Sabanto sells a box of cameras, sensors, and computers that bolts onto almost any tractor already sitting in a farmer's shed and turns it into a machine that drives, tills, seeds, and mows on its own. The company was started in 2018 by Craig Rupp, a former Motorola and Monsanto employee who grew up on a farm in Iowa, after he built and tested his own automated tractor by hauling it around the country himself.
The real story here is not the gadget. It is the labor math behind it. The number of temporary farm work visas the government approves each year has grown from about 48,000 two decades ago to close to 400,000 now, and almost none of those jobs get a single American applicant.
That is not a small shortage, it is a structural one. Farms keep raising wages and expanding the visa program, and the workers still are not showing up. On top of that, immigration enforcement is tightening at the same time that a large share of crop workers hired outside the official visa system turn out to lack valid work papers, which means the workforce farms already depend on is getting riskier to rely on, not just scarcer.
That combination, expensive to hire, hard to find, and increasingly risky to employ informally, is exactly the gap Sabanto is selling into. A new tractor built for autonomy from the factory can run close to a million dollars. Sabanto's kit costs about $75,000 up front plus $10,000 a year, and it works on equipment the farmer already owns, which is why one case study found it saved close to $13 an acre on seeding alone.
Sabanto is not alone. John Deere wants a fully self-driving system for corn and soybean farming by 2030, but is deliberately keeping only a small number of units running while it tests the technology, and separate companies like Monarch Tractor, PTx Trimble, and Bluewhite are all chasing the same labor gap with their own machines and retrofit kits. Sabanto just closed a new funding round led by Bayer's venture arm, with Sabanto's founder framing the pitch plainly: the goal is to let farmers add productive hours by running smaller, cheaper machines longer, not by buying bigger and more expensive ones. One customer is already running 25 of these units on a single operation at once.
For anyone running a business built around old, expensive machinery and a shrinking pool of workers willing to operate it, this is worth watching closely. The pattern here, bolt sensors and software onto equipment you already own instead of buying new automated machines, is a cheaper path to automation that could show up in trucking yards, warehouses, mines, and construction sites long before those industries buy fully robotic equipment from scratch.
It also raises a question insurers and equipment owners have not fully answered yet: who is liable when an unmanned machine worth a fraction of a new one is running through a field, or eventually a job site, with nobody physically present to stop it.