🧠

How Eagle Vision Works

Discover the advanced AI architecture powering your Bible study and sermon search experience

System Architecture

Eagle Vision is a sophisticated AI-powered platform built on modern cloud-native technologies

🤖 AI-Powered RAG System

Retrieval-Augmented Generation (RAG) combines information retrieval with AI generation.

We search 1,200+ sermons in real-time to provide accurate, source-backed answers.

Backend

  • Node.js Express.js RESTful API server
  • JWT Stateless authentication
  • OAuth 6 providers (Google, GitHub, etc.)

Frontend

  • React v19.0.0 modern UI
  • Tailwind Utility-first CSS
  • PWA Progressive Web App

Multi-Database Architecture

Four specialized databases work together for optimal performance

🗄️

PostgreSQL

Primary relational database for user data and sessions

⚡

Redis

High-performance caching and rate limiting

🔍

Elasticsearch

Full-text search across 1,200+ sermons

🧮

Pinecone

Vector database for semantic search

AI Service Integration

Multiple AI models for enhanced accuracy and reliability

🤖

OpenAI GPT-5.2

Primary

Latest flagship model with advanced reasoning and 400K token context window for comprehensive theological analysis

🧠

OpenAI o3-mini

Reasoning

Advanced reasoning model with 200K context window for complex doctrinal questions

⚡

OpenAI GPT-4o

Fallback

Fast, efficient model with 128K context window used as fallback when higher-tier models unavailable

Complete Technology Stack

Enterprise-grade technologies powering Eagle Vision

OpenAI Models

  • ▸
    GPT-5.2 - 400K context (primary)
  • ▸
    o3-mini - 200K context (reasoning)
  • ▸
    GPT-4o - 128K context (fallback)
  • ▸
    SDK v4.52.7

Vector Database

  • ▸
    Pinecone v4.1.0
  • ▸
    1,536-dimensional embeddings
  • ▸
    Cosine similarity search
  • ▸
    Multiple specialized indexes

Core Framework

  • ▸
    Node.js v18.0.0+
  • ▸
    Express.js v4.19.2
  • ▸
    RESTful API architecture

Authentication

  • ▸
    Passport.js v0.7.0
  • ▸
    bcrypt v6.0.0
  • ▸
    jsonwebtoken v9.0.2

Security

  • ▸
    Helmet v7.1.0
  • ▸
    express-rate-limit v7.1.5
  • ▸
    CSRF protection

Real-time

  • ▸
    Socket.io v4.8.1
  • ▸
    Server-Sent Events
  • ▸
    WebSocket support

Data Storage

  • ▸
    PostgreSQL v8.16.3
  • ▸
    Redis v5.6.0
  • ▸
    Elasticsearch v7.10.0

Caching Strategy

  • ▸
    Multi-level cache
  • ▸
    In-memory (NodeCache)
  • ▸
    Redis distributed cache

Testing Frameworks

  • ▸
    Jest v30.0.4
  • ▸
    Playwright v1.53.2
  • ▸
    Chai v5.2.1
  • ▸
    Sinon v21.0.0

Test Coverage

  • ▸
    Unit tests
  • ▸
    Integration tests
  • ▸
    E2E tests (3 browsers)
  • ▸
    Visual regression
🔍

AI-Powered Search

Semantic and full-text search across 1,200+ sermons

  • ✓ Hybrid vector + keyword search
  • ✓ Context-aware results ranking
  • ✓ Advanced filtering and facets
  • ✓ Multi-format export (PDF, DOCX, CSV, JSON)
💬

Ask the Message

AI question-answering with source verification

  • ✓ Real-time streaming responses
  • ✓ Automatic quote verification
  • ✓ Transparent source citations
  • ✓ Chart and visualization generation
📖

Study Mode

Comprehensive text analysis and annotation tools

  • ✓ Multi-color text highlighting
  • ✓ Inline note-taking
  • ✓ Organized bookmarking with folders
  • ✓ AI-powered explanations
⚡

Performance

Optimized for speed and reliability

  • ✓ Multi-level caching (in-memory + Redis)
  • ✓ Connection pooling
  • ✓ Rate limiting and DDoS protection
  • ✓ Progressive Web App (PWA) support
🎯 Quality Assurance

Every AI response goes through multiple validation steps:

  • Multi-source validation: Cross-referenced against multiple sermon sources
  • Quote verification: Every citation automatically validated against original text
  • Clear limitations: AI indicates when questions require pastoral care
  • Continuous improvement: Regular model updates for enhanced accuracy