CODEX: Beginner to Expert Course! (1 HOUR!)

GPT codecs one hour free course. So today you learn how to master 99% of codecs in the space of 60 minutes. So I'm going to walk you through exactly how to use it, how to get the most out of it, how to build and automate anything with it. We've actually tested it on literally hundreds of projects at this point.

And with the new model from ChatGPT, which is GPT 5.6. So this is their new frontier model. We've tested this across many different versions of different games, apps, tools, et cetera, and it can build some unbelievable things, right? So this is something that I've relentlessly tested.

And I can tell you now that this is one of the most powerful tools that you can get access to. And codecs can basically help you automate and build anything. So today we're going to get straight into it. If you're not sure how to get codecs, how to get started, you can just download it from their websites.

You download it from macOS. It has changed recently since GPT 5.6 Sol just dropped, and you can see an example of it over here, right? So this is what it's going to look like once you pull it up. You do need to log in once you've downloaded it, but basically this is what you're going to see as soon as you get started, and there's all sorts of amazing features with a codec.

So for example, you can build websites, tools, games locally. You can get it to actually control your computer. You can get it to connect to all your favorite plugins. And with that, we're going to get straight into this.

So what you'll see on the left-hand side is that we've got everything that we're working on, on the left-hand side. Now, if you want to use the normal version of ChatGPT, not codecs inside the same app, then this is what they call a super app, and you have the chat over here, now this can pull in your conversations from your login on ChatGPT. And, but this is just literally the chat. So inside this section, you go back and forth with ChatGPT, whereas with codecs, this is more for building big projects locally and building like automations that can run 24 seven.

So to get started, really, you need something that you have an idea for what you want to build. If you're not sure, I would just look at everything that you work on day to day and be like, right, we need to build this, or we need to build that, or we need to start working on this. So let me show you an example of this, right? So, and let's say, for example, we wanted to build a website that could essentially replace everything that I have in my SEO agency, and then people could order online, right?

So it's like a, it's a website that's kind of like e-com for my agency, and people can just order online via that website and start ordering SEO from our site. So we can start from scratch. I'm going to call this SEO agency e-com store website, just as an example to show you, and then we are going to hit save like so, and now we can choose between whether we're going to build locally or whether we're going to build in the cloud and also whether we're going to fork the conversation. So let me explain what this means.

So you see here, we've got the project folder. So it says SEO agency e-com store. This is the project that we are working inside to build out our automation. So we want to have this stored inside a folder, so it's neat and tidy and easy to come back to.

If we want to have a look at any of our other projects, so for example, like a daily Gmail automation, an agent operating system that we built in recently, a template set up, a habit tracker, et cetera, we can just scroll through this list. But if you're starting from scratch, just create a new folder. Now you can choose whether you want to do it locally or in the cloud. Cloud is great for an environment where you need stuff to be sandboxed away.

So what that means essentially is like you would have your agent working in the cloud, doing something else, and it's not like connected to your local setup. So it's not connected to your computer at all. But if you want all of your files to be available on your laptop, on your computer, on your Mac, whatever you're doing, then you want to work locally. I think it's more useful to do that because then you can come back to the files, you've got them all ready.

If we wanted to start using Cloud with the same project, we could just switch over to Cloud and open up the same project there. We could have multiple different agents working on the same project. I do think the local is better in general. But if it's like something where you need maybe a super secure setup, then Cloud is better because that way it doesn't interact with anything or any files you've got locally.

Now you also see here that you have the usage remaining section. This just allows you to figure out, okay, how much usage do you have left? How many tokens do you have left? What percentage of your plan have you used up?

And a real battle with new models, especially since GPT 5.6 came out, is that these frontier models are unbelievable. Very, very powerful. They can create amazing stuff. Biggest issue with that is that they're expensive to run.

And so now we have more problems with token usage, which means that we have to be efficient with our setup and I'll come up with more ways you can optimize that later, and we can always upgrade our package over here if we want to. So from here, we're going to get started. What you're going to do from here is you are going to click inside the do anything section, and if you have like anything that you want to switch between or anything that you want to set up here, you can see that we can click on the plus side. So if you have like a logo that you want to use, you could upload a file and folder.

If you have an existing piece of code that you need to change, then you can plug that in. And you can also attach Cloud to this. You could have goal mode running, you can have plan mode, et cetera, or come on to what those means. We can also add plugins inside there so we can connect to different apps, even like Spotify, we can connect to, and then we have our apps and plugin management over here.

So I'm going to start with the basics and what I'm going to do is we want to plan our project first. The reason that you want to plan your project before you go into building it is that it's been proven with research, Kilo Code actually did a research plan on this, and they found that if you plan ahead first, you'll get better outputs because all the decisions are made when it comes to building your project. Whereas if you have lots of decisions that need to be made later, the AI agent can go off track, it can make bad decisions, it can get lost in the context. So it's better to start generating a plan first.

So I'm going to say inside the chat here, can you come up with a plan to build out an SEO agency website for Goldie Agency? And basically what it's going to allow people to do is you're going to productize my packages that I have for SEO, and you're going to basically set those up as a web page where people can order online. So I need you to plan out every element of it. I want to give you my brand colors and everything else, my logo, and also the key links as well, so that you can use all this to help you build out.

And so I've just given it a simple prompt here. You don't need to go too in depth with the prompts. The actual plan itself is the prompt for Codex and it's going to be a simple What I'm also going to do is add a few useful details here. So for example, we're going to say, okay, here's a link to my website.

Here's a link to my website. If people prefer to book in a free SEO strategy session first, before they order online, and what we're going to do over here is we're also going to grab our sales proposal details, as you can see here. And basically you can give it any number of details. And this is really good because, you know, the more information, the more context you give to your agents, the more powerful and useful it can be.

So I'm just going to get the slide deck from here and we're going to grab that as you can see, that's going to download, and that's going to be used to train our agent on exactly what we want this project to be about. So if you go back into Codex, you can see that we've got our sales proposal, we've got that's going to be used to train our agent on exactly what we want this project to be about. So if you go back into Chachiquity here, this is good practice. We're going to go to add files and folders, go to the downloads folder, add in our proposal here.

