Overview
This POC site takes any text content (articles, social media posts, news stories, etc.) and analyzes it against a wide range of philosophical, ethical, and theological frameworks to help people understand how the content aligns with or challenges established systems of thought.
Key Features
Content Analysis
Paste in text or provide a URL to analyze
AI examines the content against major teachings, traditions, and values
Highlights relevant sources and references (including religious texts, historical documents, or philosophical works)
Compares perspectives across different traditions and schools of thought
Paradox & Interpretation Explorer
Identifies potential conflicts or contradictions in different teachings
Shows historical development of varying interpretations
Provides detailed explanations of complex issues
Offers practical implications for how these affect everyday life and worldview
Multi-Tradition Perspectives
Compares how different traditions (religious, secular, or philosophical) view the same topics
Shows where perspectives align and where they diverge
Helps users gain a broader understanding of cultural and intellectual diversity
Source Integration
Automatically finds relevant passages from key texts (e.g., religious scriptures, philosophers, cultural works)
Explains how those sources relate to the content being analyzed
Shows different perspectives within and across traditions
Identifies when texts or schools of thought present differing or conflicting views
User Authentication & Credits
Secure login system with two-factor authentication
Credit-based system for API usage
Personal analysis history
Goal
The goal is to help anyone regardless of background make more informed decisions about the content they encounter by understanding how it relates to broader systems of belief, values, and interpretation. It encourages exploration of complexity and nuance while promoting deeper critical thinking.
The Orchestration System
1. Accepting Content
What you see: You paste text or drop in a link.
What happens behind the scenes:
If it’s text, the system checks that it’s valid.
If it’s a URL, it fetches the page, strips away ads, menus, and code, and keeps only the main article.
Everything is validated so the system knows it has clean content to analyze.
2. Pulling Out Key Ideas
What you see: The system figures out what the article is “really about.”
What happens behind the scenes:
Words are broken down, short/filler words are removed, and the most frequent/important terms rise to the top.
Standard concepts (like love, mercy, justice) are added to make sure key themes aren’t missed.
3. AI Analysis with Multiple Perspectives
What you see: You get a detailed breakdown of how the content lines up with key principles.
What happens behind the scenes:
The main analysis is done by GPT-4, guided by structured prompts.
A second AI (Anthropic Claude) double-checks, providing extra nuance and spotting things GPT-4 missed.
The system may also pull in current discussions or scholarly takes from the web for context.
4. Matching Against References
What you see: Relevant passages or teachings pop up, with explanations.
What happens behind the scenes:
The system searches an in-memory database of verses and teachings using the extracted keywords.
Matches are scored for relevance, and the alignment (positive/negative/neutral) is calculated.
Explanations include historical and cultural context, plus how different groups interpret the text.
5. Comparing Traditions & Exploring Paradoxes
What you see: You see where different groups agree, disagree, or wrestle with contradictions.
What happens behind the scenes:
The system looks up denominational perspectives and checks if they align with the source text.
If contradictions are found (in verses or teachings), it builds a structured explanation:
What the paradox is
How it developed historically
How scholars and traditions try to resolve it
6. Presenting Results Clearly
What you see:
An alignment score (0–100%) with plain-language labels (Strongly Aligned, Mixed, etc.).
Highlighted portions of your text with color-coded meaning.
Clickable references that take you to source material.
Suggestions for reflection or improvement.
What happens behind the scenes:
Regex highlighting marks important phrases.
JSON parsing ensures AI responses are clean, even if wrapped in markdown.
Results are formatted into a user-friendly view.
7. Keeping It Reliable
What you see: It “just works,” even if some AI services are down.
What happens behind the scenes:
If OpenAI or Anthropic fail, the system falls back to a local keyword-matching algorithm.
Built-in retry logic, timeouts, and error handling prevent stalls.
Sessions are securely managed in PostgreSQL, while fast lookups (verses, teachings) live in memory.
8. Technical Backbone (Simplified)
Frontend (what you click): A React/TypeScript app that talks to the backend.
Backend (what does the work): An Express server that handles input, calls AI, and manages data.
Data: PostgreSQL stores user accounts & sessions; Bible verses & teachings are cached in memory for speed.
AI Providers: Multiple (OpenAI GPT-4, Anthropic Claude, web search APIs) to get broad coverage.
In short:
The system takes in your content, cleans it, pulls out the main ideas, checks it against scripture and teachings, compares multiple viewpoints, resolves paradoxes, and then presents everything back to you in a clear, scored, and highlighted report. If one part fails, it has backups to make sure you still get results.