2026-02-05 • iOS, macOS
In this online code-along session, you’ll learn how the coding intelligence features in Xcode 26 — such as coding assistant and Coding Tools — can help you write code, generate tests and documentation, fix errors, and navigate codebases. To code along, you’ll need a Mac with Apple silicon running the latest release of macOS 26 and Xcode 26. The session will be conducted in English.
Speakers: Lucas Longo, Jerome Bouvard
Unlisted on Apple Developer site
Transcript
This transcript was generated using Whisper, it has known transcription errors. We are working on an improved version.
I'm Lucas. I'm an instructional designer on the education team here at Apple. I build resources for developers to learn everything from app development fundamentals to best practices on Apple platforms. Hi, I'm Jerome. I'm the product manager for developer tools here at Apple. We're streaming to you live from the Apple Developer Center in Cupertino. And I want to welcome everyone who's joining us online. Today, we'll introduce you a brand new Code Along Jerome recorded showing the coding intelligence features in Xcode 26. Yes, I'll show you how the coding assistant empowers anyone to turn their ideas into real apps, even if they're never coded before. And speaking of building apps, submission to the SWIFT Students Challenge 2026 will open tomorrow, February 6th. We're thrilled to invite students from all around the world to participate. It's a fantastic opportunity to showcase your skills and your creativity.
Now, we set up a Slido to answer any questions you might have during the code-along. Use this QR code to sign in. We have a team of Apple engineers standing by to help. After the code-along, Jerome and I will come back to address some of those questions live. Okay, now it's time for the main event. Settle in and get ready to see some magic coding intelligence.
Hello, everyone. My name is Jerome, and I'm a product manager here at Apple. Today, we're going to do something fun. Turn an idea into a fully functional app using Xcode and its coding intelligence tools. If you're not familiar, Xcode is how developers build apps for all Apple platforms.
It comes with powerful features to help anyone bring their ideas to life. Like previews to see your changes in real time. Simulator, to simulate different Apple devices and OS versions when a physical device isn't available. Advanced debugging tools to isolate bugs, find crashes, and identify bottlenecks. And tools like instruments, providing real-time insights to deliver a smooth and responsive experience. And now, with coding intelligence built right in, it's easier than ever to learn, experiment, and create. You can use models like ChatGPT or Cloud all without leaving Xcode. All right, let's build something together. Ready?
that great apps starts with a spark, and that spark can come from anyone. I have an idea for an app. Let me tell you about it. We're always looking for little moments of discovery, right? What if every day could have a small adventure, a micro adventure? I'm not saying climb Mount Everest. I mean, go to that museum you've been meaning to check out.
Explore a park you've never been to. Find the extraordinary in the ordinary. I wanted to build an app for that. I started by sketching a design. And then I brought it into Keynote and named all the interface elements in more detail. This planning helped me define exactly what I want to create.
And in order to get the best results from the coding assistant, I carefully crafted five prompts or instructions, which I'll go over in a little bit. Now, to follow along, make sure to find the link to these prompts in the session's description so that you can use them in your project.
While you look for the prompts, let me tell you something I've learned about prompt engineering. Telling the model what you want to create is a skill in itself. The key: be specific. Tell it exactly what you want, like you're giving instructions to a a teammate. And if you know the exact name of a component you want to use, even better.
Also, give it context. Describe what your app does or what problem it solves. Include things that might inform or influence what your app would look like. And iterate. Experiment with different versions of your prompt. Now, you can also use the coding assistant to help you create and improve your prompts, or even use it to brainstorm new features for your app.
there's a lot more to prompt engineering. And like coding, there are several ways of doing the same thing. So don't get discouraged if at first, your prompts don't give you exactly what you want. Now, let's get started building my app. I'll create a new iOS project in Xcode and then set up the coding assistant.