So now it understands, okay, these are the packages we sell for SEO, and these are what we want to add to the website so that we can productize everything. Also, what we want to do here is we want to describe the style of the content. So I'm going to say, write the website and make sure it's beautifully designed, use my brand colors, which are out below. And also, can you make sure that you write in kind of like Alex or Mosey style copywriting so that the website converts nicely too.

And then underneath here, what we're going to do is just add my brand colors. So now it understands, okay, this is my branding, this is how we're going to write the website. This is the page we're going to link to if they want to book in a call first before actually buying anything. And we're going to be able to productize my agency and build out a new website for it.

So it's beginning to have a little think about that. And again, we are in plan mode. What you're going to see at the bottom here is that we can switch between the setups for what sort of access it has. Now, full access means it can just run autonomously, doesn't ask us for anything, and it will just make decisions on our behalf.

This is very useful if you want to use it fast, if you want to move fast. But again, you're not going to have much control over the process. If you want to have complete control over the process, bear in mind, this can be a lot slower. You're going to be making a lot more decisions.

Then you can switch it to ask for approval. That's actually really good if you're working on a project where you don't want to edit or make any mistakes and you just want to keep it really simple and easy. So from here, it's beginning to build out the plan. It's running the command.

If you click on these different updates it's giving you, you can see more details about what it's doing. So you can see, for example, it's coordinating here. It's running the terminal command. And the beautiful thing about this is like, if you went back a year ago, I'd still have to use Terminal.

I'd still have to understand how Terminal works. And don't be put off by the name Codex. I actually think that's why they've renamed the app itself to Chatuberty. Codex and the features of Codex make it very, very easy to be non-technical and build great things.

And I'll show you some examples of what I built in a second using the same app whilst waiting for this to be built. So you can see here it's running commands for us. It's beginning to, for example, look at the images inside the presentation. So.

It's actually opened up our presentation. This is an example. And it's beginning to create a idea of how to build out the whole presentation and everything else. It's also analyzing the web, looking for the research and you can give it more context here as well.

I'm just doing this on a quick demo to show you how to use Codex, but you could have more information inside the setup to see what it's doing. And so we've got this information in terms of what it's working on, what it's doing, et cetera. On the right hand side, you'll see the environment. And so we can see that's running locally.

And we can switch to cloud at any point as well. We can see that this is the main branch. You can actually fork the conversation. And what that means is like, if you have two different features, let's say, for example, I have one feature where I'm trying to improve the design and one feature where I'm trying to improve the plan, then I can actually create branches of this whole setup.

And we can see here, if we go over to our tasks, that this is our conversation. So we can archive the task. We can open it up and find it, which means we can actually see the files we've created and we can pin it. I would recommend that you actually pin this project just so you can cut back to it later, and it's going to be really easy to just see that later and see what we're building.

Over here, I want to show you some stuff that we're working on. Actually, it's asking for some input. So it is asking for some ideas of what we're going to do. So it's like, what should this build become relative to the Goldie agency website?

So it's actually recommending to replace my whole website with this new setup, which I'm not going to do. I don't want to replace everything. So we actually want to store this as a subdomain and keep the current site separate. Or we can have it as a single page.

So I'm actually going to say, keep it as a separate page, host it on a subdomain. And it should be a full website, but make sure it's completely separate to the Goldie agency website. So we're going to hit an X on that. And then it's actually asking, okay, like what model should we allow people to do here?

So I'm going to say, allow people to buy links on demand and also bundles and packages, so allow them to pick and choose whenever they want, and we're going to hit next on that as well. And then it says, you know, how much qualification should it go for? And I'm just going to switch to customer portal. It's quite easy to go through that decisions.

And it's good to ask me actually, cause it was, you know, if it replaced, for example, the whole setup that we have for our Goldie agency website, that wouldn't be great. So even when we're on full access here, if it's something that's like deleting or replacing existing files that we have, it's going to basically make sure that it asks for a permission test. It's like, who do you want the homepage to speak to first? So I'm going to, again, like these suggestions, not that great, but I'm going to say, just go for like established website owners and then how should they pick and choose?

So when you say like guided configurator and then what should logged in clients be able to do after this? So they should be able to see like a full campaign hub. Now you can see here it's in plan mode. We can switch to action mode later.

This is not going to build anything just yet. It's just going to plan it out. And then we also have our existing models here. Now I'm actually running this on a speed mode, which means that it's going to be faster.

The reason I'm doing that is because I'm showing you a live demo. So if you're using it on normal mode, do expect it to be about 50% slower. Speed mode makes it like 1.5x faster. We can also change the effort dials here.

Effort equals, more effort equals better results, but slower outputs. Lower effort equals less intelligent results, but it moves faster. And then we can switch between the models here. So Sol is my favorite by far for using codecs.

I just, I think at this point, like there's no contest between Terra and Luna. I don't know anyone else who's really using Terra and Luna. I do think that Sol is by far the best. And then what should this, it's asking for ideas on like, you know, what should the visual style be with?

So I'm going to go with bold. And then we're going to go with the custom brand system here. And then it says like, how should backlink profiles and everything run inside this? So we're going to create like a integrated system here where we're going to create like a login for the admin area as well.

So plan mode is really, really in-depth. Now I will say you don't always need to use plan mode. So if you just want to get straight into building system, let me show you some examples of how we did that. So what we've done actually over here is we were building out an agent offering system just as a test run using this separate project.

Now, also just whilst we're waiting for this to be done, you see how that's like buffering right here with the buffer circle. So we can come back to that conversation and it's just like, this means it's fully autonomous and it can just do things in the background whilst I'm talking to you. So in the meantime, I want to show you this. This is an agent operating system that we're building from scratch as a demo.

And we basically got a GitHub link and we were like, install this, make sure it works. Set up Hermes agent. It did that, send it all off for us as we can see here. And then we said now create a mission control dashboard, make it clean, make it interesting, et cetera.

