Bonsai Robotics has sold more than 400 units to date — roughly 75 OEM retrofit kits and the rest Amiga robots — and says demand for its new hybrid-electric Amiga Max platform has already sold out this year's production run.

The California-based company, founded by CEO Tyler Niday, is betting that a vision-based autonomy stack built on a foundation model can replace the crop-specific engineering that has long driven up ag robotics costs. The system converts ordinary 2D camera images into a scaled 3D understanding of the farm environment — elevation maps, voxel maps of trees, and semantic occupancy — allowing one model to run across almonds, strawberries, vineyards, apples, table grapes and citrus without developers rewriting code for each crop.

"Historically in robotics, you write hundreds of thousands of lines of 'if-then' statements, more or less, to get to a solution for one crop," Niday told AgFunderNews at the inaugural Ruggedize ag robotics conference at Reservoir Farms in Salinas. "If there's a plug on a plow, you have to write code around that."

The data behind the model

The company reports about a million acres of collected data across specialty crops from its 400 deployed units. Bonsai trained an Nvidia world model — using the Cosmos family of world foundation models for physical AI, originally built for on-road self-driving — on that agricultural data to simulate new environments, crops, rain and other conditions. The internal foundation model runs on a fully automated self-labeling pipeline.

The learned approach replaces lidar and stereo vision. Dust that would blind a lidar system does not stop a transformer model trained on historical data, Niday argued, because the camera system infers what lies beyond the obstruction rather than relying on a rule set.

Bonsai's early proving ground was demanding: 30-year-old almond orchards in the Australian outback, where GPS does not reach, dust is constant and sand hills complicate navigation.

The cost math

Niday frames the pitch in operating expense, not just labor replacement. The company's newest machine burns about three gallons of diesel per day instead of 30, he said. The retrofit business and Bonsai-built machines now split revenue roughly half and half.

Flexibility anchors the economics. Bonsai's autonomy platform for almond orchards runs on the Orchard Machinery Corporation AR-500 shuttle truck, a roughly $200,000 machine that historically sat parked for most of the year. Converting it into a multi-task tractor recovers idle horsepower value, and Niday said that flexibility drives buyer interest as much as labor savings.

"You'll struggle with autonomy subscriptions on just a tractor," he said. "You need to go past just a labor replacement and find other ways to drop the cost of these machines and services."

Following its acquisition of Farm-ng, Bonsai now builds its own machines, including larger heavy-duty Amiga platforms for spraying, hauling and lifting. The Amiga Max strawberry sprayer leads the new line, and larger farming applications are the stated target.

Testing access

Reservoir Farms has shortened the path from prototype to grower demo. When Niday started Bonsai, it took him seven months to find an orchard willing to host testing — "People do not want me touching their trees and breaking them and knocking them over," he said. At Reservoir, decision-making growers from the Salinas Valley arrived within days to watch a demo, and the facility opened doors in Napa Valley's grape industry the company lacked connections to.

Niday also noted a shift in feasibility: five years ago he considered robotic harvesting unsolvable, but end-to-end models have changed his assessment — a signal that specialty-crop growers, who face the sector's highest labor requirements, may see harvesting automation move closer to commercial reality.