I'll now switch over to Xcode to create a new iOS project. I have Xcode open, and I'll click on Create a New Project. I'll select iOS as the template for my new project. I'll select app as the application type and click Next. I'll type in microadventures for the product name. I'll fill out organization identifier with com.sophysun, a fictitious organization, and leave the other options to their default values. Click Next and save the project to the desktop by clicking Create.
I now have the default Hello World project with the project navigator panel on the left, the source editor in the center with my code, and the live preview on the right. Now, I'm going to turn the mini map off here and make Xcode full screen for better viewing.
Next, I need to set up Coding Intelligence. To open the Coding Intelligence panel, click on the sparkle toward the top left of the navigator or use Command + 0 to open or close it. I'll click on Setup to open the Intelligence settings. You can enable ChatGPT or Cloud where these models are available, or use another model that you may prefer. Today, I'll use ChatGPT, so I'll click on Turn On. This panel has some more information about ChatGPT.
I'll then click Next, and I get the options to use ChatGPT with an account or without. For this cartelong, I'll go without a specific account. Thank you. I'll close the settings window so I can start using the coding assistant. The coding assistant is where you interact with the model using natural language. Okay, I'm all set up and ready to go. I have my iOS Xcode project and the coding assistant configured to use ChatGPT. And now, let's start the conversation with the coding assistant and explore what we can build and learn from the process.
So for the first prompt, I start by providing some context. I'm creating an app that gives ideas for quick microadventures. Then I describe the interface elements, start by adding microadventures as a large navigation title, and add a full screen map centered on a marker. not how I'm being specific by using a navigation title and a marker, for example. Now, I'll copy the first prompt from the web page, switch over to Xcode, paste it into the chat, and press return to submit the prompt.
While I wait for the response, here's a few tips. It's very important to read the generated responses. Note that for every proposed change, an explanation is provided along with the code snippets added to the project. Click on any code snippet to highlight it in the editor. That's a great way to learn what's happening at the hood.
Now look at that, I have my title and a map with a marker for the adventure's location. Let's continue building our app. Add a filter button on the top toolbar for categories and effort levels with options to select all. Again, I'm very specific about the top toolbar and the filters I want. I'll go to my prompts, copy prompt number 2, Go back to Xcode, paste it in, and hit Return to submit. Now, it's worth noting that a coding assistant may produce slightly different results each time you run the same prompt. For example, you might get slightly different interface layouts and functionalities. However, you always have a solid starting point to build upon. And there we go. I have a button which should present a view with the filters I requested. Let's test it out.
Great, looks good. Now, it's important to note that models occasionally make mistakes. But not to worry, the coding assistant can also fix them for you. Let me show you how. I remove a bracket here, and in a second, an issue pops up. I click on the red X mark and the Generate Fix for Issue. The coding assistant starts working on it. And you can also prompt the model directly by typing, Fix all issues.
Now, onto prompt number three. And this is a big one. Let me walk you through it. Below the title, overlay a floating info card with a category pill and an effort level pill, top left, a bold adventure title, the adventure's description, and a status button, bottom right, indicating if adventure's completed or not. Lots of detail in this one, right?
I'll go to my prompt and copy over prompt number three, come back to Xcode and submit it. Now, a few more tips. You can use prompts like change the select views background color or explain the code between lines 10 and 20. And to better direct the model's responses, you can reference a specific class or file in your project. Type the add character to choose one and the coding assistant will read it along with your prompt.
Back to Xcode. I now have the information card with all the details I want about the adventure. I want to check if the Done button works. It should indicate if the adventure is completed or not. And great, it works. Next up, prompt number four. This is a short one. Add a wide Next Adventure button that floats above the lower part of the maps. I'll copy prompt number four, switch over to Xcode, and send it on its way. This should be a quick one. I'm only asking for a button and haven't explicitly said what it should do. Take a moment to read and understand what's being generated.