And then actually built the whole website and the whole dashboard as you can see. Now, if I say, re get this live again, like, so it's actually going to get the agent operating system live again, which I'll show you in a second. And in the meantime, we've got our website here being built. So it says, you know, what should be used for all of this?

So we're going to start with the details from the slide presentation, and then we're going to add some other tips here. And then it's asking like, you know, what platform should it use as well? So I'm going to go with the recommended platform. And also something to note here is like, for example, if it was deploying the website, it needs to work with, with apps across, you know, all these different apps that you can use to host it or store it and that sort of thing.

So for example, if you wanted to accept payments, you would use something like Strike when you're building out this project. Now to set Bose up, it will actually guide you through the process, but you can go to the part of the website where you want to set up the payment. So to set Bose up, it will actually guide you through the process, but you can go to the plugin section inside Codex here. You can search for plugins and then you can add them to your workflow too.

So for example, if you wanted to build a presentation using Codex, you can actually link it to the presentations feature here. And if you were, for example, trying to add, I don't know, like this, you, you could just install this pretty quickly and install the details here. Just continue here. We can log into our system, we can authorize this, and we can just add that set up here.

So you can see here that we're going to enable this, we're going to authorize this, we're going to give it permissions as you can see, and then we'll authorize. And now it's authorized that and added that plugin to Codex. So adding plugins and allowing Codex to do work in the real world, aka with your different apps, pretty simple and easy using this whole system. Now, if we go back over here, it's adding the connection as you can see, and now that's fully set up and then we've got the idea here and we'll say, there we go.

And then also how it says like, you know, it's really planning out meticulously every single part of this process. So I'm going to go with AI assisted content and then it's like, okay, and here we go. We're going to add those in. And then if you go back to that project where we have the agent operating system, we can preview it.

Now, something that's really useful inside Codex here is that we can preview our projects. So we have the chat here where we can make changes. And then on this side, we actually have the project itself that's being built and this is hosted locally. Now, this means that we can preview visually what we're building all inside the same workspace.

We can resize it as well. So if you're building an app and that sort of thing, you can build that directly here. And then we're going to use the subdomain as you can see. So I'm going to go with this one here and then I'm just going to say reuse app and we're just allowing each of those setups here.

And if you go back into the setup here, you can see that we can actually build our tools inside our project. So if we have a look at the agent operating system that we're building, we've got a section where we've added in an image generator into our project, which is over here. So you can actually create images and we've created an image tool using the same project and it's all inside separate conversations inside that project folder. So you want to think of the project folder as like this is the master folder where all the activity is in.

But then if you want to add new features into whatever tool, website, game or app you're building, you can go ahead with this. Now, if you, at some point, go and find like, this is just taking too long, then we're going to actually switch over and we're going to delete plan mode and we'll say, okay, go off and build this now. So I'm just going to hit enter like so, and this will actually start building out based on the plan that we've created earlier. And I think that's good.

You know, if you've been planning out for 20 minutes, for example, now you can actually start building and this is now creating the project for us. And usually the building section and the building system here is way quicker. Now, in the meantime, I can also show you some of the cool stuff that you can do with this. So we have computer use with codecs, as you can see right here.

So we can control our apps using codecs too. And what this allows us to do is we can set up a new project here. So I'm actually going to say done working on your project and we're going to say, okay, computer, open up school and we're going to say, okay, let's have a look here. We're going to open up our school community app.

We're just going to plug in that link directly. link into the chat here. And you'll see here that we can tag in computer use, which is the plugin, and it can actually open up our browser and start using our browser for us with Codex. And we'll see an example of what it's actually doing over here.

So it's actually opened up this page, as you can see. Now, if we go back inside the chat GPT app here, we can say, okay, great. Now create a post on how to do social media marketing with chat GPT. And so using computer use in the tool, it can actually automate, for example, our social media and that sort of thing.

So it's now using computer use. It's going to start creating the social media post and we can see it working over here. We can actually open up the browser. If you want to see what it's doing in real time, we can get a little preview window over there.

By the way, what's amazing about this is you can get the AI to control your computer and basically do anything for you in the background. And if it's struggling with anything, it'll actually ask you, but if you want to see what it's doing and what it's creating, it'll create it. And if we have a look back here, you can see that it's now created the four posts that we can approve before I actually post it. Now, what we could also do here is we could edit the post before we go live.

So we can say, add a CTA to get more training on chat GPT inside the AI profit boardroom, and then I'm going to add the link so that this post converts for us and now we can edit any sort of computer use that it's doing in the background as well. So now it's going to use computer use again. It's going to revise the post itself. And we can see that it's now using our Chrome browser.

So it's planning it out. It's going to decide what to do. And now it's actually creating the draft list. If we scroll down, it hasn't edited it yet, but it's going to add that in as you can see here.

So it's planning how to improve it first, and then it's going to add it in and we can see it working in real time here. So this is chat GPT using my browser. And it's actually added that CTA in as well. And then if we go back into chat GPT here, we can say post it.

Now it's actually going to create the social media post. Now, in the meantime, that's how to use computer use. And you can see here that we're actually at the plan. So it's created the plan step-by-step for our full SEO agency setup.

Very, very specific, detailed, great plan. And, you know, if you're using a plan system like this, it takes a bit more patience, it's not that fun, but if we look through, like it's so detailed, that it's going to create better outputs for us for sure. So then we can say, yes, implement this plan and chat GPT is now going to start creating for us. Now, also, if we have a look as well, we've talked about computer use.

We've talked about, for example, how to create websites and tools. We've created a plan system here and how to switch models. So you've learned a lot today already. We also talked about plugins and how to connect external apps to chat GPT.

So it can build and automate anything for us. Let's try something else out now. So if we go to the presentation section, we can click on try now here and we can say, okay, presentations create a slide deck about the latest AI automation news. This week.

