The Startup Teaching One AI Brain to Run an Orchard and a Kitchen

For decades, agricultural robotics meant one machine doing one job: a picker that picks, a sorter that sorts, a packing line that packs. A Cornell startup is betting that model is about to break. Ficio, founded by Cornell M.Eng. candidate Youhanna Meleka, is building FoodMind, an AI platform designed to be hardware-agnostic, one that learns from different machines and environments so the same underlying model can grade apples in an orchard and prepare potatoes in a commercial kitchen.

The company came into focus this week with two products running on that platform. FieldSorter rides through orchard rows using computer vision to identify and sort apples with defects such as bruises, punctures, or clipped stems. FicioPrep is a robotic station built to automate food preparation in large-scale eateries, where workers can spend hours chopping, slicing, and dicing ingredients for thousands of meals. One brain, two bodies, two ends of the supply chain.

Sorting at the speed of harvest

The economics of apple growing explain why FieldSorter matters. Today, growers harvest fruit, send it to a packing house, and wait months in controlled storage before learning which apples can actually be sold as fresh fruit. Ficio's system makes that determination in the field, at harvest time, recording each apple's grade and linking it to the tree where it grew. A delayed verdict becomes live data, and farmers learn about their orchards while they can still act on it.

"Growers spend a lot of time and money finding out which apples are market-ready and which aren't. We're helping them make that determination at harvest rather than months later in a packing house," Meleka said.

Earlier this year, the company demonstrated FieldSorter at the Cornell AgriTech campus in Geneva, New York, in front of roughly 300 growers. It is now running field pilots with farms in Ontario and Wayne counties, refining the system on direct grower feedback.

Then it goes to the kitchen

FicioPrep is the other end of the same idea. The robotic station uses computer vision to recognize ingredients and assess potato quality, then autonomously prepares them into a desired cut: fries or chips. The plan is for the system to eventually handle a range of vegetables, prepare ingredients on demand rather than in bulk hours ahead, and self-clean from the inside. It also tracks how much food it processes, giving kitchen managers operational data they do not currently have. The company has already demonstrated the prototype to managers at Cornell Dining to gather feedback from the people who would actually run it.

One brain, two bodies: Ficio's bet is that the winning food robot will not be a better machine, but a smarter platform running underneath them.

Why now: the labor math

Ficio is arriving as the orchard economy strains. A Cornell-led research project backed by a four-year, $7.5 million grant from the USDA's Specialty Crop Research Initiative is developing autonomous robots for pollination, fruit thinning, apple harvesting, and inter-row weeding, and it will establish a Center of Excellence for Orchard Robotics at Cornell. Nine organizations are collaborating, including Washington Fruit and Produce Co. of Yakima, Washington, which grows apples and cherries on about 10,000 acres. Field robots are moving from research projects to commercial pilots across the industry.

The numbers behind that project are stark. Fifteen years ago, labor accounted for about 45 percent of Washington Fruit's total costs. Today it is more than 60 percent. Apples are the most-consumed fruit in the United States, contributing $21 billion a year to the national economy, yet apple prices have barely moved in two decades. The squeeze lands on growers, which is exactly why capital is flowing into machines that do what hands used to.

The pattern is global. In India, the startup Segritech, named to YourStory's Tech30 list of the country's most promising startups of 2026, is deploying AI-driven fruit grading machines and developing autonomous harvesting robots with tactile sensors for delicate fruit like pomegranates. It plans to deploy more than 100 grading machines over the next 12 to 18 months.

The mission behind the machines

For Meleka, the project is personal. He grew up in Cairo, Egypt, where food is not always abundant, and he came to Cornell after studying logistics and manufacturing at Penn State and working as a global supply manager at Tesla. His framing is blunt: waste happens throughout the supply chain, and the goal is to move food from farm to fork with the least waste possible, which in turn improves cost and quality.

That is what makes the FoodMind thesis interesting beyond any single product. The orchard, the packing house, and the commercial kitchen are three stations in one supply chain, and until now they have been automated by three different industries that rarely speak to each other. Ficio's bet is that a single platform can span all three.

What comes next

If the platform thesis holds, the economics of food automation change. A grower or kitchen operator would not need a separate AI team for every machine; they would buy bodies and license brains. Meleka describes FoodMind as a foundational model that learns across different body morphologies and can adapt to new tasks without starting from scratch each time. That is the pitch, and it is a familiar one from the history of computing: general-purpose software layers sitting above specialized hardware.

There are reasons to stay measured. Ficio is early, at the stage of prototypes, pilots, and demonstrations rather than deployments. The company has not announced funding or commercial timelines, and hardware-agnostic AI in unstructured environments is one of the hardest problems in robotics. Its current scaffolding is serious, though: incubation at Rev: Ithaca Startup Works, mentors from Entrepreneurship at Cornell, an industry partnership with Universal Robots, and membership in NVIDIA's Inception program for AI startups. Whether FoodMind learns fast enough to outrun the capital burn of a hardware startup is the open question.

But the direction of travel is hard to argue with. Labor costs keep rising, growers keep getting squeezed, and the machines keep getting smarter. If Ficio is right, the food supply chain is about to get something it has never had: a nervous system.