State Labs

STL aims at removing data silos with privacy-preserving technology, enabling multiple parties to train and run AI together—without ever exposing their raw data

[State Labs @STL_AI]

Meet us in MESSE Stuttgart 10D11

#Vision2026 #AI #VLM

[State Labs @STL_AI]

Came for the chips, stayed for Jensen's new jacket. 🧥

Spent the day at #GTC26 seeing the next gen of humanoids. Atoms are the new bits.

#PhysicalAI #NVIDIA #EmbodiedAI

[State Labs @STL_AI]

☕️ Expect high-density discussions on:
🔸 Scaling AI infrastructure & optimizing compute
🔸 Moving from frontier models to production-grade apps
🔸 Bridging global AI innovation with the Asian market

We’re bringing together the minds shaping the next gen of AI. 🧠

[State Labs @STL_AI]

Are you just following AI — or building with it? 🛠️

We are hosting a curated gathering during #GTC26 for builders focused on what actually works in production.

Frontier AI meets real-world scale.
RSVP 👇
https://t.co/Xh7WX31e6R
@googlecloud @nvidia

[State Labs @STL_AI]

Day1 at #EmbeddedWorld 2026! 👋

You can really feel how fast edge AI is moving👀come meet us!

#ew26 #PhysicalAI #Robotics #EdgeAI #OnDeviceAI

[State Labs @STL_AI]

@after_ephemera Good question. We haven't published yet but i can tell you it is fast (>30Hz).

[State Labs @STL_AI]

SFT can teach robots to mimic. It doesn’t teach them to recover.
We trained a VLA SFT baseline (pi 0.5) on ~270MB (30 demos/scenario).
Imitation looked great, but small shifts dropped success to ~30%.
Full write-up: https://t.co/XzFpNcY52Q

[State Labs @STL_AI]

@physical_int impressive

[State Labs @STL_AI]

full article 👉 https://t.co/GVLD17bXDI

[State Labs @STL_AI]

SFT can make robots look competent, until conditions shift.
In our test, a minor change dropped success to 29.2%.
With RL: 74.5% (and generalization stops collapsing).
Quick summary below. Full notes in reply 👇

[State Labs @STL_AI]

Full write-up (ManiSkill + PPO + rollout/distillation loop): https://t.co/XzUGbts1zA

[State Labs @STL_AI]

SFT makes robots look competent—until the task shifts.

In our benchmark: 29.2% → 74.5% after VLA-RL.
Stress tests (never trained):
• distance ×3 → 53.7%
• appearance shifts → 43.3%
What breaks real-world robotics more: generalization, latency, or cost?

[State Labs @STL_AI]

AI belongs on the factory floor. Updates soon as we move forward.

[State Labs @STL_AI]

We are bringing secure, distributed AI from the cloud to the concrete.
By integrating OpenTMP CIS with Zekeep’s robotic platforms, we’re solving the hardest problems in industrial automation: ✅ Sub-millisecond Latency ✅ Safety-First Engineering ✅ Enterprise-Grade Security

[State Labs @STL_AI]

Also tackling deployment: pi-scale models are heavy for edge; cloud inference adds latency.
We’re pushing generalization + efficient edge running.

[State Labs @STL_AI]

Next: Reinforcement Learning to close the 70% gap, recovery, robustness, and real-world generalization.

[State Labs @STL_AI]

Pushing forward on dexterous manipulation with the Linker Hand from Linkerbot.👋

Seeing learning-based control run on high-DoF hardware in the real world has been really encouraging. Early signals look promising, and there’s a lot of potential ahead.

[State Labs @STL_AI]

ROMI Lab is a true powerhouse of young, brilliant talents in robotics control. It’s refreshing to work with a team that is so grounded, energetic, and focused on making things work in the real world.💪

[State Labs @STL_AI]

Strategic collaboration with ROMI lab (a robotics research lab under The Hong Kong Polytechnic University)！
Both parties agree to jointly research on VLA + Robotics and will disclose more details under this framework as progress is made.
Let’s build the future of embodied AI!

[State Labs @STL_AI]

@physical_int Impressive