AXIOS

A searchable database of 20,000+ laws, principles, and cognitive biases — from Newton to Kahneman — with a context analyzer that surfaces relevant patterns for any situation.

Zero Framework Multi-Agent Pipeline Context Analyzer PDF Export

20,835

Principles

326

Data Files

377K

Lines JSON

LLM Agents

10+

Disciplines

Live Interactive Tool

? What Is It?

▼

AXIOS answers the question: what are the invisible forces shaping decisions, systems, and outcomes? It's an interactive knowledge base spanning physics, psychology, economics, and 10+ disciplines — designed to help users discover patterns that apply across fields.

Cross-Discipline Discovery

See how a physics law applies to business strategy, or how a cognitive bias explains market behavior. One searchable interface for universal truths.

Context Analyzer

Paste any situation ("preparing for job interview") and get relevant principles with relevance scoring. 102KB analyzer parses free-form text.

Multi-Agent Research

12 specialized LLM agents run in parallel, auto-merging validated entries with deduplication across 326 JSON data files.

* Goal

▼

Build an interactive knowledge base of 20,835+ laws, principles, theories, and cognitive biases spanning physics, psychology, economics, and 10+ disciplines — designed to help users discover patterns that apply across fields.

The Problem

Most knowledge is siloed by domain. Researchers in psychology rarely cross-reference physics principles. Business strategists miss cognitive biases that explain user behavior. The connections exist — they're just invisible.

The Vision

One searchable interface for universal truths. AXIOS bridges disciplines, allowing users to see how Newton's Third Law applies to negotiation, or how the Dunning-Kruger effect shapes product adoption curves.

+ What Worked

▼

1
3-Panel UI: Resizable columns, dark/light themes, PDF export — zero framework dependencies. Pure vanilla JS keeps bundle size minimal.
2
Context Analyzer (102KB): Parses free-form text ("preparing for job interview") and returns relevant principles with relevance scoring. No external NLP libraries.
3
Multi-Agent Pipeline: 12 specialized LLM agents running in parallel, auto-merging validated entries with deduplication. Scales to 20K+ entries without manual curation.
4
Role-Based Curation: Health Systems PM view as proof-of-concept for domain-specific interfaces. Same data, different lens.
5
Production-Ready: Deployed on Vercel, mobile-responsive, 326 JSON data files. Zero server costs.

! Challenges

▼

Duplicate IDs

1,300+ duplicate IDs across files requiring cleanup. Multi-agent merging created collision issues that needed automated deduplication.

Content Gaps

Chemistry (65 entries, needs 200+), Biology, CS, and AI/ML domains underrepresented. Uneven coverage across disciplines.

Schema Inconsistencies

Type casing issues: "principle" vs "Principle" across entries. Required project-wide normalization pass.

Attribution Gaps

Missing major thinkers: Plato, Kant, Chomsky, Jung not yet attributed. Historical coverage incomplete.

→ Next Steps

▼

1
Deduplicate entries via automated ID validation on merge
2
Expand thin domains: chemistry, cellular biology, discrete math
3
Add new categories: AI/ML, cybersecurity, linguistics
4
Normalize schema casing project-wide
5
Build embeddings for true semantic search (currently keyword-based fallback)
6
Graph visualization for cross-discipline connections

</> Technology Stack

▼

Frontend

Vanilla JavaScript, CSS3. Zero framework dependencies. 3-panel resizable layout with dark/light themes.

Visualization

D3.js for data visualization. jsPDF for client-side PDF export.

Data Pipeline

Bash orchestration, Python validation scripts. 12 parallel LLM agents for content generation.

Storage

326 JSON data files. 377K lines of structured principle data. No database required.

Hosting

Vercel static deployment. Zero server costs, global CDN.

Scale

20,835 entries across 10+ disciplines. Context analyzer handles arbitrary text input.

// Entry schema { "id": "pareto-principle", "name": "Pareto Principle", "type": "principle", "domain": "economics", "description": "80% of effects come from 20% of causes", "attribution": "Vilfredo Pareto", "cross_refs": ["power-law", "zipfs-law"] }