And that's going to start building out here. One thing to be careful of is that when you're creating these projects, if you go to a new task here and you try and give Codec something new to do, you want to make sure that you actually select separate folder if it's a new project. So we can start from scratch or we can use an existing folder if you want to edit an existing project, but starting from scratch here is useful. And then if we were using a new task, just be careful that you don't overwrite anything that's currently being done inside the conversation.

And the way that you do that is by adding a new folder for every new task you do. And also the cool thing about all of this is that we can build in parallel. So we can have the presentation being created over here. We can have the Goldie agents website being built over here.

We've got the AgentOS that we can improve as well. And we can scroll down our list and see all the projects that we're building and build them all out in parallel. So we can easily change things around and we can have different agents working in parallel. This is really good because for example, if you saw before, like it took 20 minutes to get Codecs to just create the plan.

And so you don't want to be like prompting and going, right. Make sure that you do this and then, you know, babysitting one chat at a time. If you want to be really productive with Codecs, you would have multiple projects coding out at the same time, updating each of these chats. So you just need to be careful with your usage and be careful that you're not absolutely rinsing the tokens that you have as well.

Now, if we go over to this chat, this is a really interesting workflow. So this is basically what we've done here is we have asked chatGPT on a daily basis to manage my Gmail every day so that all my emails are drafted before I wake up. And this is probably one of the most useful tools, I think, for anyone out there. Basically what we're doing here is we've used the plugin, which is Gmail, and you can easily set that up in one or two clicks.

And then we've allowed Codecs to manage our Gmails, answer any questions that we have, answer the emails, and it runs on a schedule every single day. And the other cool thing about this is now that we've got Gmail set up, you can send emails for me any single day, right? So for example, if we are creating a presentation slide deck over here, we can actually ask, once that's done, please can you share the slide deck in the email to juliangoldy at gmail.com. And so if you were, for example, like creating a presentation on Google Drive using the presentation plugin, you can actually send that email to juliangoldy at gmail.com.

Once that's fully created, you can then email a draft over to your team or over to anyone else, and it just works directly there as well, right? Really cool. Then we have the website being created, as you can see over here. And if we go inside this section, you'll see that this is running inside a new schedule.

So it has this automation ID, has a prompt, and this is running on a loop without me. This is something called agent loops. You might be wondering, what are loops? So because we set up a workflow where the AI agent could draft my emails daily and it can answer them and everything else, this is now running on a scheduled loop so that every single day it does this without me.

It's a scheduled task. And so anything recurring that happens day to day that you don't want to handle, then the great thing about this is that you can get the email to the right person. That you can get the AI agent to handle it day for day. So if you create content, you can get codecs to create content for you daily on a schedule.

If you have emails to answer, you can get codecs to answer emails on a daily basis. So coming back to this section here, it's like, yes, I'm going to finish the verified deck. Once that's fully finished here, I'll show you how to create a scheduled task. If you're wondering what scheduled tasks are, basically these are tasks that run on a loop every 24 hours.

So if we look at this one, the daily Gmail drafts, we can see the full project over here and this is a daily looped schedule. And also what's really cool here is, you know, this is not just like a fun automation or not just like new vibe coding. These are like real work automations that you would use every single day. So once we've connected, for example, this center here, we can start a new chat and we can say, send an email to Julian Goldie just saying hello.

And we'll say via juliangoldie at gmail.com. So if we hit enter on that, you can now send emails at any time using the Gmail skill. And so every time that you set up a new skill, like we've shown before with presentations or Gmail, these plugins that you connect make codecs more and more powerful. And so this is a system that improves every single day and really levels up how useful codecs becomes.

So now we've sent the email, as you can see here. If we actually go over to our emails, we should be able to see the email inside our inbox, as you can see, and we've got the email over here and it says hello. And that was super simple. But sending emails is something you do every single day.

So it's an easy setup. We've also got the social media post here for the school post with how to do social media with chat GPT. And so, so far you've learned how to build websites, how to build automations that run on a schedule, how to get codecs to connect your emails to send emails for you, and how to manage multiple projects at the same time, plus create presentations as well, which is fantastic. So that's going to work in the background.

We've got the agency website being built over here. Once that's fully completed, we'll be able to check it on the right hand side. We also have other cool stuff here. So for example, this is a system that we've used for creating a habit tracker.

So I'm going to say, open this habit tracker up. And what we did to build a habit tracker, once we wait for that to be set up, is we said build a prototype habit tracker that's fun, easy to use, gamified, et cetera, includes beautiful UI. And you can see that's all set up over here. Now, this is basically a powerful tool that allows us to create like mobile apps.

So if we have a look at this, for example, this is a setup where we've created an app where people can log in and they can use the app that we've created on a mobile. mobile device, and again, you just describe what you want. I think it's so easy now to build out projects with Codex because you don't need to be great at prompting. You can have everything run on a loop.

It works so smoothly with GPT 5.6 SOL. You do get a decent amount of tokens with GPT 5.6 SOL as well. And just everything I ask it to do, it runs super smoothly. I mean, computer use a year ago was terrible, but now creating a school post and then editing the school post.

It literally takes a few seconds. It's so incredibly easy. Setting up scheduled tasks used to be a nightmare with something like OpenCore. You know, sometimes it breaks, sometimes it would stop working.

Sometimes OpenCore itself would break with a new update, or, you know, you just run out of API tokens. Now you can set up these scheduled tasks as you can see, and they've run really smoothly here. So let's see where we're up to. We've got the 8020 of the webpage here for the Goldie agency website being built.

If you go to the presentation that's being created and you can see here that it's creating the presentation. We also have the habit tracker that's being opened up as well. And we have the social media posts on here too. And I've talked through how to use the Gmail automation as well.

Now there's some other cool stuff here. So you can actually create videos using Codex as well. Let's have a look at this one. So this is a video that was set up using Codex and I'm showing you how to create videos using Codex as well in a second, but if you have a look at this website, you can basically get Codex to create promo videos for you.

And it's create this fully animated video. As you can see, in a really simplified way. And you don't need editors anymore. This literally replaces a whole video team.

