Web Artifacts allow you to seamlessly create and share interactive, dynamic components (HTML/CSS/JavaScript) directly in your posts. From visualizations and quizzes to lightweight web apps.
The above artifact was generated using GPT-4o and the anime.js library CSS Grid demo by Claude 3.5 SonnetLearn more at: Discourse AI - Web Artifacts
To use: hold a question in your mind, and press the magic item icon. 
Okay, I’m gonna take a break for now… so many ideas!
Any of those examples shows nothing on iPad 
And if open fullscreen on the Hub I can’t get out of it without leaving a forum.
We have a new build of hub that @pmusaraj is publishing, have you updated to latest ?
Yes! Esc to leave fullscreen is difficult on tablets, and using the back button does not take me back to the topic but one step further backwards.
I think we can push state when you click the button to fix full screen
Maybe that overlay makes no sense on touch , it does get in the way of controls at the top anyway
This is a great addition!
Just want to add a few points for those who might be hesitant about using this. For the last few weeks, I have been primarily using ChatGPT with the 4o model to create plain/vanilla HTML pages using JavaScript and CSS. What is amazing is that the generated code almost never has a bug, but prompt engineering is needed to get the desired results. Sometimes, what you think and what the model thinks you want may not be the same, so a little tweaking of the prompt is needed.
If you have access to a canvas in coding mode or similar, these are great for getting your feet wet. Also, if you are not a programmer and don’t do this normally, one of the go-to tools for fixing bugs is to be aware of the F12 (ref) key with Internet browsers.
I could write a book on this, but I’ll stop here; you’re in a sandbox, so play around!
If system has updated it, then yes. 2.0.7 two weeks ago.
Yeah, it was Hub issue indeed. Safari shows nicely.
Is an example of prompting possible for us non-coders?
Edit
First try worked. That kind of dev-work is great. Just telling what is needed. But some questions:
Simplified prompt for use as a AI Web Artifact (do not have access to AI Web Artifacts to test but should work or be very close)
Create a slot machine game, use JavaScript and CSS.
Actual prompt used with ChatGPT 4o
Create a simple HTML page for a slot machine game, use JavaScript and CSS. Create as a single HTML page for download.
Here is a more advanced prompt that was tweaked a few times before achieving a desired result. Will have to be modified for use with an AI Web Artifact.
Create a simple HTML page for a Blackjack game, use JavaScript and CSS.
- Create as single HTML page for download.
- Show the cards of the dealer after standing.
- Add option for additional bot players, show the bot players cards.
- Add option to bet and keep track of amount.
- Show total of player cards.
- Show dealer cards before displaying popup of who won.
If you want something a bit harder, create a game of checkers with a bot opponent.
That release did not include the artifact fix. New version, 2.0.8, submitted for review just now, it should be available shortly.
Tbh I’m struggling a bit how to imagine inserts that indeed enhance a post in a seamless way. Are there already examples of some practical application that would be included in a seamless way?
On my forum is lot of topics about nutrition of dogs. Now I could offer a simple planner, as an example, what would user’s dog need per day.
I would like to use an AI, but those fail every time, and are too expensive. Sure, a plugin or perhaps a component could do same and better, but I don’t know how.
If I could send the answer of an artefact to AI persona with pre-made propmpt would be nice, but I don’t know if that is possible at all.
Let’s try it! 
I got it. Kind of. Basically it is ”just” another conversation with AI where one explains what is needed, how it should look like, and other strict guidance and rules.
It is basically Visual Basic, but easier 
Visual Basic for Dummies. 
Those images are from Wikimedia; I’m looking forward to using utilizing public data interactively, for art and information sharing. 
Oh, wow, I missed this announcement. It’s amazing! 
Congrats team.
I can see already it will lead to many positive consequences, like increasing user engagement, making it easier to explain/learn things, and giving interactive feedback. Also, it’s fun 
.
