Orbital Watch: Technical Whitepaper
Building a Real-Time AI Intelligence Platform with Zero Infrastructure Costs.
1. Overview
Orbital Watch is a sophisticated missile tracking and real-time intelligence platform that showcases the power of Agentic AI. By leveraging Gemini 3.1 Pro, the system automates global data harvesting, geospatial mapping, and geopolitical event analysis to provide actionable insights while keeping infrastructure costs effectively at zero through serverless cloud patterns.
The primary objective was to launch a production-grade, highly scalable interface capable of processing global breaking news in real-time, while maintaining a near-zero cost architectural footprint.
2. The Tech Stack
Frontend
- >> React 18 + Vite
- >> TypeScript (Strict Typing)
- >> @vis.gl/react-google-maps
- >> Glassmorphic CSS System
Backend
- >> Firebase (GCP)
- >> Cloud Functions (Node.js 20)
- >> Cloud Scheduler (3h Sweeps)
- >> Google Secret Manager
Intelligence Engine: Multi-Model AI Architecture
High-fidelity extraction of breaking news with native web-grounding for event discovery.
Broad strategic sweeps and pattern matching, optimised for token efficiency.
3. Key Methodologies
A. Multi-Phase Intelligence Extraction
Leveraging a dual-phase pipeline to ensure both speed and depth:
-
1
The Breaking News Pass (48h): Immediate threat identification via Gemini 3.1 Pro.
-
2
The Strategic Context Pass (15d): Historical verification and context gathering via Gemini 2.5 Flash.
-
3
Merging & Deduplication: Custom spatial-temporal algorithm to prevent duplicate reporting.
B. The "Zero-Cost" DB Strategy
By using Google Sheets as the primary persistence layer and CMS, we achieve infinite scalability for intelligence data with $0.00 monthly cost.
C. Estimated Geocoding
AI-driven "Approximate Geocoding" uses spatial reasoning to estimate coordinates from text, removing dependencies on expensive APIs.
4. Architecture Diagram
5. Conclusion
Orbital Watch demonstrates that by leveraging Serverless Architecture, Google Cloud's Free Tier, and Multi-Model Generative AI, a company can launch a global intelligence monitoring system in days rather than months, with effectively zero monthly infrastructure fees.
Engineering the future of AI-driven intelligence.