And all this takes is teaching one single open source skill to Codex. And then after that, it can create videos for you forever. So this is a 30 second video that's created. Looks beautiful.

Looks great. It was actually based on a website that we created with Codex. You might be wondering, how did we do that? So we said, we took a open source skill for creating videos.

And this is called Reemotion. Reemotion basically teaches AI agents how to create videos for you. So you can use this for social media content. You can use this for promo videos, for ads as well.

Anything that requires content in video format, you can use this for. And it's called Reemotion. It's an open source skill and basically allows your agents to create videos for you in HTML. So what we do with this is we just take the link to the GitHub that we want to install, which is this one.

And if we have a skill we want to install, we can go back to ChatGPT. And we literally just pasted that in. So we pasted in the GitHub link. We said, based on the skill, create a video about the airport for boarding for our website.

It created the 30 second promo. It gives us all the details of that and it's created it locally for us. Now, if we wanted to create a follow-up on that, we should create a new promo. So we could say, okay, now create a another video.

And this time the video should be about GPT 516 Sol, how useful is, how to use codecs, et cetera. And because we've taught it that skill, that is fully set up and we can now ask it to create another video over here. Now, if we go back to the habit tracker, that is ready to go here. So we can open this up inside Google Chrome before that ready to go as well.

So we can build apps with it, mobile apps. We can create videos with it too. We can build agent operating systems. You might be wondering like, how did I get Codex and Hermes set up with this whole system here?

That was very simple as well. So one of the beautiful things about AI agents now is like, there's so many of them and they're really useful and you can have all these different choices of like Hermes agent, OpenCore, Codex. The biggest issue is that there's no real way to tie them together inside a simplified way, because a lot of them are running locally. So how do you fix that?

Well, you can build your own agent operating system. That's what we did. So the way that we did this is we just said, okay, now create a mission control dashboard, make it clean, make it beautiful, make it interesting to use and make sure, and test it and make sure it actually works. Now, if we go inside the center for our agent OS, it's got the mission control dashboard, this was actually based on a template it already was.

Then we have Codex and Hermes agent and the image generation tool over here. So if we say create an image of a fairy flying through the sky, we can hit generate image and that is using Codex, the app that you've got over here, but it's using it inside a system where you can add your custom workflows. So, you know, for example, a second ago, we had that email creator, that system where we could send emails, automate our email inbox, et cetera, where you could actually build a custom workflow inside the side panel here for all of the automations that you create. So for example, you could have a email agent over here and you could have the image studio.

If you were using OpenClaw, you can install that on the side panel as well. And if you were, for example, using Codex, you could have it on the side panel, anything that you want to add inside here. And so if you want to see like a really advanced version of that, that I've been working on for hours and hours, we've got Codex over here, but we also have a system where we can switch between three APIs using OmniRoot and again, you just build out by asking Codex to build this out. It's as simple as that at that point.

There's no secret here and we can switch to OmniRoot. OmniRoot is a free API that we can use inside the same agent harness, which is Codex. And so we can switch between these different APIs. So if I run out of tokens on GPT 5.6, no problem, mate, we can switch over to OmniRoot.

If we run out of tokens on OmniRoot, the free API, we can go up to HY3. And so we can use Codex in many different formats. You might be thinking, okay, you can use Codex just inside the chat GPT app. It's so not true because this is a CLI.

What that means essentially is like you can customize it in the format. So you prefer, so you can have Codex CLI, but presented as a different UI, like you can see right here. This is still the chat GPT app. It's just simplified and customized in the way that I want to use it.

And that's something that's a bit more advanced, but it means that you can build out custom workflows, like creating images in a way where you can basically build and automate anything, you know, so that it's unique to you. So that it's customized to you, so that it has all your workflows one click away. Now let's come back to Codex and we'll see right now, we've got a lot of Now let's come back to Codex and we'll see where we're up to with the presentation. So that's still being built out here.

And one thing you'll see is actually quality controls its own work before it gets back to you. So this is just running on a loop. It's self-improving its own work before it sends it to you. And this is another beautiful feature of Codex.

So back in the day, or even if you're using chat GPT directly on chatgpt.com, and I don't recommend this anymore because, you know, using chat GPT, unless you create like a quick image, it's not that great for quality controlling. And you're still prompting it. And I think this is a very old outdated way of using chat GPT. So if we open up the sidebar here and we were asking it to create something, it's not going to self-check its work before it sends it back to you.

It's just going to answer the question and then come back to you. And the problem with that is like, it takes a lot of time to go back and forth, back and forth, back and forth, and you re-prompt it and babysit it. Not fun at all. By the way, that's the image that we created just a second ago.

Whereas for example, if we go to chat GPT here, this is something that's called a loop. It is self-improving its work and it's going to make sure it's up to the right standard before it sends it back to us. So it has to be really high quality before it can actually improve the work. Which is really, really useful because it can just work in the background.

We could have, for example, chat GPT and Claude both working on something that we come back to later whilst we're doing something else inside Chrome. Now also what you can see inside Codecs here is that you have something called PET. So when we are building out with Codecs, you might have multiple different conversations, but you're like, I don't want to just sit here waiting for everything to get done. And so you can still go and do your work without enabling all the notifications from all your apps.

So this was a big problem before. It was like, let's say for example, I'm using my agent OS inside Chrome. And then when I'm using Chrome, we, for example, don't know what's going on inside chat GPT. We don't know what's going inside Claude.

And if I wanted to have notifications, I'd have to allow notifications for everything and that's super annoying. You know, I'd have to switch out of do not disturb mode and go back into allowing notifications across all my apps. I don't want to do that. It's someone who's busy.

Someone who's focusing, someone who's getting stuff done. And so what we can do instead is if we go over to Google Chrome here, we can have the PET that we can flick back to with all of our conversations. So you see where it says create border and video, we can click back to that conversation and we can review what's been done over here. If we want to, for example, go back to Chrome here, we could have a look what's going on with the current status of the presentation.

