Sri Lanka's Agricultural Crisis Is an AI Opportunity

In 2021, Sri Lanka's government banned the import of synthetic fertilizers. The ban lasted approximately seven months before being reversed. In that period, rice yields fell by 40–50% in affected regions. Tea production fell. The country, already under foreign exchange pressure, faced a food security crisis on top of the economic collapse.

The fertilizer ban was a policy decision made with limited data about its consequences. The agricultural monitoring infrastructure to predict those consequences did not exist. The climate and soil modeling tools to design a managed transition — if one were feasible — were not deployed.

This is not a criticism of a specific government decision. It is an observation about the information environment in which agricultural decisions are made in Sri Lanka.

The structural vulnerability

Sri Lankan agriculture has three primary vulnerabilities:

1. Monsoon dependence — rice agriculture in particular is highly sensitive to monsoon timing and volume
2. Limited precision in input application — fertilizer, pesticide, and water use are not optimized at the plot level
3. No real-time crop stress monitoring — disease and drought stress are identified through manual field inspection rather than satellite observation

None of these are new problems. All three are tractable with existing AI and remote sensing technology.

What ASI Research Lab is building

Monolith 03's initial research focus is on three crops: rice, tea, and coconut. These are Sri Lanka's primary agricultural exports and food crops. The research agenda:

Crop yield prediction models: Using historical yield data from the Sri Lanka Department of Agriculture, soil classification maps, and 20-year monsoon pattern data, build regression models that predict yield ranges under various climate scenarios. The goal is not a perfect prediction — it is a decision support tool for farmers and agricultural planners.

Satellite-based crop stress detection: Sri Lanka's agricultural land is largely within the coverage of Sentinel-2 (ESA) and Landsat-9 (NASA) satellite programs. Both provide free, high-resolution multispectral imagery. NDVI and derivative indices provide reliable early indicators of crop stress. Building an automated processing pipeline that converts satellite imagery to actionable field-level alerts is an engineering problem more than a research problem — and it is solvable with existing tools.

Monsoon-adaptive irrigation scheduling: The most water-intensive crop in Sri Lanka is rice. Paddy irrigation decisions are currently made on fixed schedules that do not account for real-time soil moisture or incoming rainfall forecasts. A model that integrates soil sensor data with short-range weather forecasts to optimize irrigation timing can meaningfully reduce water waste and increase yield reliability.

The DOA partnership

The Sri Lanka Department of Agriculture partnership is an active relationship, not an aspiration. DOA provides access to historical yield data, soil classification maps, and field trial data from its existing research programs. In return, ASI Research Lab provides research outputs that DOA can deploy in its extension services.

The first live pilot deployment target is within six months of site launch. The scope of the pilot is a yield prediction model for rice agriculture in the North Central Province — the primary paddy-growing region.

This is narrow, specific, and achievable. That is how research progresses.