Pinned Tweet
The first $1T Physical AI company won’t build a robot.
Introducing @safesightinc, the AI infrastructure to safely operate existing machines.
When we filmed this last month, we had $20m of LOIs and POs for our first product, Saferide. Today we’re at $40m of LOIs and POs from retailers and operators in defense, auto, and construction.
DM if you want to learn more or join our team!
Every day you're either a driver, a passenger or a pedestrian. Whichever you are today, you deserve to be safe.
Machines should get better every time AI does.
TOPS/$ at the edge has 150x'd over the last 10 years. Machines are replaced every 15-20 years.
If physical intelligence is gated by on-board compute, we'll be waiting the rest of our lives for ubiquitous automation.
We need a way to decouple rapidly evolving models and compute from machine replacement cycles. That's Safesight.
We’re turning machines into robots.
Think about how tired you feel after a three hour car ride where you ‘just sat there’.
It’s because you didn’t just sit there, you made thousands of tiny decisions and corrections every 1.5 seconds the entire ride so you wouldn’t die.
With @safesightinc’s first product, Saferide, the machine takes some of the weight off your shoulders so your trip is safer and more relaxed.
You’re still driving, still paying attention, and still in control, we’re just making your machine more intelligent so you’re not the only one paying attention.
Will Bright retweeted
Really excellent post on the economics of robots by Anthropic. Cost is the critical input for adoption. (We also wrote about this in depth here: robo.inc/blog/robotics-model…)
anthropic.com/research/what-…
Will Bright retweeted
I left Silicon Valley 6 years ago
Why?
Because it's a monoculture
The most extreme group think of any place on the planet
Show up to a dinner party and everyone is in tech
So what do they do? They start stack-ranking each other
And it's a totally ordered set: being founder of co $100B market cap > founder of co with $50B market cap > managing partner of tier 1 venture firm > hired CEO of $50B public co > founder of $10B co > etc etc
Pure toxicity that nerfs creativty
And then as a corollary, people are especially afraid to speak truth to power as everyone knows their place in
"The Hierarchy"
Reject this nonsense. Stay weird. Think independently. Be honest with yourself. Have fun.
AI can do anything that happens digitally. We're making it so AI can do anything that happens physically.
It's Christmas in August!
Today, we're kicking off the first phase of the research preview for Model Hardware Standard (MHS): a new standard for AI agents to safely operate physical equipment in scientific research and advanced manufacturing.
Read more: anthropic.com/news/model-har…
We need a "dumb" control layer between models and machines.
That's why Safesight cannot be modified by the compute it's attached to.
Separating the control layer into dedicated hardware is the only viable path for models to safely operate in the physical world.
lnkd.in/eEpGKZ8m
Will Bright retweeted
“Software has good margins.”
Ok, but why?
You’ve probably been told that it’s because there’s zero replication cost. But that isn’t true. It simply looks that way because the replication cost and the consumer price are very far apart due to the fact that you have a monopoly.
With enough competitors, Salesforce must be free, and you still need to pay for servers and SDRs. It’s the lack of competitors that allows Salesforce to create margin.
Software had good margins because it was easy to make a monopoly.
Why was it easy to monopoly in software? Because innovation always creates monopoly, at some magnitude. When you innovate, you are for a moment the only company in the world that has your capability. The world was very big, and the group people who knew how to eat it was very small.
For 30 years, software was a wide open field for innovation and therefore monopoly creation. A vista of infinite margins. In many ways it still is.
What if I told you that the same is true in the physical world?
I don’t just mean that software can be applied to the physical world (though it certainly can be, and that will yield gargantuan companies). I mean that atoms themselves are currently arranged poorly. Infinitely poorly, allowing for infinite monopolies, the same way that software was 30 years ago.
To make matters worse (or more opportune), smart and high agency people have generally been ignoring the world of atoms since the 70s. That means there’s an enormous amount of lag in the system. Many crowns lying in many gutters.
But it won’t stop there. I think that the software / hardware divorce was a singular moment in time. I’m not so sure it was a good one. For one thing, it gave everyone brainworms. We now think that software has infinite margins innately, or innately has the ability to scale.
But the fact is that the universe will be arranged in a million ways by thousands of brilliant people, and it is those *ideas* that have infinite scale and infinite margin. SpaceX took launch from $50,000/kg to $500. That was just the beginning. Beyond Valar, I can’t really tell you what that will look like. That’s the whole point. But there are things which cost $10,000 today which will cost $100 in ten years.
As energy, intelligence, and dexterity become cheaper, and as creative and genetic minds turn their attention back to atoms, you will see the next brood of leviathans grow.
This is the right path for consumer robotics: non-anthropomorphic robots with anthropomorphic behavior. Appliances that behave like humans are much better than robotic humans who don't.
machinelearning.apple.com/re…
arxiv.org/abs/2501.12493
The death of exploratory navigation and the rise of directed navigation will be studied as one of the strongest indicators of thought quality collapse.









