Agentic Coding with Xcode

Note: this starter project is used as the basis for the rest of the module, and a final project is not provided until the last lesson. This is because what the Agentic AI did for me may not exactly match what it does for you!

Here, I have a project open in Xcode. This project is actually the final project for lesson one of the Swift Data in iOS 26 module here at Kodeco. What I want to do with agentic AI is try to get to the final project for lesson two of that module, which was adding some model inheritance with the new features for Swift Data.

Let’s take a quick look at what we’ve got here. This is a Swift Data recipe app. I’ve got a basic recipe model, denoted with the @model keyword to tell Swift Data this is a model. We’ve got a unique name for our attribute, we’ve got a summary, and we’ve got some instructions. The recipe needs ingredients, clearly, and so we’ve got a relationship for the ingredients, and then we have an initializer. We also have a healthy set of sample data at the bottom of the screen. We’ve got our ingredient model as well — pretty simple, just the name and the amount. And then we also have a RecipeDetailView that gives us a pretty good view of the title, the summary, our ingredients, and then some instructions on how to make it.

So what do I want to do with AI? How can I improve this app? Well, I want to get a little bit more specific in terms of my model. Recipe is fairly straightforward — it’s kind of generic. But what if I want to do something like a baked good that might require a certain size dish to bake in? Or, what if I want to make a beverage that has a particular glass that it should be served in? What can I do to enhance this project to include models that are based off of the recipe but have these unique qualities that differentiate them? And also, I might want to have a different type of detail view for my beverages or my baked goods. I could ask the AI to help with that as well.

We’ve had the ability since Xcode 26 came out to have conversations with different AI agents. Here, I’ve got a conversation window open for Codex, which is OpenAI’s agent, but that’s a lot of back and forth — especially with the basic models they have in here as well, like GPT-5. There can be some automatic insertion of code into your project, which is great. Up until now, it’s been really nice to work with, to get some additional code in with just a little bit of a description to the agent. But what if we really wanted a more hands-off ask for Xcode? Well, to do that, we need to bring in an agentic AI, and we also need to get some help from a model context provider, which we’ll talk about in the next lesson.