Event
LA Hacks
Team
Aamani Maddukuri
Gokul Nambiar
Emily Shen
Camilla C.
Cropt is an offline, audio-first crop disease detection app for farmers in low-connectivity environments. Farmers can scan a plant or describe symptoms, and on-device AI provides diagnosis and treatment guidance by voice. It also lets farmers alert nearby communities and sync outbreak data to a dashboard when internet becomes available.
What it solves
The inspiration for Cropt stems from a devastating reality in Southeast Asia: a crop disease that decimated 2.5 million acres, leading to an 80% yield loss. Information travels slower than the disease itself, leaving farmers like Amara—one of 500 million smallholder farmers globally—utterly helpless. These farmers often face the "triple barrier": low connectivity, low literacy, and limited education. When a crop shows signs of distress, they are forced to wait for hotlines or post blurry photos to crowded WhatsApp groups, often receiving advice too late to save their livelihood. We realized that for the next Amara, the solution couldn’t depend on the cloud or a stable signal; it had to be as immediate as the threat itself.
What it does
Cropt is an offline, audio-first crop disease prevention application designed specifically for low-connectivity environments. Using an "audio-as-interface" philosophy, farmers simply point their camera at a suspicious plant. Our on-device AI instantly analyzes the image to provide a diagnosis and specific treatment steps via voice. To foster community resilience, a single tap allows farmers to broadcast alerts to nearby neighbors or send automated WhatsApp messages to community groups. When internet access is available, the app utilizes a sophisticated conversational agent capable of switching between 70+ languages, while every scan contributes to a real-time global dashboard that provides agricultural organizations with unprecedented grassroots outbreak data.
How it works
The core of Cropt is built on breakthrough on-device AI integration. We leveraged Zetic’s NPU technology to run heavy analysis models locally on low-cost mobile phones without requiring internet or cloud processing. To make the app truly accessible, we designed a minimalist, audio-centric UI that bypasses literacy barriers. For the conversational layer, we integrated ElevenLabs to provide high quality, localized speech synthesis in over 70 languages. The backend infrastructure is designed to queue data locally and sync with a central dashboard only when connectivity is restored, ensuring no data point is lost.