Embodied AI · Construction jobsites

Your materials, delivered the last 100 feet.

A tracked robot that crosses mud, gravel, and gates to drop building materials exactly where your crew needs them. The work a warehouse robot can't touch.

Talk to us See how it works Product concept · sim demo in progress
60-second film · real jobsite footage + concept visualization · sim demo in progress
60–80kg
Payload per trip
45°
Grade / rough terrain
IP65
Mud, dust, rain rated
0 drivers
Worker taps, robot runs
01 · The jobsite

Warehouse robots need a flat floor. Your jobsite is the opposite.

A residential site is mud, gravel, wet fill, scrap, and parked machines. That ground is where warehouse robots fail, and where the work still gets done by hand.

01

The material lands at the curb

The truck dumps pavers, lumber, and bags at the driveway. Getting them to the work is its own job.

02

The last 100 feet has no road

Through a 90 cm gate, down a slope, across mud. No vehicle fits, so a person carries it.

03

It eats the crew's best hours

Up to 42% of a trade's day goes to hauling instead of the work you pay for.

The power buggy keeps the bulk hauling. We take the last 100 feet it can't reach.
02 · How it works

The worker decides. The robot does the labor.

The person makes the call. The robot does the carrying.

01

Perceive

A camera scans the site and builds the map on the worker's phone.

02

Order

Tap what to bring and where to drop it. That's the whole interface.

03

Pick

Vision finds the right pile. The robot lifts a full stack.

04

Navigate

Tracks cross the rough ground and steer around whatever's out there.

05

Deliver

Sets the load on the mark. Logs the route for next time.

03 · Why now

The cheap, flexible laborer who ran materials is disappearing.

This was hand work because labor was cheap. It isn't anymore.

92%

of construction firms can't find qualified craft workers.

AGC 2026 workforce survey
1.4M

unfilled skilled-trade jobs projected by 2030 in the US alone.

ABC economic impact report
9–11%

wage growth for specialized trades in high-demand markets.

AGC / ABC 2026
04 · The machine

Purpose-built. Not a humanoid.

A tracked base carries the load and a simple lift sets it down. It cannot tip over, and it costs a fraction of an arm.

  • Tracked crawler base. 300 to 500 kg rating, IP65, climbs 45 degrees. Mud is home ground.
  • Carry a stack, not one block. 60 to 80 kg per trip keeps pace with a working crew.
  • Lift, not a costly arm. Moving a stack takes lift-and-place, not a $40k precision arm.
Jetson
Orin on-board compute
Depth + RGB
Pile ID + avoidance
Learns
Smarter every run
~$20k/yr
To run vs a $60–80k laborer
05 · The brain

No two sites are alike. So it has to think.

A hard-coded machine dies out here. This one figures the site out on arrival.

01

It sees the site

Vision picks out the pile it was sent for, plus the cords, scrap, and open gates no one warned it about.

02

It plans on the survey

The lot survey is its map. It routes pile to drop, threads the gate, reroutes on the fly.

03

It learns every run

Each delivery logs terrain and obstacles. The whole fleet gets the update.

The brain trains on years of first-person jobsite footage we shot ourselves. That data doesn't exist anywhere else.
"I spent over ten years moving these materials by hand, across mud and through gates. I'm not a robotics person who found a market. I'm the guy who did the job and got sick of it. I'm building the machine I wished I had on every one of those days."
Founder · 10+ years North American contractor
06 · Contact

The last 100 feet, automated.

Sim demo in progress. Want a pilot on your site, or want in early? Talk to us.

yockzhang@gotranno.com