I'll read through the response and find the code snippets related to the next adventure button. Here it is. I'll click on it to highlight the code in the source editor. The button's action as a to do comment instead of a real action. And that's why I have prompt number five, to tie it all together and make the app work. Create the necessary models, sample data and connections for adventures with title, description, and so on. I list in detail every piece of information I think is important for an adventure to have.
I'll get prompt number five and I'll submit it to the coding assistant. Now, remember to iterate. If what you get isn't quite right, tweak your prompt and try again. You can also start a new conversation to keep topics organized and go back to your previous conversations using the picker at the top center of coding assistant panel.
Okay, I don't expect to see any changes on the interface other than when I click on the next adventure button, I should get a new one. And voila, like magic, our app is ready. I started with a sketch and an idea. Now I have a working app, one that with a few more tweaks could be ready to publish. Amazing, right? I wish I could see what results you have on your site. I encourage you to continue work on this app. Add features, make it your own, and have fun with it. Now, before I let you go, there's another approach worth exploring.
What if you want to take a more visual approach? Can I start with an image and then go into prompting? Absolutely. I can create a prompt saying, "Create a macro adventure app based on the attached image," and attach this diagram I created previously. And to get this sketch, go to the webpage with the prompts, scroll down to part two, copy the prompt, and save the image to your download folder. So, I'll create a new iOS project, naming it My Adventures, and I'll save it and open the Coding Assistant.
I'll go get the prompt and copy and paste it into the coding assistant. Then I'll click on the paper clip on the bottom left on the chat window and select "Upload files." I'll find and select the sketch microadventures.png, click the Open button to add it into the text prompt, and then hit Return to submit the entire prompt. Models like Cloud or ChatGPT are able to process several of the most common file formats for documents, images, code, and more. This might take a bit longer. The model has to analyze the image structure and generate the code for the full app instead of going step by step like I did before. Speaking of which, those five steps I used before can be pasted all at once. The models are able to handle much longer prompts in the chat, and it helps if they are well-structured and organized.
And here we go, mine is done. And it's almost exactly what I sketched out, very nice. The next steps would be to test the app's functionality and keep prompting until it behaves and looks exactly how you want it to. And that's a wrap for the code along. I started with a concept, came up with a few prompts, and ended up with a fully functional app. The coding assistant makes this possible. you can explore, prototype, learn, and build your ideas all in one place. And there's even more in Xcode to help you build those incredible apps. Predictive code completion, in-app documentation, playgrounds, and beyond. Thanks for joining me. Now it's your turn. Go build something great.
That was awesome. Thanks, Jerome. Amazing to see how fast you can go from an idea to a working app with the coding intelligence features. Thank you, Lucas. Yeah, they allow you to quickly go from an idea to an app. And it really empowers anyone to create apps. It's also a great way to learn more about the process of developing apps. You can read the coding assistant's explanations about how we came up with the code it generated. And you can also ask questions about specific functions or even about issues you have in your code. Yes, it really is a great learning tool. And the more I've used it, the better I became at prompting. Now, during the Code Along, we got a lot of great questions from people joining us online. Right. But first, since the Swiss Student Challenge submissions start tomorrow, I want to ask you, how should students approach using these tools for their submissions? This is a fantastic tool for students. So first, I want to say that every great app started with a good idea and a specific problem it was trying to solve. Now these tools now make it easier to unleash your creativity, go from that original idea to a working prototype. And from there, you can expand, add new features, iterate on a design, test new ideas, and most importantly, learn from the process.
Yes, the more you know about the tool, the more efficient you'll become at using it. Now, for someone who's beginning their app development journey, where should they start? Well, we have a ton of resources on the Apple Developer website. And I would start with the Develop in Swift tutorials.
