I Tried the New Claude AI App Builder and I Didn’t Have to Write Any Code
Last year, Anthropic introduced a new feature called Artifacts in its Claude AI chatbot, which lets users see the results of their conversations in real time — like charts, reports, websites, or code snippets. Now, the feature has received a major upgrade , letting you create and share AI-powered apps just by describing them.
Apps you create using this ” vibe coding” approach can have Claude AI intelligence built in, and they’re available to all users, so you can try them out whether you have a paid subscription or not. And if you share your creation with other people, when they launch the app, it will count against their Claude usage quotas, not yours.
One example Anthropic gives is a card app. Previously, you could ask Claude to write you an app that showed you cards, but you would have to fill in all the details and content of the card yourself. Now, you can use Claude’s AI to create card content that is specific to individual users, as they will be able to give the chatbot hints about the text they want to see.
The updated feature should now be available to everyone, though you may need to enable it: On the web, click your account name (bottom left), then go to Settings > Profile and turn on Create AI-powered artifacts . You’ll see some example apps other people have created, and you can test any of them in your browser and make changes to them using the Customize button.
Back on the main Artifacts screen, you can see any projects you’ve made so far and tap New Artifact to start creating something new. You’ll be given a few categories to choose from, including Games , Productivity Tools , and Creative Projects — but if nothing fits, tap Start from scratch .
Create your own application
I haven’t done any coding since I learned BASIC in high school: I used it to code a soccer simulator that randomly picked scores and scorers, which was a lot of fun at the time. With Claude’s new and improved Artifacts tool, I was curious to see how good an app I could make from vibrations alone.
My first thought was to ask Claude what I should build, a troubling reminder that AI is robbing us of our ability to think and make decisions for ourselves. In the end, I used my own brain cells to make a choice: I wanted to build an app that would generate quiz questions on any given topic, with a sliding scale of difficulty and answers available when needed.
I explained what I wanted using natural language and zero code: an app that would ask the user for a topic, then display a question about that topic with the ability to click to reveal the answer. I also wanted options to adjust the difficulty of the question and switch topics. In fact, it only took one tooltip to describe everything.
Claude gave me an overview of what he would be doing before starting development, saying he liked the vision of my app (I’m sure he tells everyone that). When I gave the go-ahead, the browser window split to show the app and its raw code on the right (you can switch between these views with a click).
The resulting app was pretty good the first time around, certainly in terms of its functionality — everything worked as I wanted it to, and it was well designed. However, the AI’s question selection and difficulty rating was a bit odd: all the questions had correct answers, but the app tended to fixate on a limited set of subjects within each topic (such as certain movies, actors, or bands).
It was probably unfair of me to ask a Twin Peaks question on the kid’s difficulty level, but “where does Twin Peaks take place?” is probably too easy, and Claude generated the same question on the next difficulty level. The hardest question was about the model of tape recorder that FBI agent Dale Cooper uses, which stumped me.
But that’s more about the limitations of AI models and understanding them. As for actually building the app, I was most impressed with Claude: he was very good at answering my questions and making edits, allowing me to change colors and layouts with simple text prompts. It’s cool to see the app rebuild itself in real time, and then have the AI bot explain what changes were made and why.
There was one issue with the new “suggested topics” field that took Claude a long time to fix, despite my attempts to debug it with vibration. He eventually got the field in the right place, but it took several iterations to get it working properly. Sometimes the chatbot interface is too opaque, although the raw code is always available for serious programmers.
At the moment, it seems best suited for casual, small projects — as the examples Anthropic has on display show — but I can see it being useful for people who want to build custom tools for admin or productivity. I definitely enjoyed the experience, and felt like I learned a fair bit about coding along the way, even if the end result was a bit buggy (more due to the AI content than the code itself).