Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Kissan-Dost: Bridging the Last Mile in Smallholder Precision Agriculture with Conversational IoT

Muhammad Saad Ali, Daanish U. Khan, Laiba Intizar Ahmad, Umer Irfan, Maryam Mustafa, Naveed Anwar Bhatti, Muhammad Hamad Alizai

TL;DR

Kissan-Dost tackles the critical gap between IoT sensing and actionable farming guidance for smallholders in LMICs by delivering sensor-grounded, multilingual advice through WhatsApp. The system combines off-the-shelf hardware with a retrieval-augmented LLM pipeline, ensuring grounding, provenance, and user-friendly interaction via voice notes. In a 90-day, two-site field deployment and a synthetic benchmark of 99 queries, the approach achieves high sensor-grounded accuracy (>90% correctness) and sustained daily engagement, outperforming dashboard-based interfaces. The work demonstrates that careful last-mile design—language fit, channel choice, and transparent grounding—can unlock the latent value of Agri-IoT for low-resource farmers and inform scalable deployments.

Abstract

We present Kissan-Dost, a multilingual, sensor-grounded conversational system that turns live on-farm measurements and weather into plain-language guidance delivered over WhatsApp text or voice. The system couples commodity soil and climate sensors with retrieval-augmented generation, then enforces grounding, traceability, and proactive alerts through a modular pipeline. In a 90-day, two-site pilot with five participants, we ran three phases (baseline, dashboard only, chatbot only). Dashboard engagement was sporadic and faded, while the chatbot was used nearly daily and informed concrete actions. Controlled tests on 99 sensor-grounded crop queries achieved over 90 percent correctness with subsecond end-to-end latency, alongside high-quality translation outputs. Results show that careful last-mile integration, not novel circuitry, unlocks the latent value of existing Agri-IoT for smallholders.

Kissan-Dost: Bridging the Last Mile in Smallholder Precision Agriculture with Conversational IoT

TL;DR

Kissan-Dost tackles the critical gap between IoT sensing and actionable farming guidance for smallholders in LMICs by delivering sensor-grounded, multilingual advice through WhatsApp. The system combines off-the-shelf hardware with a retrieval-augmented LLM pipeline, ensuring grounding, provenance, and user-friendly interaction via voice notes. In a 90-day, two-site field deployment and a synthetic benchmark of 99 queries, the approach achieves high sensor-grounded accuracy (>90% correctness) and sustained daily engagement, outperforming dashboard-based interfaces. The work demonstrates that careful last-mile design—language fit, channel choice, and transparent grounding—can unlock the latent value of Agri-IoT for low-resource farmers and inform scalable deployments.

Abstract

We present Kissan-Dost, a multilingual, sensor-grounded conversational system that turns live on-farm measurements and weather into plain-language guidance delivered over WhatsApp text or voice. The system couples commodity soil and climate sensors with retrieval-augmented generation, then enforces grounding, traceability, and proactive alerts through a modular pipeline. In a 90-day, two-site pilot with five participants, we ran three phases (baseline, dashboard only, chatbot only). Dashboard engagement was sporadic and faded, while the chatbot was used nearly daily and informed concrete actions. Controlled tests on 99 sensor-grounded crop queries achieved over 90 percent correctness with subsecond end-to-end latency, alongside high-quality translation outputs. Results show that careful last-mile integration, not novel circuitry, unlocks the latent value of existing Agri-IoT for smallholders.
Paper Structure (28 sections, 14 figures, 9 tables)

This paper contains 28 sections, 14 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Kissan-Dost: Design and Implementation
  4. System Overview
  5. Hardware Design
  6. Sensing
  7. Gateway
  8. Software Architecture
  9. WhatsApp interface
  10. Chained-prompt LLM pipeline
  11. Retrieval-augmented generation and knowledge base
  12. Agronomic data repository
  13. Weather and market data integration
  14. Orchestrator
  15. Personalised chatbot persona
  16. ...and 13 more sections

Figures (14)

  • Figure 1: Kissan-Dost System Architecture: data flow from sensors to the LLM-based conversational interface.
  • Figure 2: Inside view (left) and field deployment (center, right) of the Agri Sensor module. The enclosure houses an ESP32 microcontroller, a Li-ion battery, power regulation circuitry, and an RS485 interface.
  • Figure 3: Flow of user interaction (left) and sensor data (right) in Kissan-Dost.
  • Figure 3: Judge models
  • Figure 4: LLM-as-a-Jury evaluation. Error bars show 95% CIs over three runs ($N\!=\!3$).
  • ...and 9 more figures