🍃 Smart Agriculture AI

AI + IoT assistant for farms & greenhouses

Smart farming help, in simple language.

Our assistant turns weather, soil, sensors and leaf images into clear, explainable advice on irrigation, fertilizer and pest risk — built for smallholder open-fields and modern greenhouses.

Try AI Demo I’m a Farmer I’m a Greenhouse Grower
PhD project (UK) • Edge-first (Raspberry Pi) • Explainable recommendations

MangoEdgeAI

Live Open-field orchard assistant

GreenhouseEdgeAI

Live Tomato greenhouse assistant

MultiCropEdgeAI

In progress Multi-crop expansion

Fresh green leaves with water drops

Today’s suggestion

Irrigate Plot 2 for 30 minutes this evening. Low rain risk

What the assistant can help with

Three core jobs — irrigation, fertilizer, and pest risk — designed to be practical, low-cost and understandable.

Irrigation • Save water

Decide when to water, how much, and for how long using soil moisture, crop stage, forecast rain probability and pump capacity.

  • Plot-wise recommendations
  • Rain veto to avoid waste
  • Works with or without sensors

Fertilizer • Save money

Stage-wise nutrition plans that respect local costs and reduce over-application, evolving from templates into data-driven optimisation.

  • Pre-flush → flowering → fruit set
  • Local fertilizer names & mixes
  • Farmer-first + explainable

Pest & Disease • Protect yield

Weather-based outbreak alerts plus image-based diagnosis from leaf photos and drone images — not just a leaf-scanner app.

  • Risk days via WhatsApp alerts
  • Lightweight edge vision models
  • Suggested next actions (not panic)

One platform, three products

Same edge + data pipeline, tailored to different environments. Start with MangoEdgeAI and GreenhouseEdgeAI — MultiCropEdgeAI expands next.

See features →

MangoEdgeAI — Orchard decision assistant

Built for small and medium mango orchards. Combines soil, weather and crop stage to recommend irrigation duration, stage-wise fertilizer plans and pest-risk warnings.

  • Edge-first: runs on Raspberry Pi
  • Offline-ready: syncs when internet is available
  • Explainable: every recommendation has a “why”
Open-field mode → See examples →
Farmer using mobile in the field

Example: farmer-friendly advice through simple interfaces.

GreenhouseEdgeAI — Climate & energy assistant

Designed for tomato greenhouses with multiple zones. Helps staff keep plants in comfort bands while reducing energy and CO₂ using explainable decision logic.

  • Zones: different greenhouse sections can be managed separately
  • Control guidance: heating / vents / screens / irrigation
  • Evidence: early research shows ~22% energy saving potential
Greenhouse mode → Research context →
AI detection illustration

Example: combining sensor + vision signals for risk detection.

MultiCropEdgeAI — scalable multi-crop expansion

A future platform supporting multiple crops (tomato, chilli, cotton, rice, groundnut and more) using a shared disease library + weather-risk + edge inference pipeline.

  • Same architecture: edge + weather + soil + images
  • Expandable disease library: crop-wise + severity labels
  • Goal: one practical tool for many regions
Partner with us → Project overview →

Under development

We’re expanding crop modules + datasets. Partners welcome.

Next: multi-crop rollout with the same edge-first design.

Dashboard preview (greenhouse-style)

Inspired by professional greenhouse software: zones, climate tiles, irrigation tiles and clear actions. (This is a design preview — the AI engine will connect behind it.)

Try demo →
Zone A Zone B Zone C All zones
Today External rain risk: Low • Wind: Moderate

Temperature

21.4°C

Comfort band: OK

Humidity

72%

VPD: stable

CO₂

820 ppm

Target: 800–950

Irrigation

Next: 18:30

Duration: 12 min

Climate trend

Temp vs setpoint

AI suggests minor vent adjustment to reduce humidity spike.

Energy & savings

Weekly view

Estimated saving: 18–25% this week (research prototype).

Actions

Explainable

  • Irrigate Zone A for 12 minutes (18:30)

    Why: soil trending dry + low rain risk + stable temp.

  • Check humidity spike risk (19:00–21:00)

    Why: forecast RH ↑ + limited airflow period.

  • Optional leaf image scan (pest check)

    Why: conditions match early outbreak pattern.

Request pilot access →

How it works

Collect data → run AI on the edge → send clear advice back to the farmer or greenhouse team.

  1. 1. Collect data

    Sensors (soil moisture, temperature, humidity), weather (IMD, NASA POWER), soil type, crop stage and optional images from phones or drones.

  2. 2. Analyse on the edge device

    Raspberry Pi or similar runs lightweight ML models locally so it works even with weak internet.

  3. 3. AI decision engine

    Irrigation + fertilizer + pest modules combine rules with ML to generate plot/zone-specific recommendations.

  4. 4. Explainable recommendations

    Every suggestion includes a “why” explanation, so people can trust it and override if needed.

  5. 5. Interfaces

    WhatsApp, app notifications or greenhouse dashboards — depending on what the user prefers.

Backed by research, built with farmers in mind

Ongoing PhD work in the UK, combining literature review, simulations and real-world pilots in India and the UK.

Academic foundation

Research-led design, transparent reasoning, and measurable outcomes.

Tech stack

Edge-first: Raspberry Pi • LoRa/MQTT • TimescaleDB • ML/TFLite.

Responsible AI

Explainable advice, farmer control, and safe decision support principles.

University of Portsmouth NASA POWER IMD / Local weather Raspberry Pi LoRa / MQTT