—my actual Kanban board, right? This is not made-up stuff. This is stuff I needed to do or, you know, that my other developers would have to do. In about 90 minutes, I was able to complete like a day and a half worth—worth of work that otherwise, even with current AI tools, right? Even with Cursor, would have taken me a day and a half, maybe two days. Because of Codex, because of the ability to give it many tasks at once and have it working on them asynchronously, I was able to make so much more progress on my AI startup that it's actually crazy. So if you still don't have Codex, if you're not willing to spend $60 a month to get it, and you're doing something with code, you're just missing out.
Okay, so inside of Codex, we can see that the task is already finished. So that was very fast. So let's click on it and let's see how long it took. 2 minutes 50 seconds. That's crazy. By the way, inside of the logs, you can see all of the different actions that it did in 2 minutes 50 seconds. This is all it did, right? Think about how long it would take you as a human. Even if you had all the same skills as the Codex 1 model, which is better than GPT-3, which I can certainly say I don't have all those skills and all that programming knowledge, but even if we had it, this would take us hours, guys. Right? And Codex did all of this in less than 3 minutes.
All right, let's switch back to the diff and let's see what it actually changed. So, we failed linting, and this is, I guess, because of—um, missing Next dependency, and that's because we haven't done the advanced environment setup, which as I said I'll show you later in the video. But let's read the summary. It added persistent storage keys; the last selected AI model and mode loads automatically. Okay, so at the top of the index.tsx file, it added two variables. Pretty simple. And then here it saves it to local storage when the setChatMode handler function is called. Okay. And same with the AI model. All right. So it only changed one file. It added 21 lines of code and removed seven lines of code. This looks pretty reasonable, and I think there's a very high chance that this will work. So what I'm going to do is in the top right, I'm going to click on push. I'm going to say create new PR, which stands for pull request. This will cause Codex to automatically create a new PR in our GitHub. So if we go to vector.ai and we go to repositories to the main Vectal repo, we can see all the pull requests, right, and now this is the one—remember chat preferences. So Codex has finished loading it, and now you know Vercel is running some tests for linting errors and stuff like that. And later in the video, I'll show you how to actually review the pull request as well as how you can use Codex and Cursor together. But first, I need to tell you about the Ultimate Codex Guide which we are releasing in the New Society. This is a step-by-step workshop that will take you from a complete beginner to someone who has mastered Codex. Plus, with each module, you get access to exclusive resources. So, in the first module, I give you my agents.mmd file, which is the exact system prompt I'm using with Codex to build my AI startup. So, if you want to triple your coding productivity with Codex, just like I did, make sure to join the New Society and get started on the Ultimate Codex Guide. The link is going to be below the video.
All right, let's go back into Codex and let's actually set up the advanced environment dependencies, which, to be honest, nobody else on YouTube has shown because, for some reason, most YouTubers just aren't willing to pay $200, even though it's their job. They're being cheap, and they're being lazy, and they're not showing you all the details of Codex. All the other videos I've seen on Codex, it's just the content creator reacting to some tweets or from the OpenAI live stream, which is just beyond lazy, and I hate that. So, I'm not going to do that, and instead, I'm going to show you how to actually set up the environment dependencies and all the other advanced environment settings because that's my job, and I want you guys to be on the cutting edge of AI just like me.
So again, inside of Codex, in the top right, click on environments, select the environment you want to set up. So I'm going to go with my Vectal one, and in here we need to go into edit. So top right, click on edit, and here we have again the basic stuff, right? So we have the repo name of the environment, but if we scroll down, this is the advanced stuff, right? So the code execution, you can add your environment variables. So if you want Codex to be able to, you know, interact with the OpenAI API or Anthropic API—let's say you're using Claude, right? Maybe you have an AI app that uses Claude as the model; you would need to add your Anthropic API key in here—but the main thing is the setup, because—uh, that's how you install the dependencies. So for my case, I need to do the back end, which is `require cxtd`, and the front end. So I'm going to do `cd backend` because that's my backend folder. And then I'm going to do `pip install -r requirements.txt`. For the front end, I'm going to do `cd frontend/vectal`. And obviously, this depends on the structure of your repo. So you really have to understand how your repository is structured. But if you don't, what you can do actually is you can open a second Codex instance, and you can ask it, right. This is the thing: "Explain how my codebase is structured and tell me which terminal commands I need to run in order to install all dependencies, both for back end and front end. Answer in short." So if you're working on a new repo you're not familiar with, you know, maybe you cloned an open-source repo, just ask this, and then instead of clicking code, click ask, and Codex will learn the repository, and it will tell you which commands you need to run in order to install all the necessary dependencies.
