Before you go into the Jupyter notebook and we build our blog post crew in production, let's talk about what will we do there? Because this is a slightly different from what you have done in other Jupyter notebooks. This one, you're going to create your first crew from the scratch. This means that you're going to start by using the CLI to create a new crew. Then you're going to understand the file structure that we're going to have there. You then install all the required dependencies that we have. And then we're going to be able to execute our crew. Now let's jump into the Jupyter notebook and see how we're going to build our crew from the get-go. Let's do it right now. So let's start by importing our initial classes. For this one, you're gonna notice that we're not importing to create classes because we are going to create a crew from the scratch using the CLI. We're still loading our environment variables though to make sure that we have everything that we need in order to execute the script. Now let's start by creating our project. Creating our project is as simple as running one single CLI command called crewai create crew and then we pass a name for our project. For this one we can call it New Project. So let's execute this. So you can see that we already got our initial folder created and all the different files that we need are already within it. We have a Readme file. We have a gitignore. We have a pyproject file that holds all of our dependencies references. We also can see or main files our crew files and even that custom tool example including the agents and tasks Yaml. So our project is ready to go. Now let's install our dependencies to make sure that our environment is ready to execute this crew. Installing our dependencies is very straightforward. All you gotta do is run crewai install in your terminal or command line. So let's go ahead and do that. Now that we run crewai install you can see that install all of our dependencies. Everything is already installed and set up and this crew should be ready to go. So, this might take some time. But now all our dependencies are ready installed and we're ready to execute this crew. Usually, we need to be mindful of adding our environment variables. Remember that we can be using a lot of different providers here different models. And you want to make sure that we have the right keys. For this, there is a dot env file within our project where we can put any keys that we would look like. For executing this in the Jupyter notebook, you don't have to worry about it because of dependencies already being injected for you, but this would be very interesting in case you're building it on your own terminal or your own computer. So now let's execute our crew. For us to run our crew are we going to do is execute crewai run. And once that if we do it we're going to see the logs of our agents working as usual. So let's do it. Executing our crew for the first time might take a while because this is running now on a virtual environment. But you can see our agents working task by tasks. From the get-go, when you're creating a crew from the scratch, it already come with a couple agents and a couple tasks just for you to get a feel on how to build some yourself. And this is a simple blog creation crew, where you start with senior data researcher that's going the research content, and then you pass it on for a blog content writer that will actually write the content. As you can see, writing here with the reporting analyst. So this is a very simple crew that basically do the research and then after that writes some content in a report about it. Now that we learned about this, you get a sense on how do we start a crew yourself. You can go in your terminal and just start building crews, and you're going to be able to have your dependencies. You're going to be able to push them to GitHub. You're going to be able to share them with your friends, and you got to be able to execute them yourself. And this is very powerful because you don't need to be building crews in the context of a Jupyter notebook. And this makes so much easier for you to get started with them. And the coolest part is that you can also use a command line interface to create flows as well. If you want to create a flow, it's very similar to the crew. All you going to do is type crewai create flow and the name of the flow and then will set up the initial architecture of flow for you. Creating a folder and everything that you need. Now let's take a look at the files that py created for us. You can see that the initial structure is very simple and very similar to the crew, but if you look into the actual flow folder, you now have a crews folder that allows you to have as many crews as you want to have in there, and then use them in your main.py that is your actual flow code. You also have a tools folder where you can put any custom tools that you want. This is very interesting and show how you can also use the same strategy to create flows if you want to. You can see how powerful this is. You now can not only build the crew from scratch and understand everything that is going on with them. You can create your agent, you can create your tasks, and you can also deploy this. And now that you have this API, you can use its endpoints in webhooks to actually connect this with any existing applications that you have. You can kick things off from Slack. You can kick things off from your other systems, and you can get this information back and push it anywhere that you need. This is how you can now bring the power of multi-agent systems into a production setting, where you can actually integrate this with any of the applications that you have out there. And this is so much valuable. I hope that you had a lot of fun in this class, because I know that I did. You can build so much. Now let's jump to the conclusion.