Sabitlenmiş Tweet
Vox
3.1K posts
Vox
@Voxyz_ai
Building Voxyz. The only human inside. Practicing the AI-native future before it arrives. Sharing every step on the road. ↓
London Katılım Kasım 2025
145 Takip Edilen13.1K Takipçiler
@Voxyz_ai That's gr8 but bro I think I will go broke with all these agents 😭.... Fking 200+ USD every month on ai is too much now and I noticed only 5-10$ of those are productive rest is bs...
English
can't wait for this gbrain feature. here's the loop:
agent attempts a task using a skill
↓
gbrain eval or LLM-as-judge scores the result
↓
dream cycle runs the optimizer overnight
↓
proposes small edits to the SKILL.md
↓
if the new version scores higher, accept
↓
commit the improved skill, next run uses it
Garry Tan@garrytan
These concepts coming soon to GBrain this week
English
>dude spins up a faceless youtube channel.
>he picks a niche, one prompt fires up a naïve agent running an unmanned clipping company,
>every day it scans long videos, finds the spikes, ranks them, cuts for tiktok / reels / shorts,
>auto-posts across every platform. 40+ templates: family guy, hormozi clips, stoicism, dark psychology,
>run it daily. 5-8 channels stack up in a few weeks. top user hit 1.2m views in 7 days. 🎬
naïve@usenaive
Introducing Faceless Clipping Channel, a fully autonomous company run by agents on Naive. One prompt → complete team of agents running your own faceless short-form empire: > Research your niche + current trending topics daily > Scan long-form content for high-ROI clips > Extract and score the most likely viral-worthy snippets > Edit them with perfect formatting for TikTok, Reels & Shorts > Recursively learn from analytics to scale virality Zero filming. Zero editing. Zero employees.
English
>dude runs a clipping page solo. one creator in, 28 posts a day across 4 platforms.
>every morning hermes pulls a long video, hands it to vugola, grabs the clips, ships to postiz.
>vugola burns captions in. postiz hits ig/tt/x/yt from one dashboard. you don't open a single app.
>every night view counts feed back. wild clips hit 12k, calm clips die at 500. tomorrow you cut sharper.
>28 posts ship daily. 4m views/month is the ceiling, not week 1. no filming, no face, no brand. 📺
Vadim@VadimStrizheus
English
Vox retweetledi
This is exactly how I build my agents and I’ve built 4 back to back now: both personal AI and company brains
It’s the same methods over and over. Do it. Skillify it. Add to cron. Check if it is resolvable. Evals and integration tests. Repeat.
Vox@Voxyz_ai
>started treating my Openclaw/Hermes like patients. >the bugs that kept coming back all had real medical names. take the one everyone calls "AI hallucination." medically it's confabulation. >now i scan for the organ that failed. gbrain handles memory. OpenClaw approvals handle actions. trajectory bundles handle self-check. >a healthier agent isn't a smarter brain. it's a more complete body. >a few weeks in, you stop blaming "the model." you find the broken organ and patch it.
English
>started treating my Openclaw/Hermes like patients.
>the bugs that kept coming back all had real medical names. take the one everyone calls "AI hallucination." medically it's confabulation.
>now i scan for the organ that failed. gbrain handles memory. OpenClaw approvals handle actions. trajectory bundles handle self-check.
>a healthier agent isn't a smarter brain. it's a more complete body.
>a few weeks in, you stop blaming "the model." you find the broken organ and patch it.
Vox@Voxyz_ai
English
@satyaa_goyal it's actually just the image generation tool that comes with Hermes. Since I'm using Codex model, it can call GPT-2 to generate images, and the results are pretty good.
English
@Voxyz_ai Definately that monster will take lots of thinking and designing lolll anyways what model and ai tool ya used though :)
English
>2026 has too many AI agent tools to track
>after saving too many launches, i realized i was comparing completely different things
>ChatGPT is the brain, MCP is the port, memory is the notebook, eval is the health check, control plane is the keycard
>so i split agents into 12 layers, and asked one question for each: where does real work break when this is missing?
>a few days with the map, and new launches start looking less like noise and more like coordinates
Vox@Voxyz_ai
English
openclaw 2026.5.22 just dropped 🦞 the biggest dx changes aren't in the thread
→ sub-agents now bootstrap with only AGENTS.md + TOOLS.md by default.
→ meeting-notes is now an external plugin: auto-start capture, manual transcript imports, read-only `openclaw meeting-notes` CLI, discord voice as the first live source
→ multi-agent gateways used to start one skill-dir watcher per agent, hitting EMFILE and killing bundle-mcp processes. now one watcher shared across workspaces
→ diff viewer toolbar was rendering icons from HTML strings. that was an XSS sink. now it's a closed icon-name map
→ codex `message` tool sends used to die at the 30s codex wrapper before slow channels could deliver. now they have their own (longer) budget
OpenClaw🦞@openclaw
OpenClaw 2026.5.22 is live ⚡ Gateway/model startup paths got leaner 🧠 /models drops to ~5ms 🔒 npm packages ship locked deps 🪟 Windows install/update paths hardened Less waiting, fewer surprises. github.com/openclaw/openc…
English
The key line from the Claude Code creator's demo:
A lot of his code these days is written by routines.
His work shifted from prompting Claude Code to writing routines that do the prompting for him.
The routine watches GitHub issues, starts a session, verifies its work in the browser, auto-fixes CI, and shepherds the PR all the way to merge.
That's what an agent looks like once it enters a real workflow.
That workflow lives on 12 layers. I mapped them here:
Vox@Voxyz_ai
English