Maharashtra signs MoU with WUR to scale AI and digital tech in agriculture
Maharashtra has signed a Memorandum of Understanding with Wageningen University & Research (WUR) to bring advanced AI and digital technologies into the state's farming programs. The agreement was finalized at the "AI 4 Agri 2026" summit at Bandra Kurla Complex, in the presence of the Chief Minister.
The MoU sets a framework for joint projects, with the state and universities identifying local needs before deployment. Early work will focus on sensor-based monitoring of crop growth and building AI models adapted to Maharashtra's soils, climate, and practices.
The five pillars of the collaboration
- AI and digital phenotyping: Use sensors and imaging to track plant traits, growth, and stress in real time.
- Digital breeding and seed systems: Apply data-driven methods to accelerate varietal selection and improve seed quality pipelines.
- Crop-specific pilot projects: Launch targeted trials for priority crops and regions to prove value before scaling.
- Knowledge exchange and capacity building: Train state teams, extension staff, and partners to run and maintain these systems.
- International collaboration and learning: Adapt proven European models to Indian conditions, backed by local data.
"We design systems with farmers, not for them," said Arun Kumar Pratihast, Senior Researcher at WUR, emphasizing citizen science and co-design through the full lifecycle-from data collection to model training and validation.
Why this matters for government teams
- Evidence-led policy: Continuous field data improves advisories on sowing, irrigation, fertilization, and pest management.
- Targeted subsidies and schemes: Better risk assessment enables sharper input support and climate adaptation measures.
- Extension at scale: Digital tools can amplify reach across smallholder communities while keeping advice context-specific.
- Research-to-field pipeline: Faster translation of agronomic research into farmer-facing tools.
Immediate actions for the state
- Form a joint working group: Agriculture, IT, planning, WUR, and state agricultural universities. Define a 90-day plan.
- Select priority crops and districts: Align with state programs and existing research plots for rapid setup.
- Stand up data infrastructure: Sensor procurement, data standards, API architecture, and a secure data lake.
- Agree on KPIs: Yield stability, input use efficiency, model accuracy, farmer adoption, and time-to-advice.
- Design citizen science protocols: Farmer cohorts, consent flows, incentives, and multilingual support channels (app/SMS/IVR).
- Capacity building: Train extension officers and district tech cells on tools, data quality, and feedback loops.
Citizen science: co-design with smallholders
The program will engage farmers from day one-data capture, feature prioritization, and field testing. This keeps models honest and advice usable.
- Recruit cohorts: Work through KVKs and district extension networks to onboard representative farmer groups.
- Simple data capture: Photos, short forms, and voice inputs; clear incentives and rapid feedback.
- Continuous validation: Compare model outputs with farmer observations and trial plots, then retrain.
- Local languages and offline modes: Reduce friction and improve participation in low-connectivity areas.
Data and governance guardrails
- Data ownership and consent: Plain-language consent, opt-out options, and farmer access to their data.
- Privacy and security: Anonymization, role-based access, and audit trails for datasets and models.
- Standards and portability: Open data schemas to prevent vendor lock-in and enable inter-operability.
- Responsible AI: Bias checks by agro-climatic zone; publish model cards and version histories.
- Procurement: Outcome-based contracts with clear service levels and handover of code/data artifacts.
12-month success metrics
- Adoption: Active farmers using the tools in at least three districts, with weekly engagement.
- Accuracy: Advisory precision (e.g., pest alerts, irrigation timing) benchmarked against control plots.
- Impact: Reduced input costs and fewer pest/disease losses for pilot cohorts.
- Capacity: Trained extension officers and district data stewards able to run pilots independently.
- Throughput: Time from field signal to advisory kept within agreed service windows.
Roles and responsibilities
- WUR: Model development, phenotyping expertise, tooling, and training.
- State Agriculture Dept.: Program ownership, budgets, coordination, policy alignment, and scale-up.
- State Agricultural Universities: Local agronomy, trial design, data generation, and model adaptation.
- Extension Network: Farmer onboarding, field support, feedback collection.
- Technology Partners (as needed): Sensors, connectivity, cloud, and MLOps support under open standards.
Next 90 days: a practical glide path
- Day 0-30: Charter the program; finalize data governance; pick pilot crops/districts; start procurement.
- Day 31-60: Deploy sensors and data pipelines; launch farmer cohorts; begin baseline measurements.
- Day 61-90: Adapt initial WUR models with local data; field-test advisories; publish first KPI dashboard.
For frameworks, checklists, and templates that help structure MoUs and pilot governance, see AI for Policy Makers.
Learn more about the research partner at Wageningen University & Research.
Your membership also unlocks:
