❌

Normal view

There are new articles available, click to refresh the page.
Before yesterdayGeekWire

Data farmers: Ag-tech startup Aigen grows AI-generated weeds to train its solar-powered farm robots

23 October 2026 at 10:33
Aigen’s autonomous, solar-powered robot Element. (Aigen Photo)

For years, Aigen has been building and tweaking solar-powered robots designed to pull weeds on farms. But the Redmond, Wash., startup is also raising a crop of its own: a growing, AI-generated collection of digital weeds and plants that teaches those robots what to pull.

That synthetic data, which Aigen calls Alchemy, is changing what the company can sell. Training its robots on a new crop used to require three to four years of collecting and labeling real-world images, co-founder and CTO Richard Wurden said. Now it takes about a week, helping Aigen attract growers of asparagus, hops, flowers, lettuce and more.

Wurden spoke with GeekWire by phone from FIRA USA, a three-day agricultural robotics expo at Yakima’s State Fair Park that wrapped up Thursday. He said Aigen made several sales at the show, including to growers it previously couldn’t serve.

β€œNow we’re saying yes to a bunch of crops, where previously we were locked into just doing a few,” he said.

Aigen is nearly sold out of robots for the 2027 season following the show, according to the company.

Aigen’s solar design is also getting more attention as fuel costs climb. U.S. diesel prices are up roughly 70% from a year ago, according to EIA data, and averaged nearly $6.30 a gallon this month, according to AAA. Many rival weed-control machines are pulled by diesel tractors, Wurden said, while Aigen’s robots run entirely on solar panels with a battery backup.

β€œIt just so happens running a solar-powered robot without pesticides is the best possible timing right now,” he said.

Aigen co-founders Kenny Lee and Richard Wurden. (Aigen Photo)

Wurden and CEO Kenny Lee founded Aigen in 2020 after meeting in a climate-focused Slack group and discovering they lived about five minutes apart. Wurden, a mechanical engineer, previously worked on battery systems at Tesla and at Seattle electric boat maker Pure Watercraft. Lee is a cybersecurity veteran who co-founded Weblife.io, which Proofpoint acquired in 2017.

The mission is personal for both founders. Lee had non-Hodgkin’s lymphoma and Wurden has type 1 diabetes β€”Β diseases Wurden said have been linked to herbicide use.

The startup raised a $4 million seed round in 2022 and a $12 million round in 2023 led by ReGen Ventures, with participation from NEA and others. Wurden said the company has raised about $26 million in total. The startup now has about 15 employees.

Aigen’s Element robots are about 6 feet wide and topped with a roughly 400-watt solar panel. Cameras and AI distinguish weeds from crops, and mechanical arms pull the weeds out at the root. The robots cost $50,000 to $70,000 each depending on configuration, and Aigen typically sells them in packs of five, Wurden said.

Aigen has built more than 100 robots, all assembled in Redmond, and logged more than 15,000 hours in fields in California, Minnesota and North Dakota. Customers include organic soybean growers in North Dakota and tomato growers in California, Wurden said, and the company is expanding a recent pilot with an unnamed Fortune 500 company.

Alchemy has also become a product in its own right.

Large agribusinesses are using Aigen’s synthetic data, Wurden said. Because the plants are generated in simulation, each one comes perfectly labeled. That matters when weeds and crops overlap in ways that are hard to sort out from real photos, especially under shifting daylight. Aigen’s robots don’t use artificial lights. Wurden said the startup built the system out of necessity, because it couldn’t afford to pay people to label images by hand.

β€œOur AI now labels the data for us,” he said.

The company is choosy about who gets that data. Wurden said the company would work with precision-spraying companies that use less herbicide, but not with operations that broadcast-spray fields.

Next up: robots on Washington farms early next year, the company says. It’s also developing an orchard version with arms that reach out to weed between trees. And a next-generation robot is in the works.