So while this is running, let me continue my setup because I know what to do for the front end. We need to do `npm install`, and this should be good. So this is actually interesting. We can see how Codex runs and how it completes our task, right? So first it—um, sets up the environment, then it turns internet off, then it reminds itself of the prompt we sent. It reads the agents.md file in its entirety. I'm checking the agents for instructions, understanding what's needed. Okay. Then it looks for requirements.txt. So it listed out all the files in the backend folder, and it found requirements.txt. Then it opened it and starts reading it. So basically, it did everything it needed to understand what commands we need to run, and it told us exactly what I wrote down. So you know, `cd backend pip install -r requirements.txt`, and yeah, this is what we have: `cd backend pip install -r requirements.txt`, and then `cd frontend/vectal npm install`. So this is correct. So inside of our environment right here, we can save this. But one more thing I'm going to do here. I'm going to click on edit again, and I'm going to give it our OpenRouter API key. So click on plus. I'm going to do OpenRouter API key. That way the Codex agent can actually test and run different tests with OpenRouter because that's what I use to power Vectal. So I'm going to create a new API key here. I'm going to do Codex testing key. I'm going to give it some limit because it's an autonomous AI agent. So I'm going to do $50. That way it doesn't go too crazy. And I'm going to copy this API key and I'm going to go into Codex. I'm going to paste it in. And then I'm going to scroll down and click on save environment, which is conveniently right underneath my webcam. Big white button, save environment. So now it has the OpenRouter API key. So if Codex wants, he can run different tests, different unit tests using the AI we have inside of Vectal, right? Because before it couldn't do that since it didn't have the environment key, but now it can.
Now, before we give Codex any more tasks, let's look at the last one we did and let's figure out what happened because it failed some Vercel deployments. So, I'm going to click here, and let's actually read what happened. Okay, there's an error here. I'm going to copy this. I'm going to go back into Codex. I'm going to click on the task. I'm going to say: "The code changes you did resulted in an error when Vercel ran its checks on it. Here are the details. Fix this like a senior developer would." I'm going to ask you to code. And now Codex will get back on it, and it will hopefully fix these linter errors so we can actually implement this change because I do want to put this into Vectal. This is something I actually want to add. And again, I'm trying to use this for my AI startup. This is my main project, right? This is what I spend most of my time on. It's not just some like, oh, vibe coding some front end for some made-up app, right? This is something that over 50,000 users have used. We have thousands of daily active users currently. We have hundreds of paying members that have Vectal Pro. So this is a real project. It's a startup that multiple developers are working on that I'm dedicating most of my time towards and that I have hundreds of things I need to do. So let's give it another task. Let's see how good Codex really is.
Okay, this is an interesting issue. So "Project created by an AI agent had a longer description than we limited to words." So let's do that. Let's copy this. Go back to Codex. Okay. So, seems like we have put the environment set up incorrectly. So, what I did—and by the way, this is what you should do as well when you run into errors—is load up GPT-3, the most powerful model right now, and just describe the issue. Right? So, I gave it the screenshots, and it told me what to do. I asked about clarification. So, I'm going to go back into our environment, and I'm going to change it. And honestly, I could easily cut this out from the video, but I think it's valuable to show you that there are likely to be errors and how to fix the errors, right? So, for this, I'm going to do what GPT-3 gave me. This should be the correct way because we need to `cd` back into the root. It's a mistake I did because otherwise you cannot do this because it tries to do it from the backend folder, right? So, you need to `cd` back to the root folder and then install the frontend dependencies from the frontend folder. Okay, so that's a rookie mistake, but make sure to save it. Save environment below my cam. Big white button, save environment. And—uh, make sure that this actually saves because—uh, before I had an issue where I tried saving it and it didn't save. So let me reload the website and let's check that this has saved. Okay, because this is a bit buggy. So let's click on vector.ai. Yes, we have the setup script saved. So now if we send it as a new task, it should work.
