Produced by Warren E. Clark · [email protected] · +1.847.836.5100 mobile: voice, text & WhatsApp
Weekly Roundup: PerPlant’s sensor becomes a Valtra factory option, BloomX raises $13M for robotic pollination, Google opens a sorghum genetic atlas, and laser weeding reaches Indian farms.
The Signal: This week’s useful AI news was about route to market and public data, not demos. PerPlant’s crop sensor became a factory option on Valtra tractors. BloomX raised $13 million for robotic pollination. Google and ICRISAT opened a sorghum genetic atlas. Canada and China funded AI breeding that cuts years off variety development. Distribution and data are deciding which tools reach a field.
This week rewarded the companies that solved the boring part. PerPlant’s sensor is now fitted at the Valtra factory, which removes the dealer-integration step that stalls most precision-ag products. BloomX’s round included a pollination-services company and several kibbutzim, not just financial investors. Google’s open sorghum atlas and Canada’s funding for Céréla attack the same bottleneck from the data side: better varieties, found faster. And in India, Harvested Robotics is building a laser weeder sized for tractors Indian farmers already own. The pattern is consistent: AI in agriculture is moving from a demo to a distribution problem.
— Warren
Today’s Sponsor
AgPR — the news release distribution service for agriculture shares your news with top ag journalists, editors, broadcasters, bloggers, podcasters and content producers.
PerPlant’s roof-mounted AI sensor is now fitted at the Valtra factory. It will be offered on A105, A115, G, N, T, Q and S Series tractors in selected European markets. Sales start this month with Valtra’s sixth-generation T Series. Two cameras map crops and drive spot spraying, fertilizer, fungicide and growth regulator passes, day or night.
Why it matters: Factory fit removes the dealer-integration step that has slowed precision-ag adoption. PerPlant says users have cut input use by 40% to 87%, at about $2 per acre on a 494-acre farm.
AI Angle: Edge vision models turn every pass of the tractor into a crop map; the value shifts from the sprayer to the data layer.
BloomX raised $13 million in a Series A led by Elah Fund. It has now raised $20 million. The Israeli company builds robots and AI for blueberry and avocado pollination. Growers face unreliable bee supply and rising rental costs. BloomX says yields rose about 20% on average, based on its own figures.
Why it matters: Pollination is a bottleneck that capital can fix: the round included BioBee, a pollination-services company, and several kibbutzim, signaling industry operators are hedging the bee shortage.
AI Angle: Vision-guided robotic pollination needs per-flower detection and timing; the AI is the difference between a mechanical shaker and a reliable pollinator.
Google DeepMind and Google.org are building an open genetic atlas of sorghum with ICRISAT, HudsonAlpha and the University of Washington. It uses more than 44,000 seeds from over 90 countries in ICRISAT’s seed bank. Machine learning helps find promising plant crosses. Google also opened its farm models as a public preview API in 11 countries.
Why it matters: Open genomic data plus a public ag AI API lowers the cost of building crop models, especially for public breeders and startups that cannot license proprietary data. The Gujarat figure shows the API is already in production, not a demo.
AI Angle: Machine learning on genotype-phenotype data shortens the search for useful traits; the API puts the model layer in reach of third-party developers.
Harvested Robotics is developing RAKSHAK, a tractor-mounted laser weeder. Cameras and onboard computer vision tell weeds from crops. The machine carries up to four Class 4 lasers. It burns a weed in under 0.5 seconds and covers 80 to 100 weeds a minute. The company reports above 90% detection accuracy and about 0.2 inch targeting. It is in beta, two versions from production.
Why it matters: Laser weeding avoids both herbicide resistance and the soil disturbance of mechanical cultivation; an India-priced tractor implement could reach smaller farms than current autonomous weeders.
AI Angle: The models run on an onboard computer without internet, so the machine works in fields with no connectivity.
Canada will give Céréla Inc. up to C$571,521 over three years. The money comes through its AgriScience Program. Céréla will use AI to breed soybean and barley varieties years faster than older methods. It aims for barley that resists Fusarium head blight. It also wants soybeans without anti-trypsin factors for Eastern Canada.
Why it matters: Public money for AI breeding is a bet on cycle time: varieties that reach farmers sooner are the cheapest form of climate and disease adaptation.
AI Angle: Predictive breeding models rank crosses before field trials, cutting the number of seasons needed to find a viable variety.
China Agricultural University’s Shennong Guxin model combines genotypic data with farm knowledge graphs built on tens of millions of entries. CGTN reports it has cut breeding cycles by more than 30%, from five to eight years down to two to three. The report also cites 1,811 acres of AI pest monitoring and a 494-acre rice farm run by four people.
Why it matters: The claims are state-reported, but the scale is the point: breeding, pest monitoring and autonomous machinery are being deployed as one system, not as pilots.
AI Angle: Knowledge graphs plus genomic prediction are the breeding layer; computer vision and Beidou-guided machinery are the field layer.
CropX is supplying soil sensors and weather data to a McCain Foods research farm in New Zealand. Lincoln University runs the farm. The eight-year trial compares regenerative and conventional potato growing. It uses a 1:8 crop rotation. CropX probes at 24 or 35 inches will sit in every crop.
Why it matters: Long-term, whole-rotation data is rare; it is the evidence base growers need before changing practices at scale. McCain has committed to regenerative practices on all its potato acreage by 2030.
AI Angle: Sensor fusion plus models turn soil moisture, weather and imagery into irrigation and disease decisions, not just a dashboard.
