We carry more than the software. The whole problem of getting machines to do real work is ours, end to end, so you never build a robotics team.
What they learn on the floor, we feed back to you with insight, and your operation gets sharper for it. They also work in really cool ways that humans can't.
A room and the things in it sit at human scale because humans put them there. A machine is only useful here if it can meet that scale on the world's terms. The hardest part of putting a robot to work was never the robot. It was the room.
The real question is about judgment. Can a machine walk into a space it has never seen and finish the job a person would have done?
FIG 01 · A body holds up in a structured space and falls off as the world gets messier. The distance back up to what the work needs is the gap, and closing it is the layer's job. It is largest in the real world, where the work is.
Most of the money in this field never reaches the field. It pours into the robots, and into the tools for the companies building robots. All of that is real work, and all of it will commoditize the way every layer of hardware and software eventually does. The part almost no one is doing is the hard one: getting a machine to do real work inside a real operation.
That is what we specialize in, and it runs on a layer above the hardware. That layer is what we build. Hand it an objective and it reads the room, then turns a plan into coordinated work across whatever machines are standing in it. The robot underneath can be whatever fits the job, and we can change it the day a better one ships without touching the intelligence on top.
FIG 02 · Your business hands the layer a goal. The layer reads the space and directs coordinated work across whatever machines are standing in it, what they see makes it sharper every shift, and the value comes back to you. Whatever robots are in the room, the intelligence driving them is the same.
The models are converging. The largest are trained on the same public internet, and Larry Ellison put it plainly: “they’re all trained on the same data.” When the intelligence is shared, the only edge left is the data no one else has.
That data does not live on the internet. A machine that works in the real world produces something no lab can fake. Every shift it learns the building it works in, and the exceptions a real space throws at it that no dataset contains. That record is the asset, and no one else can download it.
FIG 03 · This kind of capability only comes from real deployment, never a demo. Every shift in the field feeds the layer, and the curve bends the longer the system runs.
It does not reset when the next machine arrives. The system gets sharper the longer it runs. The moat is that record. It compounds with every deployment and cannot be bought off a shelf. It is the one moat the models cannot share.
The UAE is the best market in the world for deploying robots.
The UAE has the world's highest rate of AI adoption. Microsoft's 2026 AI Diffusion Report puts it first on earth, the only country past seventy percent of its working-age population, and growing at nearly four times the global pace.
FIG 04 · The UAE is the first country past 70% of its working-age population using AI, ahead of every advanced economy. Source: Microsoft AI Diffusion Report, Q1 2026.
That number is about software. AI on a screen that answers and decides. The frontier is already moving off the screen and into the physical world, where most of the work actually happens.
FIG 05 · A society reaches the top of the first curve on conditions that do not move when the technology changes. The second curve is barely off the floor, and the same place is first in line to climb it.
A society does not reach the top of that chart by accident. It gets there on conditions that hold whatever shape the technology takes, and those are exactly what a robot needs to get to work inside a building.
Software AI proved the appetite. Robots that work are the next surface for it, and the UAE is first in line. Three conditions put it there.
One · Demand
The region runs some of the largest and most demanding human environments on earth, and the institutions that own them are done with pilots. Dubai has put a number on it. The government's target is 200,000 robots deployed by 2032, with robotics lifted to nine percent of the emirate's economy. That demand is set by the state itself, and it runs on a timeline measured in quarters.
Source: Dubai Robotics & Automation Program.
Two · Will
The UAE runs like a startup. It reasons from first principles and moves with the decisiveness of a founder. Decisions are made at the top, and fast.
The government also builds alongside the private companies it works with. When the state wants autonomy inside its own infrastructure, it gets in the room and partners to make it happen, clearing a path in months that would take years elsewhere.
Three · Position
Capital here thinks in decades, and the country trades in every direction at once. The work is not boxed in by any single bloc, and it sits beside the people who actually own the buildings we want to walk into.
And we are not arriving as strangers. The person who leads our commercial work has spent more than twenty years building businesses and relationships across the UAE and the Gulf. He knows who makes the call, and he has the relationships to reach them.
Peter Thiel's rule for building a monopoly is to start small. Win one market completely before you reach for the next, because a small market is winnable and dominance there compounds. You expand only into the territory next door.
We run public and private customer acquisition in parallel because here the territory next door sits across the public-private line. The same names own both sides of the table. Win one institution completely and its counterpart on the other side is already watching, and so are its peers in the backyard. Each win becomes the beachhead for the next.
This is where the work is, and where the people who own it want it done.
Headcount is the wrong number. Put a great builder next to a great team and everyone around them levels up, so output grows closer to n2 than to n. It decides how fast everything else gets built.
Our roots are in San Francisco. Frontier engineering, and people who have closed real business. Dubai is the most builder-friendly place on earth, and we build here.
The team is small. We are growing intently, locally.
And we are beginning to work alongside the best AI research labs in the country.
Owning the layer only matters if the work gets done, so we own the rest of it too. We choose the machines for the job and learn the operation they walk into. The result is on us.
We run the fleet and keep it current. When a more capable robot comes out, we put it to work for you, and your service gets better without a new contract or a bigger bill.
What you buy is the outcome. Coverage where you are short, and a picture of your operation you could never see before, built from everything the fleet takes in on the floor.
We do not measure this by demos*. The test that counts is a Tuesday morning in a building that does not care who you are, full of people with their own work to do, where a machine has to actually be useful.
* but we do have some really exciting demos coming.