DEV Community

Gabriel Mahia
Gabriel Mahia

Posted on

Building parametric crop insurance with wapimaji-mcp and bima-mcp

#insurance #mcp #africa #climate

Building Parametric Crop Insurance with wapimaji-mcp and bima-mcp

Traditional crop insurance in Kenya requires an adjuster to visit the farm, assess the damage, file a claim, wait for processing, and — maybe — receive a payout months after the harvest has already failed.

Parametric insurance replaces the adjuster with satellite data. If drought reaches Phase 3 in your county, the policy pays out automatically. No adjuster. No waiting. No dispute.

Here's how to build a basic parametric insurance agent using two MCP servers.

The tools

wapimaji-mcp provides Kenya NDMA drought phase classifications across all 47 counties:

# Tool: kenya_county_drought
# Input: county_name
# Output: current drought phase (1-5), affected population, area
Enter fullscreen mode Exit fullscreen mode

bima-mcp provides parametric risk scoring and insurance product matching:

# Tool: parametric_crop_risk
# Input: county, crop_type, area_ha, coverage_amount
# Output: risk score, premium estimate, trigger conditions
Enter fullscreen mode Exit fullscreen mode

The agent workflow 

User: "I have 2 hectares of maize in Turkana County. How much will crop insurance cost?"

Agent:
1. wapimaji_mcp.kenya_county_drought(county="Turkana")
   → Phase 4 (Emergency), 67% area affected

2. bima_mcp.parametric_crop_risk(
     county="Turkana", 
     crop_type="maize",
     area_ha=2,
     coverage_amount=50000
   )
   → Risk score: HIGH
   → Estimated premium: KES 4,200/season
   → Trigger: Phase 3+ for 3+ consecutive weeks
   → Payout timeline: 14 days after trigger
Enter fullscreen mode Exit fullscreen mode

Why the combination works

Neither tool alone answers the question. The drought data provides current conditions; the risk tool prices the forward-looking exposure. Together, they give a complete picture in a single agent session.

The farmer gets: a current drought assessment, a premium quote, and the exact trigger conditions — without calling a broker, visiting an office, or waiting for a human intermediary.

This is the practical argument for coordination infrastructure. The data exists. The tools exist. The value comes from connecting them.

Demo data disclaimer

Both tools use synthetic demo data. In production, these would connect to real NDMA APIs and accredited insurance product APIs. The architecture is production-ready; the data sources need to be swapped.

Source code:

Top comments (0)