Aigen is one of several Washington startups using AI to fight weeds. Issaquah-based TerraClear maps individual weeds so farmers’ existing sprayers hit only those spots. Seattle-based Carbon Robotics zaps weeds with lasers and trained its AI on 150 million labeled images of real plants. Aigen is betting synthetic data can teach its robots new crops faster.

Wurden said it’s refreshing to see technologists working on these problems rather than focusing on ad clicks or another gadget or service that may not be good for society.

β€œI support every company working on ag technology and things that help farmers, because farmers have felt like they’re left behind for a long time,” he said. β€œIt’s a good way to cross the aisle in politics, too. Everybody needs food.”

Agility’s new Digit 5 robot lifts 50 pounds, works 20+ hours a day and operates alongside people

15 September 2026 at 07:00
Two views of Agility Robotics’ new Digit 5, which is designed to work alongside people without safety barriers. (Agility Robotics Photo)

Agility Robotics unveiled the next version of its Digit humanoid robot, which it says is designed to work safely alongside people in warehouses and factories without the barriers that are usually required.

The Salem, Ore.-based company, which is preparing to go public through a $2.5 billion SPAC merger, said Digit 5 is its first humanoid built for β€œcooperatively safe” work at scale.

The robot uses AI and multiple sensors to watch for nearby people. Depending on the situation, it steers around them, stops or sits down. Visual and audio cues signal what it’s about to do, and a separate safety controller triggers its responses when someone gets too close.

Digit 5 can lift up to 50 pounds, 40% more than its predecessor, and reach objects as high as 7.2 feet off the ground, up from 5.5 feet, according to the company. A new battery runs for 90 minutes and recharges in 9 minutes, which Agility says allows more than 20 hours of work in a 24-hour day. Its grippers can be swapped for different jobs.

β€œWe built it to the exact requirements our customers gave us after three years of Digit 4 working on their production floors,” said Agility CEO Peggy Johnson, a former Microsoft executive and Magic Leap CEO, in the announcement.

Digit 5 signals that it has spotted a nearby worker in this rendering. (Agility Robotics Image)

Digit 4, the current model, has mostly moved totes, logging more than 65,000 hours at customers including GXO, Schaeffler, Amazon and Toyota’s Canadian manufacturing arm, according to the company. Agility says Digit 5 will take on more of a facility’s work β€” unloading pallets, tending machines, gathering parts and inspecting items before loading them back onto pallets for shipment.

Amazon, an early Agility investor, has run multiple pilots of earlier Digit models, including one GeekWire observed at a warehouse south of Seattle in 2023. It bought five Digit robots for $1.05 million, according to Agility’s S-4 registration statement, filed with the SEC this month in connection with its merger with Churchill Capital Corp XI, a special purpose acquisition company.

Agility says it had more than $300 million in multi-year orders for Digit 5 as of May, contingent on hitting certain milestones. According to the S-4, those orders are from one customer, for 1,000 robots over three years. The filing doesn’t name the customer, but it also describes a May deal for at least 1,000 robots with a β€œrelated party,” an accounting term for those with close ties to a company, such as major investors.

Agility had $1.78 million in net sales in 2025, up from $310,000 the year before. Its loss from operations doubled to $140.2 million.


Its auditor, Grant Thornton, included a going-concern warning in the report, citing substantial doubt about the company’s ability to keep operating. Agility said in the filing that the cash forecasts behind that conclusion didn’t take into account the SPAC deal, which the companies said in June could provide more than $620 million in gross cash proceeds before any shareholder redemptions.

In the S-4, Agility names Boston Dynamics, Tesla, Apptronik, Figure AI and 1X Technologies among its competitors, many of which it describes as well-funded.

Early access to Digit 5 is expected to begin in the first half of 2027, with general availability by the end of the year. The robots are assembled at the company’s RoboFab factory in Salem, which is designed to produce up to 10,000 robots a year.

❌
❌