The challenge
Farmers face mounting pressures: rising input costs, climate volatility, and increasingly complex decisions about maintaining profitability whilst transitioning to sustainable practices.
While at Agreena, I led research and experience design for an AI assistant designed to earn their trust.
Introducing Agatha
Agatha is an AI-powered assistant designed to provide farmers with data-driven insights and support for daily decision-making and farm management. Unlike generic large language models, Agatha is built specifically for agricultural contexts, integrating proprietary remote sensing data (such as field practice detection), AgreenaGro platform data, and web search with large language models.
Understanding the trust barrier
In early research interviews, scepticism about AI was widespread: "This wouldn't work for me because it can't know everything that's happened on the field." In one session, Agatha even recommended a product that was illegal in the UK. Learning from failures like these informed our approach to building trust.
When we reframed Agatha as verification not replacement of judgement, and led with time-saving features, sentiment shifted markedly. To improve specificity, a modular architecture was developed to access remote sensing data, grounding recommendations in farm-specific conditions whilst maintaining clear separation between systems.
We worked with data science to validate how Agatha balanced profitability vs. sustainability trade-offs correctly. But the real validation came from farmers: with transparency built into the user experience, they could cross-check outcomes against their own judgment.
Designing for adoption
In order for farmers to adopt any new system, they need to understand and evaluate it at the point of use. This shaped core design principles:
Structure and scaffolding. Scaffolded journeys guide early usage, helping farmers see value quickly. Information that's structured, visual, and interactive is more useful than walls of text. Workflows combine deterministic UX with probabilistic exploration elsewhere.
Transparency in data and usage. During onboarding, we provide clear summary of how Agatha works, its limitations and how data is used. Data provenance and usage is visible so farmers know exactly what’s being used at key moments. Internally, processes across teams were aligned to achieve a certification with Farm Data Principles (similar to GDPR for farming).
The competitive advantage: data and experience design
Proprietary data is Agatha's foundation. Whilst LLMs are commoditising, our remote sensing capabilities and integrated AgreenaGro platform data create insights no generic AI can replicate. But data alone isn't enough — the real advantage lies in how it's presented. Farmers want insights that integrate into their workflows, built on a genuine understanding of how they work. That's what's difficult to copy.
Agatha proves that the most powerful AI isn't the most complex — it's the most trusted. By embedding design thinking early and grounding decisions in real farmer research, we've created a system that makes farmers' expertise sharper, not their role redundant. That foundation of trust is what unlocks sustainable adoption, and at scale, that's how AI can create positive environmental impact.