life these days:
still not where i wanted to be.
still figuring things out slowly.
but less panic now. less noise in my head.
maybe thatβs how it was supposed to be.
Debugging used to be a whole mood.
Now it's just β AO start β AO spawn 1954ββοΈ
No prompts. No commands. AO just hands it off to Claude Code, Codex, Grok β and comes back with the fix. π«‘
@aoagents
prompt β HTML scene (three.js + GSAP) β headless Chrome (Puppeteer) β JPEG frames β FFmpeg β .mp4
each frame is rendered by seeking the timeline to i/fps, not wall-clock. same prompt β pixel-identical mp4, every run.
https://t.co/osQTW9Dwnp
what if `prompt β mp4` actually worked, deterministically?
built Renzaflow: describe a shot in plain English, get an exact, repeatable video.
three.js + puppeteer + ffmpeg.
https://t.co/uD08rTo3U6
what if `prompt β mp4` actually worked, deterministically?
built Renzaflow: describe a shot in plain English, get an exact, repeatable video.
three.js + puppeteer + ffmpeg.
https://t.co/uD08rTo3U6
@heynavtoor https://t.co/oU1PllNiQX
I've been building mine (GWEN) for a while now. missing these layers. Blueprint's solid, gonna wire them in this week.
Built my own JARVIS.
Voice-first. Always-on. Claude thinks, ElevenLabs talks, macOS obeys.
Introducing GWEN π€ :https://t.co/JteFdba3Sj
Video by Renzaflow β’ Audio by @HeyGen#AI#BuildInPublic#ClaudeCode#MacOS#OpenSource
My coffee took longer than my PR β
ao spawn β walk away β done
agent orchestrator is live .
check it out here :https://t.co/Zl8VgML5Pe
@aoagents@agent_wrapper
I built Gwen β a personal AI that actually runs my life.ποΈ Voice-first
π» Controls my whole Mac
π§ Fixes her own code & relaunches herself
"Hey Gwen" and she just... does it. πΈοΈ
https://t.co/oU1PllNiQX
Multi-agent coding workflows in @aoagents.
No terminal babysitting, no guessing which agent changed what, no manually tracking every branch.
Run different agents across different projects and keep the workflow moving from one board.
Use Claude to plan.
Use Codex to implement. (endless possibilities)!!
Let each task run in its own workspace, branch, and PR.
Built for:
β Multi-agent sessions
β GitHub issue queues
β Isolated workspaces
β Branch and PR tracking
β CI feedback loops
β Review-driven fixes
Someone in our Discord decided their coding agents needed emotional support.
They asked AO to build a tiny screen pet that watches PR activity, reacts to urgent stuff, sleeps when inactive, and remembers where it was dragged.
Now their agent dashboard has a little companion living inside it.
10/10, give your AI agents a Tamagotchi.
Whatβs the funniest or most useful thing youβve built using AO?
Server link in the comment. Feel free to hop in.
My coffee took longer than my PR β
ao spawn β walk away β done
agent orchestrator is live .
check it out here :https://t.co/Zl8VgML5Pe
@aoagents@agent_wrapper
Tired of using everyone else's AI assistant. So I built mine.
Meet MJ β voice-controlled,runs on Mac reads your calendar, sends messages & controls your Mac.
Powered by @AnthropicAI Claude + FishAudio
#BuildInPublic
https://t.co/oU1PllNiQX
@jamesroberts@aoagents@NousResearch Each one runs in its own worktree so they parallelize without stepping on each other. CI fails? The agent reads the logs and pushes a fix. Reviewer leaves a comment? Agent addresses it.
You supervise. They ship.
https://t.co/Zl8VgML5Pe