Okay, I was running into some errors here. Even though when I sent it as a new task, it still failed. I don't know why. So, let's copy it. Let's go back to Vectal—I mean to Codex—and let's run it not on the main branch, on the David branch. Okay, let's describe the task again and click on code. So, a new task is going to deploy here, and we need to make sure that it doesn't fail in the dependencies install process. Okay, so there it is. Let's click on that—working task, starting container. Usually it fails with like—if it fails in the dependency setup, it's in the first 15–20 seconds. So if it doesn't, we'll see very quickly. Okay. So `cd backend pip install -r requirements.txt`. That's good. That's good. Now while this is going, I need to say that we are hiring. Obviously, what's better than one person using Codex is five people using Codex. That's why I want to expand my Vectal team with an experienced backend developer. So if you are someone with five plus years of experience, make sure to apply; the Google Form is going to be below the video.
All right, let's switch back to Codex and see how it's doing. So it seems like it did not fail yet. Fingers crossed. Let's see if the frontend dependencies installed correctly. Unknown environment config. Okay, so nice. The environment setup did not fail. So whatever I did before was incorrect. Make sure to `cd` back into the root just as I showed you like this, right? Whatever you do, you need to `cd` back, or you need to do `pip install requirements.txt cd` at `backend`. There are different ways to do it, but the order of these commands matters because remember if it sees `backend`, it's in the backend folder, and then you have to `cd` back to the root to install the frontend dependencies. Okay, so now it's actually running the `npm run build` command to prevent any linter errors. So if we had installed the environment properly before, this would not have happened because Codex would have coded, and we would not have to, you know, send another prompt and debug it. So this is very important why you should do the advanced environment setup. Why you shouldn't skip that is because you'll save yourself a lot of time; instead of, you know, Vercel catching these errors, it'll be Codex itself when it runs these tests to find these errors.
Okay, so here one of the tests failed because I didn't give it all of my environment variables, which honestly I probably will do. I just gave it the OpenRouter API key. But if you want it to be working as a senior engineer would, you need to give it all of the API keys and environment variables for your development setup. Obviously, do not give it your production stuff, right? Keep your production API keys safe. I wouldn't trust Codex yet with your production setup, but—um—seems like giving it all of the development or staging environment API keys and environment variables makes sense. So definitely do that. Anyways, it finished running. So let's see—let's see—uh, the solution summary. Added date type to calendars date. Okay, so four lines of code added, and—uh—that should fix the—fix the issue. So let's push this to GitHub as well, and we'll see if that passes the Vercel `npm run build`. So let's go to pull requests again. And here's our new one: Fix calendar week start type. Okay, so the Vercel checks are running. Let's see here. Okay, this is actually funny. Now I see that one of my developers is trying to use Codex because, by the way, I purchased the team plan for everyone because you need—you need your people to be on the cutting edge of AI. Guys, don't be cheap; like, please don't be cheap. This is the AI revolution. Now is not the time to save $20 or $30. So what I did is the moment I knew that Codex is on the team plan, I instantly purchased the team plan for everyone, and I put all of my developers on it, and I encouraged them to use it. And as you can see, one of my devs right here—fixed project description link—he started working on the same GitHub issue with Codex, and you can see that with the branch CodexFixProjectDescription, right? So this is how Codex creates it before you merge it into one of your branches. So that's kind of funny because he tried to fix the same issue with Codex, or he maybe—maybe he already fixed it actually. It's pretty simple. Yeah. No linter errors. So he—he probably did it with Codex. So this is amazing. Now we're able to ship as a team two to four times faster. I mean, personally, I can easily say it tripled—three to four-exed my coding productivity because I think I'm like very well positioned for it. I'm not a senior developer with, like, you know, 15 years of experience, but I'm also not a complete beginner. The issue with being like a super experienced programmer is that usually they have a lot of ego, right, and they don't use AI enough because they want to write it themselves. But the issue with being a complete beginner is that you're simply—like, you don't know anything. You don't know what to accept. You don't know, you know, what pull requests are, how to review pull requests. You don't know any of that. So you'll get—you'll quickly vibe code some frontend MVP. Sure. But you don't know how to continue. You don't understand any code; you know what functions are. So being in the middle is like the perfect balance. You're not—you don't have too much ego to just not trust the AI, and you use the AI enough, but you also understand the basics of programming, and you are technical enough to build fast. So that's why I'm—I already can tell you certainly for the last two days of testing that Codex will three to four-x how much work I get done per day. The crazy part is that now I'm putting all of my developers on it, and we will just build Vectal way faster than any—any other team. So again, if you want to join a rapid AI startup on the cutting edge of AI that doesn't have any bureaucracy, any middle management, and that's shipping updates to production every single day, make sure to apply to the Google Form. It's going to be linked below the video. I'm just waiting for Vercel to complete these checks so we can actually—uh—test it.
