ChatGPT Work 1 HOUR COURSE: Automate Anything!

**Julian Goldie** (0:00)
ChatGPT Work, full course, one hour, how to build and automate anything. So today we're going to be running through exactly how to use a new app, which is ChatGPT Work. We can download it like so, if you want to get started, so you can get it from openingai.com/chatgpt-work.
And we're going to be walking through exactly how to use this new tool step by step, some of the best use cases for it. It's a really powerful system. Now that GPT 5.6 with Sol is built in. So you can do all sorts of amazing things with this app. Very impressive. One of my favorite apps right now for AI. And I'm going to talk you through exactly how to use it, how to get started with it. We actually created this because we had a question from Jason four hours ago, who was saying, you know, hey, I'm using the new ChatGPT app. Like so, I don't know how to use it like the old Codex app. And to be 100% clear here, ChatGPT Work is the new version of the Codex app. So previously, this was called Codex. Now it is called ChatGPT Work. And we can get access to it over here, as you can see. And this is what it's going to look like for you. So once you got started with it, once you've downloaded it, you just need to make sure you signed in. And once you signed in, then we can start using this. So the first thing that you're going to see is you're going to see the new task at the top, right? So you can see over here, you've got new task. And over here, you have chat. So there's two different versions inside one app. You've got the chat, which is like you can go back and forth inside the chat with ChatGPT, and then you can actually build stuff. Now, depending on what you want to do, you've got two different options right there. I'm going to walk through the Codex side of things first. So this is the new task section, as you can see. And from here, what we can do is we can just make sure that we have a new project set up. So what you'll see on the folder here, and you can click off it and then choose a new project if you want to. So we can click on new project, start from scratch, or we can use an existing folder. Now, you would use an existing folder if you wanted to start a new project. However, if you don't have something new, sorry, an existing project to change, then you can go to start from scratch. So we'll click on start from scratch, and super basic example, let's start with the basics. Let's say, for example, you wanted to build out like a basic app, and you can build apps, tools, games, websites with this, whatever you want. It's a very powerful tool. So from here, what we're going to do is we're going to type inside here.
Habit Tracker, like so. And you can see that we can switch between working locally or creating a parallel copy of our project as well, or we can work in a cloud environment. So work locally is you work on your Mac. So for example, I have a Mac Studio. If I click work locally, that means I am building on my Mac Studio. If we select new work tree, that is going to actually clone our existing local project, and we can start building inside there as a new version of our existing project that's been cloned that we can edit later. And then we have cloud. So cloud means that we're working in a environment in the cloud sandboxed away, and the agent can work directly there. That's good if, for example, you wanted to share the task or you wanted to sandbox the task. And you also see this usage remaining section. So what this usage remaining section means is that you have a limited number of tokens that you can use inside your app. And the more you use and the higher efforts you dial your model to, which I'll come on to in a second, the quicker you're going to go for your usage. And your usage depends on which package you're on. So if you're not sure which package you're on, you can learn more here. You can find out what codex plan you're on, and you can go from there. So what we're going to do now is we are going to tell it exactly what we want to build. And before we do that, we're just going to select the model that we want to use. Now there's two options for selecting the model. The model is the brain behind ChatGPT. You can change and customize the brain behind it as much as you want. So for example, if you wanted something really fast and lightweight that doesn't use many tokens, but probably wouldn't give you such great outputs, then you can actually change this to TeraLite. If you dial this up and you drag this forward, the higher it goes, the slower it will be to respond, but the better responses you'll get and the more intelligent the effort levels you're using. There's actually three models with the new GPT 5.6. So Sol is the frontier model, the most powerful. I would recommend if you're building something good, then you want to go with Sol. If you were just using the chat basically, then you go with TeraLite. So we're going to dial this up to extra high. I think for you watching this, high is probably good as well. And you can also switch to advanced. And if you switch to advanced here, you can switch between the models. So Sol, Tera, Luna, 5.5, 5.4. These are older versions of ChatGPT. And then we can change the effort level as well. And we can change how fast it replies. So for example, if you want this to reply fast, if I'm doing a demo, which I am today, then I will switch this to 1.5x speed, which means I'll use more tokens. But I'm going to do that for you just to help you and so you can see results faster today. But for you using this normally, you know, this can code in the background, it can work on multiple projects, and you can just go with the basic settings here, as you can see. So we've got the fast mode switched on. We have Sol, High, Setup. And then what we're going to do from here is we're just going to add, you know, we're going to start building this project. So whatever you want to build, you start going from here. Now, you can actually click on these different presets as well. So you'll see, for example, fix issues and failures. Good if you have something you're already working on that you need to debug. Then you have review code and