AfyaRisk 2.0

Enterprise Insurtech & Actuarial AI Platform for Africa

Afya means health in Swahili. AfyaRisk 2.0 is an AI-powered actuarial engine built specifically for the African health insurance market — unifying five critical workflows into a single real-time API platform that any insurer can plug into their existing operations.

The Problem

Risk Mispricing

Underwriters price premiums using static age-based tables with no clinical data — leading to loss ratios exceeding 80%.

Fraud Leakage

Manual claim review catches less than 30% of fraudulent submissions — costing African insurers billions annually.

Reserve Delays

Actuaries calculate IBNR reserves quarterly in Excel — creating regulatory non-compliance risk and delayed disclosures.

Policy Opacity

Agents answer policy questions by searching PDFs manually — slowing claims and degrading policyholder experience.

Siloed Systems

Each workflow runs independently with no shared risk signal — no unified view of a policyholder's true risk.

The Result

Mispriced risk pools · Fraud leakage · Delayed reserves · Underserved policyholders

The Solution — 5 Engines, 1 Platform

Clinical Risk

Random Forest

Scores patient biomarkers — Low / Medium / High risk tier + 3yr survival probability

Policy Pricing

Actuarial Pure Premium

Risk-adjusted premium using (Frequency x Severity) / (1 - Loss Ratio - Expense Loading)

Fraud Detection

Isolation Forest

Anomaly detection on claims — 0 to 100% fraud score + APPROVE / REVIEW / REJECT

IBNR Reserving

Chain-Ladder

Projects ultimate losses from development triangles — reserve requirement per accident year

Policy RAG

FAISS + Transformers

Semantic vector search over policy documents — matched clauses in milliseconds

What Makes AfyaRisk Different

Real-time

All five engines respond under 500ms via REST API

Explainable

Every output includes reasoning — flags, factors, scores

Integrated

Risk score feeds directly into premium calculation

African-first

Premiums in KES, ICD-10 codes for local disease burden

API-first

Any insurer system integrates via standard REST endpoints

No cold start

RAG model preloaded at startup — instant first query

Built with IBM Bob

AfyaRisk 2.0 was engineered end-to-end using IBM Bob as an agentic AI software development assistant — covering architecture specification, backend ML engine development, frontend dashboard, bug diagnosis, performance optimisation, deployment configuration, and documentation. Every engine, every fix, every commit in this repository was produced through Bob.