← All projects
SIT223 Design thinking · Deakin

AgriSense Cambodia

A voice-first concept that turns farm sensor data into spoken advice in Khmer, for farmers who can't use a dashboard.

IoT Design thinking React prototype Khmer + English
See the prototype ↓ The project it builds on
AgriSense prototype: a soil sensor alert in Khmer and English
3
new layers on an existing IoT system
5
prototype screens
2
languages, Khmer first
01

The problem

Farm sensors can tell you when soil is too dry. But most IoT dashboards show graphs in English and assume you can read data.

  • Many smallholder farmers in rural Cambodia can't use that kind of dashboard.
  • Numbers alone don't tell a farmer what to do next.
  • Farmers often sell through middlemen, with no direct way to reach buyers.
02

Where the idea came from

In 2022 I supported an agriculture hackathon program at Impact Hub Phnom Penh and saw these gaps up close.

For this unit, I extended DataBytes' Intelligent IoT Data Management project at Deakin, which already detects unusual sensor readings. My question: how would a farmer who can't read a graph actually use it?

03

The idea

  • Simple dashboard: icons and colours in Khmer instead of graphs.
  • Voice advisor: the farmer asks a question out loud in Khmer and gets clear advice back.
  • Marketplace: list a crop and connect directly with verified buyers.
04

The prototype

Five screens, built as a clickable React prototype. Every label is in Khmer first, with English underneath.

Home screen with soil, temperature, light and rain tiles
Home: field status at a glance
Soil sensor alert showing 18% moisture and advice to irrigate
Sensor alert: what's wrong and what to do
AI advisor chat in Khmer and English with a microphone button
AI advisor: ask out loud, hear the answer
Sell crops screen with crop, quantity, price and total estimate
Sell crops: set quantity and price
Buyers screen listing four verified offers for 120 kg of rice
Buyers: compare verified offers
Honest note: I generated the prototype with Lovable (an AI tool), then moved it into my fork of the DataBytes repo with Vite. It's a concept, not a tested product. I haven't put it in front of farmers yet.
05

How I planned it

  • A Lean Canvas covering problem, users, solution and key metrics
  • A Trello product plan with user stories and sprints
  • A pitch video and a design thinking reflection
06

What I'd do next

  • Test the voice advice with real farmers before building anything else.
  • Connect the screens to live sensor data from the DataBytes system.
  • Work with a UX designer. An AI tool got me started, but it isn't a substitute for good design.