And this also saves you a lot of time because then you don't have to like scroll through looking, you know, where's, where's that last conversation I was just working on. Makes it way easier to just have the PET working in the background. And then it's not interfering with anything you do. It's not slowing you down and you can just click back to it.

Underrated feature that most people don't use to be honest, but very, very good. Now, if we go back to the video that we were creating over here, which is this one, we can now see and review the video that we created, right? So I'm just going to say, okay, let's have a look here. We've got the video.

This is the GPT 516 explainer that we just created a minute ago. And literally all I said, as you saw, it was a very, very simple prompt. So. If we open this up and we'll start playing this, it looks really cool.

Right? So it's created a full presentation that's fully animated. It talks about all the key elements of GPT 5.6. What it can do, how it can be used, what it works with, et cetera, is branded as the AI profit boardroom.

And this literally created the whole thing, the script, the animations, the design, the editing, put it all together, and it did it in the background. And 99% of people don't know how to do this, but you've just learned a new skill where you can fully automate videos in like one single click. You can also change the aspect ratio. So you could say, okay, now create this in portrait mode.

So if we do that now create this in portrait mode and we can get codecs to basically edit the video and make it vertical plus add more cool effects and animations, so the video skill, I think is really, really powerful. You also see like, if you go onto like somewhere like fiverr.com, you'll see a lot of freelancers who are still selling animated videos. So this is something that's in demand. 99% of your clients or people that you're working with don't even know about this.

Let alone using, like 99% of people don't even know how to use codecs, let alone build out animated videos and stuff like this. So if you look at all the automations that we're building today, there's a lot of businesses out there who would absolutely love someone to come in and implement it for them. So if you're watching this and you're like, well, you know, how can I get the most out of the skills that I learned with codecs, all of these workflows, the GMO automations, the videos, building out a dashboard for an agency, building out, for example, the agent OS, creating the presentation, content creation with social media and the computer use automation on school, these are all examples of automations that you could actually sell to a client. So now we have the PowerPoint presentation created by codecs, and it's actually emailed that out to our emails, as you can see here.

So if we go over to GMO and we'll find the email that's created, let's have a look. There we go. So it's created the presentation and it's attached the slide deck. This is something that you imagine, like if you went back five or 10 years ago, and I know this person because I used to be a marketing manager, I would be like, I'd be quite scared at this point if you're not using, if you're not learning how to use AI, right, because at this point I can create 10 or 20 different projects with codecs, it's fully automated, it works first time round, the quality of it is better than I can do.

Like if I had to create a presentation that was as nicely designed as this, like it's clean, it's got beautiful images, this might take me a whole day, right, to think, figure out the content, how to structure it, make all the decisions around it, do all the research, put it into a nice presentation, format it, and then email it over. Like even basic things, like, I don't know if you used to send emails back in the day, but quite often you'd forget to attach them, like nuts. Like people would send out an email and they're like, see the attached and there's no attachment. You remember when that was a problem?

That sort of stuff has all been replaced using systems like this. And so you can be 10 times more productive. You've got an advantage that most people don't have and you can get way more done. So if we have a look, for example, inside this system, we've got all the research that's plugged into here, it's created a beautiful presentation.

And the other thing that I would say is like, the more tools you connect to Chatsheepity, the better and the more powerful it becomes. So if you look at this in a simplified way, it was like, create a presentation, then link to presentation skill, then learn how to send emails. And then send the email and attach everything. So it was actually like a multi-step project, like an AI agent, but it's done it inside one quick prompt.

So this can do, this is capable of doing very complex stuff, but doing it in a way where it feels easy, where it feels effortless. We've now got the video ready to go. This is a vertical video, so it really didn't take long as well. And we can just open that up.

So if we click on, open that up in QuickTime here, we've got a video that we just generated. And that has now created a beautiful vertical video. So it's adapted all of the content, but it's turned it into a, you know, a responsive version of it as well. Again, just one prompt and it does it perfectly.

It's unbelievable at this point. And the presentation itself looks super nice. I honestly don't think that anyone in my editing team could create something as nice as this with the animations. And Cheapty 5.6 does it.

Now, one thing that I want to say here is that when you are running any of these workflows, when you create a website, whatever, you want to make sure that you save this skill, save this workflow and build out a positive feedback loop. Now you might be wondering, okay, what is a positive feedback? So essentially with any of these tools, the more you use them, the better it should get if you train it in the right way. So what we can say here is create a skill for video generation based on what you've just learned today, and then also make sure that every time you create a video, you watch the video and you reflect on how you can improve every time and make sure that's part of the skill as well.

So that you become a self-improving feedback loop. And so that I don't need to quality control you. You just get better by yourself. You become more autonomous.

So we're going to build out that skill for video generation now. And what that means is this just improves and gets better on its own. And for every single workflow that you use that's repeatable. So for example, a lot of people create video content on a daily basis.

You want to think like this. You want to think in terms of how can we set up a loop, not a prompt? How can we set up a skill that gets better, not worse every time we use it? The great thing about this is that you can actually set up a loop that is that every time you ask it to do something, it's going to reflect and be like, Oh, we could do this, we could add that, we can make this better.

And so this just turns the whole of Codex into a much more autonomous loop. So we've created presentations. The website is actually taking quite a long time, but that's okay. Cause it's all running in the background.

We've automated our emails and I want to show you as well, like how you can create a schedule task here. So we were talking about schedule tasks before. This really takes Codex from something that you build stuff with to something that is an AI or agent that's far more autonomous. So in the same way that you can build out loops here, we can say, okay.

Create a schedule task where you research all the latest AI automation news in the industry, and then you turn that into a video and do that on a daily basis at like 3am whilst I'm sleeping. So that when I wake up, you have the video ready to go and I can just check it out. And so if you're doing, for example, social media content, if you're doing videos, you could set up this workflow. If you're doing, for example, some social media posts where it's written, you could do the same thing, but you just change the instructions and you personalize it to you.

And this works for any industry, not just AI automation news. That's just an example because we have an AI community. So I want to show you an example of how I would personalize it to me and how I'd actually use it in reality. So we're going to hit enter now.

