Aegis-docs

Roadmap

Dependency-aware phases to deliver Aegis from MVP to production-grade risk automation.

Overview

This roadmap delivers Aegis from deterministic MVP to production-grade risk automation, in dependency-aware phases.

Phase 1 — Foundations

  • worker onboarding and KYC flow
  • profile storage and H3 zone mapping
  • basic policy entity and tier activation
  • baseline weekly premium calculation service

Diagram for this phase is covered across the onboarding, architecture, and workflow sections.

Phase 2 — Trigger Engine MVP

  • integrate weather + AQI sources
  • build normalization pipeline
  • define threshold rules for initial covered events
  • publish disruption events over event bus

Diagram for this phase is covered across the disruption monitoring and tech/data sections.

Phase 3 — Claims and Payouts

  • eligibility checks against active policy + zone/time
  • auto-claim creation from disruption events
  • payment execution integration (Razorpay / UPI)
  • worker notification service (FCM/SMS/WhatsApp)

Diagram for this phase is covered across the claim automation and adversarial defense sections.

Phase 4 — Fraud Controls

  • GPS spoof and impossible-travel checks
  • duplicate identity and payout-link detection
  • manual review queue + claim status tooling

Diagram for this phase is covered across adversarial defense and the fraud gate decision logic.

Phase 5 — AI Risk Intelligence

  • worker risk scoring model rollout
  • dynamic premium model rollout
  • disruption forecasting for ahead-of-time pricing updates
  • fraud ML scoring layer for subtle anomaly detection

Diagram for this phase is covered across the AI models and risk scoring sections.

Phase 6 — Risk Pool Governance

  • loss ratio dashboards
  • portfolio exposure monitoring by city/zone/tier
  • premium margin and tier factor recalibration controls

Diagram for this phase is covered across performance/impact and the risk pool governance loop.

  • start with deterministic triggers and transparent eligibility logic
  • add ML gradually behind observability and fallback logic
  • keep payouts fast while maintaining fraud safeguards

On this page