正在加载视频...
视频加载失败
introducing Ralv, the 3D interface to manage AI agents ◆ run multiple coding agents in parallel ◆ infinite 3D canvas instead of chat window ◆ use with claude code & codex subscription we also just launched on producthunt 👇
15,951 次观看 • 5 个月前 •via X (Twitter)
22 条评论

upvote on producthunt, means a lot 👉👈

congratulations, this is epic!

appreciate you man

first impressions are so slick Need to get these guys to work Congrats

cheers!

Looks cool but won't it get busy if you have a lot of threads? Working on a similar native macOS app and got very far already. Matches Codex App's UX as much as possible. Uses native Ghostty terminals. Supports T3Code. Has Chromium embedded with devtools. Includes a light VS Code based editor/git manager. Lots of agent features. Please let me know if you want to work together 🙏🏻

I'm not sure the 3D canvas solves the actual pain point though. Managing agents feels like a coordination problem, not a visualization one.

in what sense do you mean coordination problem, e.g. what is the bottleneck in parallel orchestration for you?

Does this use upstash boxes for sandboxes? Or is this all local? (Sorry if you said in video I didn't watch it)

right now it’s local and i’m experimenting with upstash box remote agents today 🦖

Is it open source? Also if you end up using upstash boxes you should write a blog post about them. I'm in the upstash oss program and super close to using upstash boxes for a open source background agents project.

this is awesome

🫡

looks so cool congrats on launch

vr agent management when?

imagine 👀👀

this is cool!

Running agents in parallel is smart. Verifying what they produce is the part nobody has solved yet. We found 24 bugs in documenso with 0 false positives using a debate layer across 3 models.

Congratulations on your launch @joshtriedcoding Ralv looks solid. did you run a "devil's advocate" pass on this before you started? i'm curious what was the one 'green flag' that made you pick this over your other ideas?

STARCRAFT

is the 3D layout mapping anything specific (file tree, agent dependencies) or more just giving each agent its own workspace?

This would be super helpful. Kindly check your dm