And whilst that's running, it's designing the workflow as you can see here. And we'll also see what it's working on here. So it's now building out a self-improving video workflow, but it's also going to run on a schedule and when that's being built, it will actually show up inside our scheduled section over here. So we have, for example, all of these scheduled tasks.

We can also add to, we can, we can delete any of the scheduled tasks that we've built. We can run them now if we just want to test them out, we can pause them and we can delete them. So you can easily manage this. If you set up a scheduled task, one thing that I'd be careful of is that if you want a scheduled task to run, you need to make sure that you have the Codex app open and your computer on.

So if you don't have your computer on, or you don't have Codex open, the scheduled task will not run in the background. So this is really important to remember, because if you don't have that, you'll wake up in the morning and be like, why didn't that run? You know, why is that not working? Why is that not working?

And now you can see that the scheduled task for videos has been set up and it says next run is 17 hours, it's going to run daily at 3am and then we can open this up and we can actually see the full details of the scheduled task. So it runs inside a new task, it's going to open up inside this project folder. We can change all of these, by the way, we can edit them at whenever we want. And we have the, we have the scheduled task, we have the scheduled task, we can change all of these, by the way, we can edit them at whenever we want.

And we have the reasoning mode here and we have the model that it's going to use. Also, if you want to hide the pet at any point, you can just click on close pet and then you can go back to your settings and show it whenever you need to. And then we can set the frequency. So do want it to be weekly, weekdays, weekly, custom.

When's it going to run out? So what time do you want it to run? And then if we have a look, we can review the skill itself. So if we have a look here, it's, it's tagged in the skill, autonomous video producer skill to create a finished video.

And then it breaks down the skills here. We can edit the skill and we can also see where it's going to be saved. So it's going to be saved inside this folder locally each day. And then it's going to add the date to the note here.

And we'll basically automate video automation forever. There's also some interesting stuff here. So if we have a look inside the plugins, you might be wondering, okay, what else could you connect this to? It's just, it's so easy and so much fun to do.

And I would agree with you by the way, like automating things with AI has never been so smooth and easy. I feel like this. which is probably the best update that's come out from ChatGPT in years in terms of GPT 5.6.0 that just dropped recently as of creating this video. So you can see all these amazing things.

You've got like entertainment topics here. Let's have a look for social media. I want to see like what sort of tools you can connect to in terms of actually posting the content. So, yeah, we've got the social hub that we can connect to.

Let's just have a look as well over here. Let's see, we can link it to. Yeah, there we go. All right, cool.

So you can link this to Chrome, to computer use, to spreadsheets, presentations, data analytics, all sorts of cool stuff here. Website is still being built out, which is useful. And. What else would I recommend checking out here?

So you can also link this to GitHub. So if you're working on GitHub, if you're creating something open source or even if you create any of these projects, you're like, you know, I want to make that open source. I want to share it with the world. GitHub is a really good place to do that.

So if you go over to GitHub here. What this allows you to do is like create projects inside Codex and then share them with the world. So, for example, like Hermes agent, obviously it wasn't created with Codex, but it just like anything like this, it can get really popular. And so when you share your projects with the world, you know, number one, like more you can connect with more people.

Number two, like some of these projects go really, really viral. And number three is a great way of like testing out what you've built and seeing what people's feedback is like on it. So you can see how many people started and favorited it. You can see what's trending.

Also, another really good idea here is if you're not sure what to build, but you want to work on something really amazing, have a look through the trending GitHub and see what tools people are building and use that for inspiration. And there's all sorts of amazing projects. You can see what's trending each day and you can use that to come up with new ideas in terms of what you want to build next, which is great. Now, one thing that's also really useful is you can connect Codex to a second brain and a second brain is essentially your brain turned into a series of projects and files and folders that are very organized that you can come back to later.

Now, you might say, why would you want to do that? Well, the thing is that Codex doesn't have a lot of insight into the details. It knows what you worked on yesterday. It knows what you've built with Chatsheepty, but it doesn't know what you've done inside Claude and it doesn't know what you've done inside Herman's Agent and it doesn't know what you're working on day to day.

So what we've actually created is something called a memory galaxy. And this is fully visualized into a second brain. So this is the brain and the galaxy and all these are different memories and different markdown files. It's beautiful as it looks.

It's literally just a file that looks like this and it has everything stored. Now, if you're wondering how to do that, you can go to Obsidian MD. This is what I'd recommend for storing your second brain. And it will basically create all of these connections in the web on every single detail inside your life.

So you can get Obsidian for free here. And then once it's set up, you can go over to Obsidian. So let me open that up for you. And this is enough positive feedback loop that makes Codex really, really good.

So if you look at my database here, you can see that we've got all of these different projects that we're working on in a knowledge graph view. And then we can also, for example, see all of the details in terms of what we're working on and who we're working on and who's in our life and that sort of thing. So if we go to this knowledge graph, you can see that it's color coded. It looks beautiful.

And it's got details on all of my projects. Plus, I can see how it all links together. So if we have a look at the chat GPT module here, we can see that that links to multiple other projects it's actually built for us. And so this helps you sync the context, the details and the memory of everything you work across every different project.

So if we go to a new task here and we're like, OK, connect to my local Obsidian folder and give me some ideas based on my Obsidian folder of new keywords that we could create SEO content around. I'm going to create a new project for this. So create a new project there, save that. And then we're going to get that to connect locally to my Obsidian folder.

So that it can look at my knowledge base and instantly it understands everything that we've been working on. So if a new tool comes out tomorrow, let's say, for example, you'll be doing loads of cool stuff inside chat GPT, but then Claude comes out with a better update. You're like, you know, I want to switch to Claude, but I don't want to worry about losing all the progress I've made. No problem.

You can take everything that is inside the brain of Codex, all the activity that you've got over here. And because you've got Codex to update it inside Obsidian, you can then transfer that to Claude. So you can say the same thing, connect to my local Obsidian folder and use that for your memory. So you can see here that it's found the vault, it understands everything about us, and then it's going to come up with the ideas for our keywords.

