AI Plant Management App ('Let Us Farm') Progress Report

"Let Us Farm" is our in-development mobile app that puts an agronomist in every farmer's pocket. It uses on-device machine learning to identify common crop diseases and pest damage from a phone photo, then recommends locally-available treatments.

The app is being built together with students from the QS Future 17 AI team, who are co-designing the diagnostic models and Luganda voice interface alongside our field team in Uganda.

Why offline-first

Most smallholders we work with have intermittent data connectivity. The diagnostic model runs entirely on the device — no upload required — so farmers get an answer in seconds, in the garden, on any Android phone.

Progress this quarter

• Trained the maize, cassava, banana and tomato disease classifiers on a locally-labelled dataset of 12,000+ images.
• Field-tested v0.4 with 60 farmers in Wakiso and Mukono; achieved 84% top-1 accuracy on real-world photos.
• Added a Luganda-language voice interface for non-literate users.

What's next

Beta release in Q3 2026 with a soil-management module built jointly with the AKS II Zombo team. If you'd like to test or fund the app, get in touch.