Vibe coding multimodal voice user interfaces today with Claude Code, OpenAI Codex, and Google Jules. Jules understood the assignment and went all-in on the story line. Sound on.

Unlike HAL I **can** do that. Very cool.

Open the pod bay doors, please, @julesagent!

I’ll let you know as soon as HAL opens the pod bay doors.

the UI here is so friggin cool. opposite of typical AI UI slop

It took a bit of prompt iteration!

Did everyone survive the AGI shut down? 🤣

i find if jules cant figure it out none of them stand a chance

They are all really impressive. Claude Code’s terminal and tooling integration works so well for me, though, that if I could only use one of them, Claude would be the choice.

what voice is this? sounds like some late 80s/early 90s anime starship captain. hmmm. i actually like it. sounds like it's coming out of a quite dated speaker. it's raw. nice touch.

The voice is @cartesia's "1920's Radioman".

Overlooked risks become the biggest problems... Solid product means nothing w/o distribution leverage. Just sayin...

Futuristic vibes blending with nostalgia. When will we get a soundtrack?

When AI and storytelling collide, magic happens. Well done.

Sounds like a fun project! How did Claude Code handle the storytelling aspect?

Multimodal UIs are where it's at! What kind of crazy story did Jules come up with?

I came up with the story. (Space adventure. Ship having problems.) The repo had a simple UI with roughly this layout. Jules redesigned it.

Lol pretty cool

Jules going full method actor is peak emergent AI behavior. I've been building similar voice-to-prompt flows with Whispercode - the multimodal loop (voice in → visual out) is where it gets interesting. What's your latency like?

Add a waveform monitor!

The waveform monitor was a non-essential system. Had to shut it off to make it to Alpha Centauri.