They introduce app development with Swift and Xcode for anyone, learning how to develop apps for Apple platforms. And they cover everything from app design to distribution. Great advice. Use the developments of tutorials where you learn by doing. It's a resource that everyone should look at, even if you're an experienced developer who's new to Apple platforms. OK, so let's move on to some questions asked during the code along. Sure. The first one is, what models does the coding intelligence features support? Oh, it's a great question. So in Xcode, you can use any models you want. So we have these great integrations with models from OpenAI and Anthropic, but you can also bring your own model. For example, if you're using Gemini with an API key, you can bring that. And you can also run models locally on your Mac. So there are tools that exist out there, such as Ollama, OLM Studios, that allows you to run a model locally on your machine or on your private network, and you can just connect Xcode to those models and use them in the coding assistant, but also in coding tools. Wow, that's very flexible, very amazing. The second question is about conversations and the chat history. Tell me a bit more about what happens when you start a new conversation, how does the context for the model work? Sure. So a great way to organize your IDs is to start a new conversation every time you want to try a new feature or start a new design, something like this. And so each time you start a new conversation, it resets the whole context of the conversation. But you can always go back to all the previous conversation you had and continue and iterate on a previous ID. Now, there's also a great trick, a great tip is actually don't be afraid to break things, right? Model sometimes can produce errors or it can go in a direction you might not be happy with. You can use the history, which I think in the coding assistant is a little clock icon, to always go back to any point in time. So before or after a conversation with a model and before the model makes changes. Got it. So it rolls back the changes kind of like GitHub. It's like a Git history, exactly.
Right. Very cool. And well, we got tons and tons of questions, obviously, about Xcode 26.3. What can you tell me about that? Interesting. This is perfect timing. So on Tuesday, we released the release candidate of Xcode 26.3. And this is a very exciting release because we're adding a genetic coding to Xcode. So we partner with OpenAI and Anthropic to bring their coding agent directly into Xcode. So now, very easily, you can download Codex or CodeAgent directly from Xcode in one click. Really, it's just one click. You know, it's impossible to do simpler. And you can start chatting with a coding agent immediately.
Now, a coding agent is great because it can talk to a model, continue to generate code and fix errors, but it also has access to tools, and that makes the entire difference. And so in Xcode 26.3, we added tools in Xcode for the agent to use. For example, it can build the project. It can add entitlements. It can go through the entire project structure to understand your source code and how your project is organized. But there's more. It can also grab a screenshot from the previews to verify its work visually, right? And we added fantastic tools to search for Apple documentation, code snippets, code samples, and past WWDC sessions. So you can, for example, ask for a new API that was released in iOS 26, right? And if the model doesn't have this knowledge, the agent will actually call the tool to search for the documentation and be able to bring that context into the conversations. Now, there's also something really important in that those tools are using an open standard protocol. It's called the Model Context Protocol, MCP, right? You might have heard of it. And so, you know, those tools use MCP inside Xcode. but you can also connect Xcode to other MCP tools, right? So if you're using other agents, or if you're using other tools that are compatible with the MCP, you can now configure all those tools to work together in Xcode to unleash your creativity and productivity. Wow, what a game changer. Yes. Very good. I really encourage everyone to test it. It's available right now on the developer apple.com for members of the Apple Developer Program as a release candidate, and it's going to be available on the App Store this month. Very cool. Great questions, everyone.
And thank you so much for taking the time to join us. If we didn't get a chance to answer your questions, I encourage you to check out the developer forums at developer.apple.com/forums. Yeah, thank you for joining us. Remember to check out all of our learning resources on developer.apple.com/learn, Be sure to click on the Notify Me button if you're interested in getting updates about the SWIFT Student Challenge, SWIFT Coding Clubs and other student developer activities and resources.
And as I mentioned, the Develop in Swift tutorials are a great way to get you started with app design, development, and distribution. You can even learn more about advanced topics like spatial computing, machine learning, and AI. Finally, remember you can submit your project to the Swiss Food Challenge starting tomorrow, February 6th, until the 28th. I can't wait to see your ideas come to life. Thanks for letting us be part of your journey as a developer. I hope you leave today learning something new and that you'll join us again soon, whether online or at the Developer Center near you. Bye.