The improved AI agent access in Xcode has made vibe coding astoundingly simple for beginners, to a level where some apps can be made without manually writing any code at all. Here's what we found.
The concept of vibe coding is one where the user defines to an AI chatbot or agent what to code, and then the agent codes it for them. The intention was previously to be an extension of the commonly found code suggestion and completion systems that you would normally find in a development environment.
Over time, and with the increased sophistication of said AI tools, it's become more involved. A person can describe to an agent a general concept for what they want to produce, and it will come up with functions or even complete apps, some of the time.
These don't usually behave as intended the first time the prompt is made, so there's usually a period afterwards where either the user adjusts the code, or they ask the AI to fix it for them.
As an idea, it's not a bad one, since it can help people create swathes of code quickly, with functions that work well enough to be acceptable to the end user.
It's also something that has developed in a number of ways. Entities such as OpenAI's ChatGPT and Anthropic's Claude have tools in place to help produce code, including chat windows that work within Xcode and other environments.
This is a concept Apple really wants people to adopt in their work.
Vibing with Xcode
On February 3, Apple updated Xcode to version 26.3, initially in a build available to developers but with a wider release expected shortly after.
The update increased support for agentic coding, allowing OpenAI's Codex and Claude to be fully integrated with Xcode. Apple made it so that developers could install their preferred chatbot and quickly iterate on projects using it.
This included one-click installations for Claude and Codex, as well as additional capabilities for other Model Context Protocol-compatible agents.
Apple included some new built-in tools specifically with models in mind. For example, it's possible for an agent to take a screenshot of how an app appears in the in-Xcode device simulator, which can then be used to work out what the agent should change, based on the current request.
Models can also search through documentation, examine file structures, update project settings, and iterate through builds and fixes.
This sounds like the agent can do a lot for you. To a point, it can, and to a fairly amazing degree.
Painless infiltration
I am already well-versed with using ChatGPT to code in Xcode, albeit in a more limited fashion. Through my game development efforts documented on AppleInsider, I have used ChatGPT to edit single pages of code in Xcode at a time.
Aside from getting the Unity export to work on an iPad, I've not created an entire Xcode project from scratch. It's also safe to say that, while I have a bit of experience in C#, the language used for Unity scripting, I've never made a native Swift app at all.
For the purposes of this test, consider me operating as a person who is completely unaware of how coding works where Swift is involved, and is making their first iPhone app.
Installing AI agents in Xcode is simple enough for most users.
After installing the release candidate of Xcode 26.3 via the Apple Developer Program, I was given a heads-up of the new agentic coding changes before being able to create a new project. Blundering through the initial elements, like coming up with a name, I was presented with a blank project to work from.
At this point, I went to the Settings section of Xcode and selected Intelligence. This was where the Coding Intelligence section lived, complete with options to allow agents the use of integrated internet access tools, and more dangerously, to allow agents to use the Bash command line without asking first.
This second option feels best kept off for inexperienced developers. I immediately imagined somehow allowing an agent to run a command that wiped my Mac's storage completely for no good reason.
That said, it would be remiss of Apple not to include some level of guardrails for what the AI can do. While Apple does expose Xcode functionality to the agents, Apple will have ensured that it would be extremely difficult, if not impossible, for things to go horribly wrong.
I selected OpenAI since I have a paid subscription there, and was immediately warned that Codex was third-party software, and it would have access to a ton of stuff if I wasn't careful. I agreed quickly, in part because I had made similar agreements with ChatGPT's previous usage.
Once it was installed, I headed back to the project and clicked the Star icon in the top left corner to bring up the agent column.
This was all extremely simple to set up, almost suspiciously so. Short of directly demanding users install an agent straight after launching Xcode, it's hard to imagine a much easier and relatively pain-free installation process.
Agent mission
For this test, and my first ever Swift project in Xcode, I thought about a very simple application that could take a few hours for a beginner to make. The immediate thought was a Pomodoro timer.
With the agent window open, I provided a relatively simple project brief and expected more questions.
Hi. I want to make a simple Pomodoro timer for my iPhone. It should give options to set a varying block of time according to Pomodoro principles, and then to run. Alerts should play when it's time to stop working and to start back up again.
That was the entirety of the prompt, but that was enough for Xcode to be quickly populated with code. The responses asked if I wanted to add two changes, which I agreed to each time, and I was quickly seeing an iPhone simulator with a working Pomodoro timer.
From an empty project, a brief description became a fully-working app within minutes.
It had a basic interface, sure, but it also had a settings button. Opening that up, it showed options to change the work and break cycle durations, the number of work sessions, and even a toggle to enable and disable alerts.
More astoundingly, the time from submitting the prompt to seeing a working app on my Mac's screen was just under two minutes.
A few button presses later, and the app was shown on my iPhone display, since I had previously registered it as a developer device. It worked perfectly too.
To further press the tech, I asked how I would go about setting it up for TestFlight. Cue a list of instructions on how to get that done, as well as proposals of other changes that could be done to the app, such as onboarding screens for notifications.
Good-ish vibes
This was a simple test of the changes made to Xcode, but one that demonstrates the power of agentic coding.
That it made a functional, albeit relatively simple, app in the space of a few minutes is astounding. If you know what you want to make and can describe it in text form, it will spit out a version of an app that you can use as a base.
Those further changes could also be via prompts if you don't have much coding experience. But for those who know what they're doing, it's a fast way to get to a stage where there's a basic app that you can then build upon.
Beginners to coding, and frankly, those who don't want to know how to actually code at all, have the opportunity to build their app ideas quickly and reasonably well, too.
Developers who have experience in producing apps and probably know how best to employ AI agents in their workflow should also find this to be a positive.
This does significantly lower the barrier to entry for app production, though. That is both a blessing and a curse.
Thinking on the negative side, it's not hard to consider the possibility of people creating lots of simple apps quickly, without due care and attention. All followed by ramming them onto the App Store to make a quick buck or two.
There will be a deluge of simple and very similar apps hitting the App Store Review Guideline reviewers in the coming weeks and months. While this could flood the App Store with quickly made apps, it may be a worthwhile price for allowing the more well-rounded apps to get new and improved features quicker.
You may have to wade through a lot more junk apps in the App Store in the future. But agentic coding could lead to app gems being made brighter and better than ever.