Or you know what? Let's not do that. Let's just uh let's just yolo it. Right.
So, this is going to my branch. So, that's safe. So, I'm going to merge this pull request and I'm just going to send it full send it because I want to show you how to actually test it inside of cursor as I promised earlier in the video. So, what I'm going to say is I'm going to switch to David branch and make sure to hard reset it to match and pull latest changes. Okay. So uh cursor agent can easily do this. You know it can run terminal commands. It it knows how to use git. So it will do that. It will switch to my David branch and will quickly run git pull origin David just to ensure that it matches with you know the latest deployed version. Okay. So good. So I'm going to clear this and now we can actually we should actually run probably also merge the current state of dev into David. Some of the features are missing. My branch is older, I guess. But when it gets stuck like this, just tell it to run git status and to finish it. Okay. So, seems like no conflicts. All conflicts are fixed. Yeah. Okay. Good.
So, now um we should have those changes, right? So, again, we're on the local host here and by default GBD4.1 is the default model and agent mode is the default mode. But now since we have merged the pull request from codeex, we should be saving the last state. So let's switch the latest model to Gemini 2.5 Pro and let's switch agent to chat and let's reload to see if this is saved or not. Okay, it doesn't seem to be. Um let's check what happened. Did we merge the pull request properly? Okay, pull request has been merged to David. Okay. So, I'm going to tell Kursa to run git pull origin David again. And perhaps uh this is not being saved properly. Okay. Let me actually reload this. This is a good idea. I should reload this. And I should probably restart my server as well. So, this is already up to date. I know this should be automatically restored, but I don't trust uh the front end. So, I'm going to restart this automatically. I mean manually, what I meant to say. So, that should be installing. And now when I reload, we should have the latest latest version. Okay. So let's try again. Let's do Gro 3 and let's switch to agent and let's click around and when we reload the website. No, it's not saved. Wait, let me see. So what I would do here is I would go back into codex and say the changes you did to actually what we should do we should give it the original long prompt which was um this I'm going to copy this the changes you did to implement the following did not work okay I'm going to send the original log prompt investigate why upon loading our app, we still default towards GBD4.1 and agent mode instead of just selecting what the user had last selected. Okay, so I gave it a bit more context and I'm going to ask it to fix it.
