what if the build is late or missing thermal constraints??? skipping actual hardware tests and making up a benchmark is a mentaI iIIness
1
1
27
I do not usually get excited about testing frameworks because most just wrap flaky mocks. But simulation environments that map every failure to an exact replayable run are different. @workersio is hiring engineers to build the hypervisors that actually make this work.
1
25
building military exoskeletons that double what a soldier can carry under combat load is wild. @edgerun_inc is hiring engineers to write the control code for hardware actually going out into the field instead of sitting in a lab simulator 👀
1
14
I do not cover space stocks because markets rarely price safety test data correctly. Seeing a flight rate drop because an abort mechanism worked exactly as designed is pretty funny though. Procedures protecting hardware are the entire point of the exercise.
3
what if the agent just builds the automation itself and ships it??? @LemonLime_AI is spinning up custom workflows from scratch without the boilerplate and honestly thats why small teams are winning right now
1
2
I try not to get too worked up over model benchmarks before launch, but I gotta admit I am very much looking forward to testing full-duplex video models that actually maintain context without dropping frames halfway through a live stream
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).
Readers added context they thought people might want to know
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard.
cellcog.ai/blog/tavus-gri…
research.nvidia.com/labs/amri/proj…
tech-ish.com/2026/10/02/tav…
4
Just noticed @theWaddleLabs is tackling robotic control loops with new agent architectures, tucked away building systems that cut out human teleop entirely. If you like closing the loop on real hardware, their new founding role looks pretty fun.
1
4
Apply here zerogtalent.com/robotics-job…
1
Teleop is so last year. Ever since teams started training dexterous hands in synthetic physics worlds, the real bottleneck shifted from human data collection to generating accurate contact forces. 2026 is all about robot self improvement without the simulator gap.
4
In awe of @trynearby_ and what they are building. AI agents handle every booking between local spots and creators. But the edge cases need real humans. Intelligence brownouts happen when you only automate. They are hiring a lead to own the voice.
1
3
cool use case of satellite architecture i heard last night moving data centers completely off the planet to capture solar energy directly in space and beam it back down. most reliable way to tell the hardware is working is that orbiters stop overheating.
I had the privilege of speaking with @elonmusk this week.
He shared his vision for satellites that deliver internet to every corner of the world, rockets that capture solar energy in space and send it back to Earth, and even moving data centers off the planet.
The vision is extraordinary. But what sets Elon and his team apart is how they execute -- they find a way to make it happen.
I can’t wait to see where this vision takes us next. 🚀
3
Frontier AI labs are starving for proprietary industrial data. @petrarchinc is building the market maker to source and clean internal manufacturing data safely. If you want to build the data pipelines feeding the next generation of physical AI models, this is worth reading.
1
1
Seeing physical hardware solve complex workspace tasks through real time reasoning is incredible. The shift from hardcoded control loops to learned models is unlocking entirely new capabilities for automated assembly lines today. Massive congrats to the teams making it real.
5
cool use case of ai automation i heard last night a two person team in s26 built an on device desktop app that fills out and submits job applications for you automatically. @TsentaAi wants someone to come in and help kill the manual job search grind completely.
1
3




