Case study
MundaGuard AI: agricultural intelligence that works in the field
Agricultural intelligence for disease detection, field operations and data-driven decisions — from single farms to enterprise scale.
- Status
- Live at mundaguard.com
- Mobile
- Android app coming to the store; usable in the phone's browser today
- Languages
- English, Bemba and Nyanja
- Built on
- The Tech Crafter Labs platform
The challenge
Help that reaches the field before the disease spreads
Crop disease spreads faster than help can reach the field. Growers need to know what is wrong, how sure that answer is, and what to do next — often on a basic phone, without a reliable signal, and in their own language.
- 01
Fields where the mobile signal comes and goes
- 02
A wrong diagnosis costs a harvest, so certainty matters as much as speed
- 03
Farmers, agronomists, cooperatives and extension officers all need different views
- 04
Advice must name the product, the dose and the timing — not just the disease
Our approach
Four decisions that shaped the product
01
Honest AI
MundaGuard runs its own crop-vision model. When it is not confident enough, the case goes to an agronomist instead of a guess going to the farmer.
02
Offline first
Crop checks work with no signal. They are kept encrypted on the phone and sent automatically when the connection returns.
03
In the farmer's language
The app speaks English, Bemba and Nyanja, so advice is understood the first time.
04
Independent services
Each module runs as its own service, so one can improve without putting the others at risk.
What we built
Ten modules, one season
Each module covers one part of a farmer's season and runs as its own service.
Crop diagnosis
A photo of the affected crop in, a finding out; unsure cases go to an agronomist.
Treatment advice
Next steps for every confirmed diagnosis: the product, the dose and the timing.
Smart crop advisor
A season plan for each crop, from land preparation to storage.
Farm management
Farms, fields, crop cycles and yields in one record.
Soil intelligence
Soil estimates for every farm from global soil maps, plus your own tests.
Weather intelligence
Hourly forecasts and a week-ahead crop-risk outlook per farm.
Learning library
Guides, videos and downloads linked from the advice they explain.
Extension network
Extension officers and experts publish alerts to the farmers they serve.
Notifications
Push, email and in-app messages, including district outbreak alerts.
Offline sync
Checks captured without signal are stored encrypted and sent when it returns.
How it works
From the field to a decision with MundaGuard AI
Offline-first by design
Step 01
Capture
Photograph the affected crop in the field.
Step 02
Diagnose
Models identify the likely disease from the image.
Step 03
Advise
Treatment and cost guidance, with weather and soil context.
Step 04
Sync
Works offline and synchronises when a connection returns.
Architecture
Engineered on the shared platform
MundaGuard AI runs on the same engineering foundation as every Tech Crafter Labs product.
- 01
Experience
- Web app with workspaces for farmers, agronomists and administrators
- Offline-first Flutter mobile app
- 02
Services
- Fourteen bounded-context services in Go and Python
- API gateway and event streaming with Kafka
- 03
Intelligence
- Self-hosted crop-vision model with a confidence threshold
- Feedback loop and retraining from confirmed cases
- 04
Platform
- Kubernetes with GitOps delivery through Argo CD
- OpenTelemetry and Prometheus observability
Work with us
Have a product like this in mind?
We bring the same engineering to partners: from first scoping to software running in production.