Project Case Study: 001

Orbital Watch: Technical Whitepaper

Building a Real-Time AI Intelligence Platform with Zero Infrastructure Costs.

Agentic AI Serverless Gemini 3.1 Pro Firebase

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

Primary: Gemini 3.1 Pro REAL-TIME

High-fidelity extraction of breaking news with native web-grounding for event discovery.

Secondary: Gemini 2.5 Flash HISTORICAL

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. 1
    The Breaking News Pass (48h): Immediate threat identification via Gemini 3.1 Pro.
  2. 2
    The Strategic Context Pass (15d): Historical verification and context gathering via Gemini 2.5 Flash.
  3. 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

flowchart TD A[Cloud Scheduler] -->|Every 3 Hours| B[Cloud Function] B -->|Phase 1: 48h| C[Gemini 3.1 Pro] B -->|Phase 2: 15d| D[Gemini 2.5 Flash] C --> E[Data Merger / Dedup] D --> E E -->|Write| F[Google Sheets API] F -->|Persistence| G[(Google Sheet)] H[React Frontend] -->|Read| G H -->|Manual Fetch| B H -->|Visualization| I[Tactical Map Overlay] style A fill:#D31719,stroke:#fff,color:#fff style B fill:#333,stroke:#fff,color:#fff style C fill:#444,stroke:#4ade80,color:#fff style D fill:#444,stroke:#4ade80,color:#fff style G fill:#064e3b,stroke:#059669,color:#fff style H fill:#1e3a8a,stroke:#3b82f6,color:#fff

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.

Developed by Casson Media
Engineering the future of AI-driven intelligence.