Robots still haven’t had their “ChatGPT moment.” Why? 🧵
It’s not the hardware. It’s the DATA.
LLMs train on ~15 trillion tokens. The biggest robotics datasets are just a few thousand hours of video.
@PrismaXai is building the fix 👇
Since my last post, Voice Keyboard has gained:
• Auto-stop after a pause, with on-device speech detection
• Expanded vocabulary and local word replacements
• OpenRouter speech-to-text preset
1,240+ stable APK downloads on GitHub. Thanks for trying it!
https://t.co/QqeUVTxswk
You can train robots with just your phone. 📱
@PrismaXai is collecting egocentric data: first-person clips shot on a smartphone, GoPro or VR headset.
No robot needed. Clips that meet their Operator Intent standard get paid.
Opening in waves. Robots learn from human hands 👀
PrismaX in 30 seconds 🦾
Problem: LLMs train on ~15T tokens. Robots get a few thousand hours of video.
Fix: @PrismaXai lets people control real robots and validate the data.
Edge: $11M led by a16z CSX + plans to share data revenue with creators.
Physical AI's data layer 👀
What concrete on-chain or economic mechanisms (beyond basic wallet age/activity filtering) can make large-scale Sybil reputation farming for ERC-8004 AI agents on Monad unprofitable long-term, given that leaving reviews is still cheap and doesn’t require proof of a real interaction?
@PrismaXai 5/ Why it matters: robots learn from these clips.
One “almost right” episode labeled as perfect = a robot that learns the wrong move.
Validators are the quality filter for physical AI. More notes soon 👀
#PrismaX#ValidatorWatchWednesday
Validator Notes #1 🦾
I gave a robot 97/100.
The consensus said: FAIL. ❌
Here’s what my first @PrismaXai validation taught me about how robots are really judged 🧵
@PrismaXai 4/ My lessons for next time:
Read the instruction word by word
Check every sub-step, not just the ending
Watch for re-grabs, hesitation, spilled drops
Rewatch before scoring
90+ only for truly flawless runs
@PrismaXai 5/ Backed by an $11M round led by a16z CSX, PrismaX is betting the next AI breakthrough happens in the physical world.
I’m starting as a validator and will share my notes here. Follow along 🦾
#PrismaX#PhysicalAI
@PrismaXai 1/ What is PrismaX?
A base layer for real-world robotics, built on 3 pillars:
🦾 Teleoperation: humans remotely control robots
✅ Data validation: the community checks data quality
🧠 Models: better data → smarter robots
@PrismaXai 4/ The flywheel 🔁
More data → better models
Better models → more efficient teleop
More teleop → even more real-world data
And a fair-use idea: revenue from data should flow back to the people who created it.
Robots still haven’t had their “ChatGPT moment.” Why? 🧵
It’s not the hardware. It’s the DATA.
LLMs train on ~15 trillion tokens. The biggest robotics datasets are just a few thousand hours of video.
@PrismaXai is building the fix 👇
They chased him through all of SOLANGELES. 🐂💨
"what happened?!"
"bull run is coming."
When the bull himself says it — you listen.
$RED — launching soon on https://t.co/T9WC9uGwVo 🎩🟢
#Solana#PumpFun#Memecoin#AI#Agents