Claude Design just launched.
In their demo, Anthropic showed an interactive globe built for a website.
I recreated it in
@Rive in š® šŗš¶š»ššš²š š°š® šš²š°š¼š»š±š. Almost for free.
Round 1: Claude Design. Prompted, iterated, tweaked the sliders. Beautiful globe running, fully interactive.
ššÆšŗš¶š»ššš²š ššµšš²š°š¼š»š±š. Also 64% of my weekly plan, gone.
Round 2: Rive Agent alone. Opened Rive, prompted the AI Coding Agent, built a Luau script with state machine and momentum physics. Native-ready for iOS, Android, web.
š±šŗš¶š»ššš²š šÆš³šš²š°š¼š»š±š. Around $10 in AI credits.
Round 3: Both, composed. I asked regular Claude (free) to write the Luau logic. Pasted it into Rive. Errors everywhere, Claude doesn't know Rive's types. So I asked Rive Agent to fix it.
š®šŗš¶š»ššš²š š°š®šš²š°š¼š»š±š. Fix cost around $3.
The fastest workflow wasn't just one "AI-native" tool. It was combining two tools, using each for what they do best.
Stop asking: "Which tool replaces my workflow?" Start asking: "Which combination wins?"
What tool stacks are you testing for production-ready work? Curious to hear, especially from anyone building for production, not just prototypes.
(If you're working on something interactive and want someone to both push back on your ideas and ship them, let's talk)