What is TerraGuard?

Getting Started

New to TerraGuard AI? The User Guide is the best place to begin — a complete, screenshot-led walkthrough of every part of the platform, written for everyday users with no technical background required.

Overview

TerraGuard is a disaster event tracking and response coordination platform built for humanitarian organizations. It continuously monitors global data sources for natural disaster events — earthquakes, tropical cyclones, floods, volcanoes, droughts, and wildfires — and enriches them with AI-powered analysis, news discovery, and population exposure assessments.

The platform replaces manual monitoring workflows with an automated pipeline that ingests events from authoritative sources, correlates them with web intelligence, and delivers actionable alerts to response teams within minutes.

Key Features

Real-Time Event Monitoring

TerraGuard ingests disaster events from three authoritative data sources:

• GDACS (Global Disaster Alert and Coordination System) — All disaster types with alert levels
• USGS (United States Geological Survey) — Earthquake-specific data with precise measurements
• NHC (National Hurricane Center) — Tropical cyclone advisories and forecast tracks

Events are deduplicated, correlated across sources, and enriched with geospatial context automatically.

AI-Powered Analysis

Every event is processed through a multi-stage AI enrichment pipeline:

• News Discovery — Automated web searches find relevant news articles, situation reports, and official advisories
• Content Validation — LLM-based filtering ensures only genuinely relevant articles are indexed
• Knowledge Base — Crawled content is vectorized and stored for semantic retrieval
• AI Summaries — On-demand summaries synthesize all available information about an event
• RAG Chat — Ask questions about any event and get answers grounded in indexed knowledge

Population Exposure Assessment

The GeoPop service provides reverse geocoding and population analysis for every event:

• Affected population estimates within configurable radii
• Country and administrative region identification
• Land/sea detection for maritime events
• Exposure scoring that feeds into alert prioritization

Automated Notifications

Teams receive targeted alerts based on configurable matrices:

• Alert-level routing — RED, ORANGE, and GREEN events route to different teams
• Country configuration — Per-country deployment likelihood determines notification relevance
• Scheduled follow-ups — Automatic notifications at +2h, +12h, and +24h after event onset
• Multi-channel delivery — Email and Telegram with templated, information-rich messages

Comprehensive Reporting

An AI agent generates structured reports using a ReAct pattern with access to five MCP tool servers:

• Database search, vector search, document retrieval, geocoding, and web links
• 10-section reports covering situation overview, impact assessment, response recommendations, and more
• Rich text editing for manual refinement before distribution

Team & Organization Management

Multi-tenant architecture supports multiple organizations, each with:

• Teams and role-based access control
• Organization matrices defining event handling rules
• Country configurations with deployment likelihood settings
• Clerk-based authentication with Google sign-in

How It Works

1. Ingest — The Event Processor polls data sources on a schedule and writes normalized events to PostgreSQL
2. Enrich — The Backend API triggers Inngest jobs that search the web, crawl articles, validate content, and generate vector embeddings
3. Analyze — AI agents score events, assess population exposure, and prepare knowledge bases
4. Alert — Notification engine evaluates organization matrices and delivers targeted alerts
5. Present — The frontend displays events on an interactive map with filtering, detail views, and AI tools

Technology Stack

Next Steps

• Architecture Overview — Understand how the services connect
• Quick Start — Get the platform running locally
• Dashboard Guide — Learn the main interface