SYSTEM DESIGN

Knowledge Graph Design Brief

Executive Summary

This document proposes a semantic knowledge graph architecture to store the relationships between all content, concepts, decisions, and operations in the Kai Hamil ecosystem. The graph enables RAG (Retrieval-Augmented Generation) constrained to Kyle's actual thought patterns, replicating his mindset in agent form.

Core Philosophy: Every entity is connected. Abstractions (frameworks) link to implementations (products). Decisions link to outcomes. Content links to the thinking that produced it. Transparency is achieved through auditable relationship chains.

Foundational Design Principle:

Systems serve presence. The infrastructure we build — automations, agents, workflows, knowledge graphs — exists to create space for physical presence with other humans. Digital scaled communication, record-keeping, content production: these should happen without consuming the attention they enable.

— Does this system increase or decrease Kyle's capacity for presence?

Entity Types (Nodes)

Content Layer

Entity Description Examples
PostPublished essays"Physics of Love", "5 Filters"
ResearchRaw notes, draftsFuture of Work, Open Source Society
FrameworkThinking tools5 Filters, System // Self, 5:1 Ratio
BriefQuick insightsMorning briefs, session summaries
IdeaCaptured conceptsDental lollipop, meal planning

Operations Layer

Entity Description Examples
DecisionDocumented choiceMac Mini, OpenClaw adoption
TaskKanban-tracked workDeploy #14, V1 baseline testing
MetricMeasurable dataToken usage, deploy frequency
AgentAutomated systemStitch, orchestrator, executor
DeployPublished changeDeploy #14 (research HTML)

Relationship Types (Edges)

Implementation Relationships

Causal Relationships

Use Cases

Implementation Phases

  1. Bootstrap: Define entity types, create initial triples from existing content
  2. Populate: Backfill all posts, link decisions to outcomes, map concepts
  3. Query: SPARQL endpoint, RAG integration, concept visualization
  4. Evolve: ML patterns, predictive recommendations, automated summarization

Success Criteria