Beyond Disruption: Earning Trust and Driving Impact in AgTech Through Grower-Led Innovation

Editor’s note: “Cultivating Tomorrow” is a special series that shares insights from C-suite executives at leading AgTech companies, presented by AgTech PR. Its aim is to highlight the experiences of AgTech leaders driving agricultural transformation today. In this installment, Danny Bernstein, CEO of Reservoir, an early-stage venture capital firm, discusses why in the rugged, unpredictable world of open-air farming, the next leap in agtech isn’t just smarter software — it’s physical AI built to work in the dirt, alongside growers, from day one.

Innovation in agriculture has taken big steps forward — with drones mapping fields, satellites delivering real-time data, and software guiding decisions at scale. But there’s a layer of complexity closer to the ground for open-air farms — rugged terrain, fragile crops, unpredictable labor — that demands something more tactile. That’s where physical artificial intelligence (AI) comes in: technology built to operate in the dirt, alongside the crop, in real time.

Silicon Valley software culture shaped my career, where things move fast, break often, and scale is king. At Google and Microsoft, we obsessed over clean UX, seamless code, and shipping quickly. But agriculture plays by a different clock. The unpredictability of open air and fields doesn’t care about your roadmap or your next release. And the stakes — grower balance sheets, rural economics, ag communities, food systems — are deeply real.

That said, there’s something from Silicon Valley tech that feels worth bringing with us — not the hype, but the habit of questioning assumptions. What if startups worked in sync with growers from the start? What if we built tools that adapt to the timing of the crop, instead of forcing the crop to fit our schedule? What if the people designing the tech were in the field when it gets used?

That’s the thinking behind the Reservoir — an “Olympic Village” innovation community for agtech, starting in California, where startups, growers, and retailers work side-by-side. Ideas get pressure-tested in the field, not in isolation.

From Disruption to Ground-Level Results

Too often, agtech is built in isolation — by teams removed from the realities of the farm, designing for ideal conditions that don’t exist. But agriculture isn’t always a controlled environment or a software layer you can iterate on overnight. It’s shaped by unpredictable variables — weather, labor, regulation, crop biology — that don’t respond to roadmaps or pitch decks.

What we keep hearing from growers, retailers and producers is that real traction happens when technologists are hands-on, immersed in the day-to-day of the farm, not just the boardroom. And that’s backed by data. According to the 2024 CropLife/Purdue University Precision Agriculture Dealership Survey, the percentage of ag retailers using drones to apply crop inputs is projected to grow from 35% today to more than 50% by 2027.

AI-driven weed identification, once cutting-edge, is on track to be a standard offering for a quarter of dealers in the same window.

Yet for all the progress, the reality is sobering: less than 2% of specialty crop production is currently automated, according to the Western Growers Association. That’s a massive gap — and a massive opportunity.

These numbers make one thing clear: growers adopt technology when it’s developed with their input, not handed to them after the fact.

CropLife’s research reinforces a core truth: in agtech, trust isn’t optional — it’s the foundation. Retailers continue to be key advisors, helping growers navigate an increasingly complex landscape. But trust doesn’t stop at the grower. Farmworker buy-in matters just as much. The tools that succeed aren’t just aligned with business goals — they work on the ground, with the people who use them. The most effective innovations are built collaboratively and tested alongside those in the field — not handed down from above.

If there’s one lesson tech folks need to absorb, it’s humility. The land is the real proving ground — and it doesn’t care how slick your deck is. The best ideas start with listening. That’s why we embed startups directly on working farms, shoulder-to-shoulder with the people who know the land best.

Making Physical AI Real

At the Reservoir, we’re focused on closing the gap between early prototypes and field-ready products. Our incubator helps startups go from “this should work” to “this actually works” — in real dirt, with real constraints. Like Missouri is the Show-Me State, agriculture is the show-me industry. Trust is earned through proof — tools need to work, in demos, and then in real conditions.

Physical AI is central to this shift. It’s software-enabled hardware — robots, sensors, machines—that can survive and adapt to the messiness of open-air farms. These aren’t controlled environments. They’re chaotic, living environments that demand fast decisions, nimble movement, and rugged design.

So why is now the right time? Robotics costs are dropping. Machine vision and edge compute are getting better. Simulated environments are a powerful starting point — they help refine ideas quickly and safely. But real-world testing is the critical next step. Dust, wind, moisture, shifting light, and labor flow introduce variables that no model can fully capture. That’s where theory meets reality — and where most tools are truly tested.

That’s why we partnered with Tanimura & Antle, Western Growers, and Naturipe to stand up a 40-acre site in the Salinas Valley, planted specifically for iteration. Here, startups build in sync with growers, soil, and harvest cycles — shortening the loop from idea to usable product.

And this isn’t just about California. We’re already hearing from growers across the U.S. and internationally. The appetite for a more connected, grounded model is everywhere.

Real progress in agriculture starts in the field — with people who understand the crop, the season, and the stakes. That’s where good ideas take root, and where the future of agtech is being built.