Now, this is the real beauty of Codex. While it's fixing on this task, what I can do is I can just go back to our canvan board here, find another thing that it needs to do and launch a new task. Right? This is what a lot of people don't realize. It's not like you have to wait for this one. You can launch 10, 20, 30 different tasks at once. You know, go for a run, go to the gym, go take a sona, come back, and they'll all be, you know, in progress. Obviously, not all of them will work on the first try. Obviously, however, a good 20 30% might. And for the other ones, you can give it follow-up prompts saying like, "This didn't work. You know, here's a bit more context." Or, "This didn't work. Try updating this file." Like, I'm telling you, Codex is very, very powerful. All you need to do is just remove your ego. Stop wishing for these oneshots, right? Like there's a lot of these YouTubers that like one-shoting different games with AI model and then think that's impressive. Nobody cares. Nobody who's developing actual software cares about this. People care about how good these AI models and AI agents are on production level code bases such as the one for Vectal, right? And by no means this isn't the biggest codebase in the world, but as I said, it's software that thousands of people are using every single day and that hundreds of people are paying for for the Vectal Pro paid plan. It's a codebase with thousands of different GitHub commits and with hundreds of open GitHub issues that we want to do. So, it's a real project is what I'm trying to say. It's not some imaginary oneshotting of a snake game like everyone else does, right? So, this is what I'm trying to show you. Codeex is powerful on actual projects. This is actual productivity boost. This is a like a step change towards AGI. This is one step closer towards a fully autonomous agent that can do everything. And nobody is cares. I've se most YouTube even Fireship like the you know number one most watched technical coding AI YouTuber didn't even make a video about this. Most of the content creators have simply no clue just how insanely revolutionary this product by OpenI is. And even those that did make a video on it weren't willing to pay $200 to test it. Even though they make like 20, 30, 40K, and this is literally their job. So yeah, it's crazy. Uh hopefully you guys realize just how big this change is. This isn't just another AI model. This isn't just some overhyped release. This is the most underhyped released I've seen since the release of CHGBD. And as I said on Twitter, follow me on Twitter, David Andre, this is the biggest AI product since Chad GBD itself. It's insane and it's beyond powerful and it will save me multiple hours per day.
Okay, so now it ran for 14 minutes, which usually this doesn't happen. Usually it runs for like two to three. So it really put in a lot of work now. But let's see what what happened. uh added local storage keys, introduced constants, loaded store values, processed changes. Okay, environment variable. Yes, we know it's missing, but what changes did it actually do this time? It loads the previously selected models. Okay, so I'm going to push this again and we're going to test it locally through cursor to see if it works or not. We should see the new pull request here coming any moment. Okay, now let's reload. Okay, fix calendar week start type. This should be the one. Yeah. Now, so this is by Codex. And by the way, you can see the tags. Codex. Codex. Codex. That way it's clear what the AI agent did. So I'm just going to quickly merge it inside of Curs. I'm going to say run git pull origin David again to pull these changes because we need to make sure our local matches. Okay. So now when we reload, let's see if things work. So again, GPD4.1 is default for me and agent mode. Let me zoom in a bit more. So what we expect and what we want is to save the last selected model. So let's do GI 2.5 and we switch to chat mode. And now if I close the tab and I do local 3000 again, this should be saved. There it is. It is saved. Gemini 2.5 Pro and chat. So this is a quality of life change that I will definitely push to production because why not? You know, let's say somebody really likes using cloud 3.7. It makes no sense for them to be reset back to the default model. It makes sense for for vectal to remember this and anytime they visit vectal, this is saved. So again, this is something that I would have to do or one of my developers would have to do. But now I just gave it to Codex and yeah, there was like one error or whatever. I had to send it another prompt. But again, it works autonomously. You know, I didn't have to wait for these 14 minute 14 minutes when it worked. I can do something else. You know, this is running asynchronously. So I can give it 20 different tasks to work on and then when I have time I just come in quickly review them. The ones that look good I do pull requests right? Then if there are no conflicts I merge them into my branch. I I test it locally. If it's good I push it to staging environment where our testers can test it and then we push it to production. So again, I'm that's why I'm beyond excited for Codex because this is truly the most powerful coding agent ever created and it's unlike anything we've seen before. And if you know how to use it, it will easily triple your coding productivity. And again, that's why I created an entire workshop on Codex, which I don't do for most AI tools like this is the first one I did for any coding agent because I truly believe this is revolutionary. Oh, inside of a new society, inside of the classroom, you can access the ultimate codeex guide. This is a step-by-step workshop that will take you from a complete beginner, even if you don't know anything about programming. And even if you only have a few hours per week to dedicate to this, this will take you from zero to someone who has mastery of Codex, the top 1% of Codex users. And I can confidently say that if you implement the lessons in this workshop, it will easily double or triple your coding productivity. So again, the ultimate codex guide is releasing in the new society and if you want to join the new society and access all of the exclusive content we have inside as well as copy paste templates and presets such as my AI automations or my AI agents which you can implement into your life right away. Make sure to join the new society. The link is going to be below the video. With that being said, thank you guys for watching and have a wonderful productive rest of the week. See you.