suggest changes. So you can actually get the AI to look through what you're creating. And then it can go from there. And we'll have build a new feature, Apple tool. And then we have explore and understand code. So you can switch between each of these as you go along. So from here, I'm going to go to build a new feature app. I'm going to say build. And you see how we can build a feature, a UI, a prototype, an internal tool. We're going to build a prototype. I'm going to say habit tracker. That's fun, easy to use, gamified, etc. You can say little things like beautiful UI. You actually with the new models, with ChatGPT like back in the day, you'd be very detailed in the prompts that you give it. Nowadays, you don't need to do that so much. You can keep it simple and you can go from there. So we're going to hit enter like so. Little tip here, you can press command and enter, and that will just enter it for you so that you don't need to click this button over here. You can also control the AI with your voice. So if you click on to dictate, then you can actually speak with your voice and we can ask it to do things like that. It will need access to your microphone, but we can allow that now. And you see how it's recording my voice. And then if we click on stop, it will actually translate what we said. So you can see it's actually translating over here. So you can type with your voice as well inside ChatGPT. And now you can see here that it's going to ask a bunch of questions. So it's like, okay, before I build, I need three choices. And these are basically folks that help the AI understand exactly what you're building, how you want to build it, and if and else. So for example, we've got, where should the prototype run? And so what I'm going to do for here is we can set what type of app we want to build. So for example, if I'm building an app for the iPhone store, then I'm going to go with native mobile app. If we go with mobile first, this is like a web app, but it's going to be responsive first, and then desktop is secondary. And then if we wanted to build an app for our desktop, then that's going to be the fastest to test, but it's not going to be native to mobile. Now, if you're wondering, okay, what does native mean? Well, for example, if we have a look at the agent operating system that we've built with all our agents plugged in, and you can build stuff like this if you want to use in this system. Well, all these custom automations, they were built for desktop first, and then it can be mobile responsive later. But the point here is that when you're building with this stuff, you want to help it understand exactly where you are building for. And then we have the style to use. So you can change the vibe, the feel of this and everything else. So we could go with like dark neon, calm organic, pixel game, playful editorial, et cetera. So I'm going to go with playful editorial, and we can just copy that. Then we can go over here and we can say, okay, I want a responsive web app. And again, if you're building like a mobile app, then you could do that later by selecting native mobile app. And also all these things can be edited later. From here, we can switch between what type of prototype we want to build. So do we want an interactive demo? Do we want a functional local app? Do we want a full product foundation? Now, the difference between this is like an interactive demo is like, you know, pretty simple, pretty fun to use, quick to build. But it's not going to be as functional as a functional local app, right? Where you can, for example, like save stuff, you can create and edit things. And then we also have the full product foundation. That means people can log in, they you can have a database, you can store details for anyone who's using it, and they can save their progress as you go on. So, for example, you might start off with like an interactive demo. If you like it, if you like, well, this is good. You could build something out with that, then you could build later the full product foundation. For you, because I'm doing a quick demo in the course, I'm just going to go with interactive demo. So we're going to hit enter like so. Now that it's thinking and soul is going to start coding. Now, one thing to note, as we go along here, is that you see how it's got the full access here, we can change the approval. Now, some people watching this will be like, AI coding agent sounds great, but a bit scary that it's coding locally for me. What if it does some crazy stuff? So you can actually change the approval levels here. So you can switch between ask for approval, approve for me or full access. Full access means it's not even going to ask you for changes. That's usually the default that I go with because I'm an advanced user. But for you, you might prefer to ask for approval just so you can check things and understand things before you allow it to do anything else. You could say approve for me, right? Only ask for actions detected as potentially unsafe. That could be, for example, like deleting stuff, editing existing files. If you've got an existing project you've coded out, you might not want Codex to make any crazy changes. You might want to approve everything before anything goes live. So you can also switch with that. And now you can see that it's planning the whole strategy. So what you can see is it's reading all these MD files. MD files are just simple markdown files. Very simple way of the AI understanding skills. So for example, it can read this MD and you can click on it over here. And this allows you to understand what it is doing, right? What's going on here. So if we have a look, for example, at the typography, this is basically like a skill that helps ChatGPT build things out. So for example, we have the typography section over here and it has a prebuilt skill where it understands, okay, these are the principles of typography. This is how it works step by step. This is what it works with. This is what it should avoid, right? Mistakes to avoid. And it's basically like you level up the AI with skill sets. And if you have certain things that you want to train the AI to do, what's really good is you can create a skill.