Hey, everybody, and welcome to the course. I bet I know exactly why you're here. You want to add AI power to your view applications, right? After all, that's basically the name of the course.
Well, the way to do that most easily, at least in my opinion, is with the AI SDK from Vercel. With it, you can create AI chat interfaces, you can add AI text generation and image generation to your apps, you could add AI powered autocompletions to your inputs and text areas, and a whole lot more. Best of all, the AI SDK allows you to work with many different AI providers without having to learn the interface for each of the different providers APIs. It has support for Gemini, for OpenAI, for Grok, for Anthropic, plus a lot of other less well known models.
Let's take a look at the AI SDK website for a quick overview of what they offer. Now, do note before we start that this course will be using version six of the AI SDK. At the time of this recording, it is still in beta, but it should be reaching production fairly soon. That being said, we wanted to get ahead of the curve and teach you the latest and greatest.
But if anything changes between beta and the stable release, please let us know in the comments so that we can update the video. The four core features of the AI SDK are put on display right here on the homepage. The first and most important is kind of what we just talked about. Instead of juggling different AI APIs, you get to write one set of functions, and then switch models with a tiny configuration change.
This allows you to test different models for different use cases, easily switch out between them and saves so much time. Next, generating predictable and beautiful UIs with the AI SDK is a breeze. Allowing you to elevate user experiences with AI beyond just simple text. Feature number three is its framework agnostic.
Even though we're using Vue and Nuxt for the course, the SDK also works for React, Svelte, Node, Deno, basically whatever you want. And then of course, finally, they support streaming, text streams, JSON streams, streaming data to power your generative UIs, you name it. The SDK makes it incredibly easy to stream both unstructured and structured data to your front end. Okay, now let me walk you through exactly what you'll learn as we progress through the course.
Number one, you will learn the essentials of the AI SDK backwards and forwards. That is, you'll learn how to set up the SDK and then choose models across different providers. You'll get hands on generating text, images, JSON objects, even custom UI elements. I'll teach you how to track token usage.
So you know exactly how much each interaction costs. You'll see how to stream responses of both text and structured data. You'll use tools to get the AI superpowers that can be called from both the server and even the client. We'll build agents with dedicated system prompts, looping tool calls, and stateful reasoning.
And yes, we'll even connect the SDK to an MCP server. That is a model context protocol server for plug and play tool driven workflows. Once you've mastered the AI SDK itself, this course is specifically designed for V developers. So we're going to get hands on integrating the SDK with a Nuxt application.
Mainly, we'll learn how to use the use chat composable, which is a super simple way to create chat UI in seconds. I'll also show you how to persist chat messages across sessions. We'll display generated UI on the front end with dedicated view components. We'll build text inputs that feel AI aware with AI powered autocomplete.
Something like GitHub Copilot, but inside of your app's UI. Finally, once you've mastered the AI SDK, and how to integrate the AI SDK with Nuxt, we'll take it one step further with a couple of bonus lessons. During those bonus lessons, you'll learn how to deploy your AI powered Nuxt app to Vercel and get edge ready performance. Plus, we'll use Vercel's AI gateway offering to make sure your app is production ready.
This service provides key features at a great price. You get a unified API that even further helps you switch between models and providers. You get high reliability that automatically retries requests to other providers if one fails. You have support for generating embeddings to power things like search retrieval and other tasks, as well as a spend monitoring dashboard where you can see your spend across different providers in a single place.
There is absolutely no markup on tokens. They cost the same as they would if you went directly to individual providers, and you can even bring your own key if you want to. Whoo, that is quite a list of takeaways. I don't know about you, but I'm ready to get started.
Let's go.