A farmer I spoke to recently told me he could imagine running an electric weeder and a laser weeder on the same pass. Not competing, just doing different jobs. The electric weeder could handle the bulk of the weeds while the laser would pick up what's left. The combination would make the overall job faster and more effective. More broadly, he saw the electric weeder replacing broadcast, shield, and banded spraying, while precision tools would still maintain an edge in other use cases.

He wasn’t necessarily working from integrated pest management theory. He was just describing his farm and what he'd actually use.

That observation has stayed with me because it captures something I think the weed tech debate mostly misses. The farmers closest to the problem already understand that no single tool is going to solve it. The question is whether the founders and investors building the next generation of solutions are listening.

Beyond seductive silver bullets

We need new weed management tools urgently. Vermont just banned paraquat, joining more than 70 countries worldwide. Sixty percent of Australia's highest-volume pesticides are banned in the EU, representing a significant trade risk for a country that exports nearly $80 billion in agricultural produce annually. Resistance is documented in 23 of the 25 commercially relevant herbicide modes of action. New chemistry is not coming fast enough. The pressure is real and mounting, and non-chemical weed management is a key part of the answer.

But non-chemical approaches can't escape the silver bullet trap just by being non-chemical. Shane Thomas has written about this recently, highlighting how weeds adapt to physical and mechanical mechanisms as well as chemical ones. 

Remembering the lesson from Bt cotton 

The history of Bt cotton is worth revisiting here. When insect-resistant Bt cotton came to market, growers naturally wanted to plant it everywhere because it worked and it maximised returns. The system only kept functioning because regulators and seed companies enforced refuge plantings: a small portion of every farm planted with non-Bt cotton to maintain susceptible insect populations and slow resistance. Individual incentives pointed one way. System health required something different. The intervention worked: resistance has been slower to develop in countries where compliance is high.

If we adopt non-chemical weed technology today, what is the equivalent of the refuge requirement for this system? How can we include more diversity in weed management systems and ensure that these technologies get deployed to maximise the benefits to individual growers while supporting long term efficacy for all? 

These questions aren’t reasons to stop backing these companies. But they are questions I want us all, especially founders, to have thought about. 

The investor view

I understand why many founders might prefer a bold claim about breaking the herbicide resistance treadmill over the more measured ‘we're one useful tool in a complex integrated system.’ The lack of nuance is a frustrating reality of how early-stage funding works. The conversation after the pitch needs to go beyond the initial simplification. 

The first thing I'm looking for is systems thinking. It’s a red flag when a founder can tell me exactly how their technology kills a weed but not how a grower would fit it into a season, or what other tools it needs to work alongside, or what might happen to weed populations over a decade of use. What gives me confidence is a founder who can move from ‘here's our wedge, here's our beachhead’ to ‘here's where we sit in a grower's rotation, here's what we do that herbicides can't, here's what we'd run alongside.’

The second thing I'm looking for is platform thinking and technologies. There's a meaningful difference between a company that has optimised a single product for a single use case and one whose underlying technology can flex. In biological crop protection, that's the difference between a company with one product and one with the capability to screen or design across many targets. In physical weed control, it's the difference between a mechanism suitable to a single cropping context and one that is designed to adapt and be adjusted for different weed types, growth stages, or field conditions. The commercial case for the latter is straightforward: a larger addressable market and lower dependence on any single use case succeeding. But there's an agronomic argument too. A narrow mechanism or deployment creates concentrated selection pressure and risks early resistance. A platform that can evolve alongside the problem has a longer runway in both senses.

As an example in our portfolio, Azaneo has built a physical weed control mechanism that can be adjusted, via electrode geometry and specific dosing models, to work across different cropping systems. The technology can adapt to the problem, providing the kind of flexibility that makes a business interesting over a longer time horizon.

Getting to a clear articulation of this is a difficult thing to ask of an early-stage founder, and I understand why. The specific use cases might not yet be mapped. The platform might still be theoretical. But the founders who can talk candidly about this, even provisionally, are the ones I trust to build something that lasts long enough to matter. 

The farmer already knows

The farmer who imagined two machines doing different jobs on the same pass wasn't describing a product roadmap. He was describing how farming actually works: a system of layered, complementary interventions, managed over time, with no single answer.

We need the tools he's imagining. We need founders building them with that picture in mind. And we need investors who know the difference between a great pitch and a company that can back it up.

‍