Distributed Knowledge Architechture
A self-organizing knowledge graph with emergent query capabilities.
Eidolon Mesh is a local-first knowledge management system that uses parallel resonance field querying to create coherent responses from fragmented information. Unlike traditional search, which returns isolated documents, Mesh synthesizes answers by combining semantically-related knowledge nodes weighted by connection strength and usage patterns.
Core Innovation: Resonance Field Querying
Traditional knowledge bases treat documents as independent units. Eidolon Mesh treats knowledge as an interconnected field where queries activate multiple resonance points simultaneously. The system:
-
Propagates queries across semantic neighborhoods - Your question activates not just direct matches, but conceptually adjacent nodes.
-
Weights responses by coherence decay - Frequently accessed connections strengthen; unused pathways naturally fade using biologically-inspired decay
-
Synthesizes multi-node responses - Instead of returning documents, the system composes answers from the collective activation pattern.
The Quorum Effect
When enough semantically-related nodes activate simultaneously, the system exhibits emergent behavior analogous to quorum sensing in distributed systems. This "quorum effect" enables:
-
Context-aware retrieval - The graph "remembers" previous queries through usage-weighted decay.
-
Self-organizing structure - High-coherence pathways strengthen automatically; noise attenuates.
-
Organic knowledge evolution - The graph topology adapts to your actual usage patterns, not predefined categories.
Technical Architecture
Knowledge Graph: Custom graph database using coherence-scored synaptic connections between semantic nodes (neurons). Each connection has:
-
Base coherence (semantic similarity).
-
Usage-based decay coefficient.
-
Last-access timestamp for fractal forgetting.
Query Engine: Parallel field activation with configurable strategies:
-
max-coherence: Strongest semantic matches.
-
max-connectivity: Most interconnected hubs.
-
diverse-hubs: Balanced coverage across domains.
-
balanced: Coherence + connectivity weighted.
Synthesis Layer: LLM-powered composition from activated node set, maintaining source attribution and coherence scoring.
Use Cases
Research & Academia: Build a living literature review that surfaces connections between papers automatically as you query. The graph learns your research domain's structure through usage.
Technical Documentation: Create self-organizing documentation archives where frequently-accessed pathways strengthen, while outdated information naturally attenuates.
Knowledge Work: Transform fragmented notes, articles, and documents into a queryable knowledge organism that provides synthesized answers rather than search results.
Enterprise Knowledge Management: Deploy organization-wide knowledge graphs that adapt to team usage patterns without manual curation.
Key Features
-
Local-first architecture - Your data stays on your machine. Full functionality without cloud dependencies.
-
Multi-repository support - Separate knowledge domains (work, research, personal) with independent graph topologies.
-
YAML-based storage - Human-readable, git-compatible knowledge units ("proteins").
-
Gemini API integration - Leverages modern LLMs for synthesis while maintaining local graph control.
-
Usage-based memory - Fractal decay system prevents information overload while preserving frequently-accessed pathways.
-
Visual graph explorer - WebGL-powered 3D visualization of your knowledge topology.