This is also really useful because if you're struggling for automation ideas, you can go inside a new task here using that same project for Obsidian. And we can say, based on my Obsidian folder, what do you think I could automate with Codex that would save me a lot of time? And again, it can look through your second brain and give you personalized, unique, relevant ideas of what you can automate as well. So it's just so much potential to this.

Now, if we go over here, we've got the website that is ready to go. All right, so this is the website that we've built out. Let's have a look and open it up. I'm going to say, open it up and preview it for me.

And we can get that. If it doesn't show you like a visual representation of what's being created and that sort of thing, then you can actually click open it up or you can allow it to preview. And sometimes if you're building out something locally, particularly with Next.js, it has to create a local environment. And you might be wondering, OK, I don't want to do that.

That sounds technical, blah, blah, blah. No problem. Just get Codex to do it for you. So we said, open up, preview it for me, and it's going to open up that local build and then it's going to preview it for you.

I don't have to do anything. It's just working in the background, doing everything for us. So that's how it works. Now, at this point, you're probably like, Julian, there's so much going on here, right?

You know, you probably haven't learned so much today if you're brand new to this sort of stuff, because you've learned how to build a responsive website, which we've got over here. This is the dashboard and the portal for our SEO agency that we started off with. We have the video creation skill over here. You've learned how to create your own builds and video.

You learned how to create your own skills and also connect to different apps. You've got the video. Over here, you've learned how to save skills, you've learned how to schedule tasks, how to automate your emails, how to create an agent operating system, how to build apps like the habit tracker, basically mobile apps, the website and everything else. Now, one thing that's really useful is like, you're going to work on stuff and you're just going to forget it's there or you're going to struggle to find it from this whole list.

So if you go to the search icon here, you can search across all of your projects, everything you've built, and also all of the custom sort of pre-made tasks. So suggested, you'll see new task, open folder, go to your search files. Then you've got, for example, like you can have project list tasks. So you don't need to create a new folder every time you create one, every time you build a new task.

You can archive stuff as well. So if you don't want to see some of this stuff, you can actually archive it and that means it would just disappear. But it would still be saved locally. It's just like you don't see it inside your chat history and that sort of thing, which is great.

Let's have a look at this website and what we built. You see how it's all customized for me. And also if we open that up in the browser. So I'm just going to take that like so.

Let's just make sure it actually works. Open it up. And if we go back to the setup, just open it up in Chrome for me, we can preview inside a browser, not just inside here. The one thing I don't like about Codex is like when you're previewing the browser, it's useful for like a quick glance, but you can't see it in full screen mode.

Whereas, you know, if you open up inside Chrome, it's going to be way easier to see and everything else. But yeah, that's basically how to build and automate anything. We've talked about so much. Schedule tasks, video creation, social media content, computer use, app creation, website creation, building mobile apps, scheduling tasks, and using all those elements inside one workflow.

So, for example, like researching the latest news, then creating a presentation about it, then sending the email and actually attaching the presentation as well. So there's so much stuff that you can do with this. And then. If you go to our Chrome browser here, we can view that project that we had a look and here we go.

Look at that beautiful design, super clean, doesn't feel like AI at all. It just looks super nice. Modern, interesting. It's actually taken some example case studies from the presentation that we added and then even added some quotes.

This is a real testimonial, by the way. So it has actually taken the quotes and it feels and it looks like our website. I mean, how beautiful is that? And we created that in one single click using that whole system that I showed you before.

You plan it and you build it. And that's how you can build something this unbelievable way. Also, if we check the links here, it does actually take us through to the right page and if we click on build my campaign, we can actually see what the page looks like here. Beautiful.

Unbelievable, actually, that worked out way smoother than I imagined. In fact, everything I've shown you today has worked out really, really smooth. Now, if you like this sort of stuff, we actually have an AI automation community you can connect with me personally. And inside there, I actually answer any questions.

So you're going to start building with this. Then you're going to have questions. Something's going to break. That's totally natural.

So don't worry if that happens. If you have any questions, you can post inside the community of the AI Professor Boarding. Link in the comments description and go to the AIProfessorBoarding.com and inside the community, you can ask questions and I personally answer them with a video tutorial every single day. And then what I also do.

Is I actually record a tutorial and also the whole community answers as well. So there's always people online 24 seven and it's just a really positive community full of people who are awesome, learning, very positive, sharing their journeys as well with AI automation. Something that's actually happened recently is like people are actually building out automations for people. So three people asked for this automation for like automating LinkedIn follow ups.

Sheena built it for everyone, then created the whole automation and shared it with the whole community. Like it's unbelievable. And just the positivity and the vibes inside this community as well. Inside the classroom, you can get access to all my best training.

So if you like this sort of stuff, you're probably going to like the ChatTuberty 5.6 one hour course we have here. We also have a full one hour course in building agent operating systems so that you can build these AI automations like so. And also, this is just an amazing system. It's called the Agent OS we've built to get the most out of all of your AI agents.

So if you're using Claude, Hermes, Codex, if you want to use free APIs inside Codex, which we actually give you a full system for this, as you can see, with your memory system plugged in. And then we have all these custom automations, like a loop engineering system, an SEO automation, a video agent, a music agent, game studio, and even notebook on end plugged into the same system. So you can get the most out of this stuff. So you get so much inside the AI Profitable Room.

If you go to the Agent OS, you can get the new system here with the video tutorial, the last update date. And this is the file for installing it. Inside the calendar, you can actually jump on four weekly coaching calls. So you can share your screen, ask questions in real time and meet other community members who have been doing other stuff.

Inside the map, you can connect with people locally who are building with AI agents like you, using Codex, et cetera, and that's all available inside the AI Profitable Room. Plus, you can direct message me and answer your questions too, as well. So feel free to get that link in the comments description or go to the AI Profitable Room dot com. Thanks so much for watching.

That is our Codex course, one hour, how to build and automate anything.