NEW Claude Code Update is INSANE!

**Julian Goldie** (0:00)
And CLI has just been launched for Claude, and this is a really powerful way to use AI agents and manage them inside a Claude. So I wanna show you exactly what it works, what's different now, how to use it, et cetera, and how to implement it inside your Business Plus. A powerful way that brings it all together. So essentially, we've already plugged and CLI directly into our agent operating system. You can see a bunch of agents that we've created right here. And this is very different from using Claude Code normally. We'll come on to that in a second, but you can basically manage your AI agents in the cloud, which is very different to when, for example, you would use your terminal with Claude Code. And so essentially before, when you use Claude in, for example, at the chat, you know, you're a person typing into a box, which is great for you. You got one conversation at a time. But if you're a business, for example, you don't want to sit down and type into a chat. You want Claude to build and be built into your own software, doing tasks automatically as an AI agent without a human watching. And the API is how you do that. It's a doorway that lets you build an app or a system talking to Claude directly. So you don't need the chat window. You don't need someone typing. You can have a software that sends Claude a request and Claude sends back an answer with the API, right? So for example, if you're just using the chat directly here, it's kind of like you ordering coffee at the counter. Whereas if, for example, you use the ANT CLI, which is the new way of doing things, the API is a pipe that pumps coffee straight into a factory all day with no barista needed. So why would you actually want the API? Well, let's say, for example, we've got these agents set up. We set up a research agent, support triage, a code reviewer, and an SEO strategist using this whole new system. And so you might say, well, why would anyone want the API? Because they want Claude working inside their products or in their business at scale. So, for example, you could have a customer support tool that uses Claude to auto answer emails daily. You could have an app that summarizes all your documents or something that other people upload. You could have a company that runs Claude across all of your documents overnight to find patterns and trends. You could have a startup whose entire product is Claude wrapped in a nice UI. And so none of that works by typing into a chat box. It only works through the API. Now, if you wanted to use the API previously, the old way was that you had to write a program to do that, right? Maybe you have a Python or JavaScript or that sort of thing. The new way is the ANT CLI. So you can now just type a command in your terminal and the request goes through directly, which means like you don't need scripts, you don't need JavaScript or anything like that. You just literally tell it what you want to do. So the API is kind of like the doorway, but the CLI is a much easier way to knock on that door. Now let's talk about actual use cases here. So let's say, for example, you wanted to try a prompt against the API right now without building the whole script, you could type one command to see the result done. The automation can run itself. So say you have a system that runs every night on a server with no human around. Well, you can't have that person type in, and that's why you want AI agents in the cloud instead, right? They can run autonomously in the cloud using this managed service. And also if you are plugging Claude into existing terminal tools, you know, people who live in the terminal can now pipe Claude's answer straight into their other tools. So for example, feed it into a search, a filter, the next step in a chain. And finally for managed and managing automated agents. So you can set up a managed agent, which is Claude running tasks on its own, store its settings as a simple file with version control, which we've got right here, and use the CLI to update it or check exactly what it did step by step. Now you might at this point be going, OK, why wouldn't I just use Claude over here versus ANT CLI? All right, what's the difference? Why would you use one over the other? The way you want to think of it is like Claude Code is like an agent that can build stuff for you. So it could build a tool for you. You could say, OK, hey Claude, build this website for me. And then it writes and edits the code. ANT CLI is actually like a remote control for the API. So you could say run this API request right now and it actually does it. They're not really competing. In fact, they actually work together. So when you ask Claude Code to build something now that uses API, Claude Code uses the ANT CLI to do it because the CLI is simple, predictable. One builds, the other runs. Now the other thing as well is like you could upload files and use that across your managed agents. So let's say, for example, we've got these agents here. If you are just using Claude Code on your local laptop, well, then how do those agents in the cloud get access to files that you have? Right. Well, that's the power of ANT CLI. So you can upload files and you can also session trace, which means if you kicked off an agent session, you could send it an event and then pull the trace, right? And see all the reasoning steps that we've done right there. Additionally, you can manage all your agents in the cloud. You can explore the different models and you've got the raw API console as well. And so just to recap here and to really summarize it, the API is how your tools and your websites use Claude. So businesses can run Claude on their own tools in the websites. And even like, for example, your agents, your managed agents, to use the API, you have to send a request. The old way was writing code to send those requests. The new way is that you can type into a single terminal command and you use that command for like, for example, quick tests, managing your agents, seeing what your agents did in the cloud, et cetera. And we've actually built this into the agent operating system. You can see an example right here. So for example, if we type in models with the ANT CLI, we can pull in the models that we've used. If we check the author status, we've got here. If you wanted to see all of our managed agents, we've got those here, previous sessions as well.