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 |
|---|---|---|
| Post | Published essays | "Physics of Love", "5 Filters" |
| Research | Raw notes, drafts | Future of Work, Open Source Society |
| Framework | Thinking tools | 5 Filters, System // Self, 5:1 Ratio |
| Brief | Quick insights | Morning briefs, session summaries |
| Idea | Captured concepts | Dental lollipop, meal planning |
Operations Layer
| Entity | Description | Examples |
|---|---|---|
| Decision | Documented choice | Mac Mini, OpenClaw adoption |
| Task | Kanban-tracked work | Deploy #14, V1 baseline testing |
| Metric | Measurable data | Token usage, deploy frequency |
| Agent | Automated system | Stitch, orchestrator, executor |
| Deploy | Published change | Deploy #14 (research HTML) |
Relationship Types (Edges)
Implementation Relationships
- implements — Product implements Framework
- builds_on — Post builds_on Post
- deploys — Deploy deploys Content
- documents — Post documents Decision
- captures — Memory captures Session
Causal Relationships
- caused_by — Outcome caused_by Decision
- blocks — Friction blocks Goal
- resolves — Framework resolves Friction
- enables — Framework enables Goal
Use Cases
- RAG-Enabled Responses: "What did I decide about X?" — trace from question to documented decision
- Framework Recommendation: Map friction to proven solutions
- Content Discovery: Follow idea evolution from origin to current form
- Progress Tracking: Goal → Tasks → Blockers → Next steps
- Pattern Recognition: Identify recurring frictions, successful frameworks
Implementation Phases
- Bootstrap: Define entity types, create initial triples from existing content
- Populate: Backfill all posts, link decisions to outcomes, map concepts
- Query: SPARQL endpoint, RAG integration, concept visualization
- Evolve: ML patterns, predictive recommendations, automated summarization
Success Criteria
- Can answer "What did Kyle decide about X?" from graph alone
- Can trace idea evolution (origin → development → current form)
- Can identify patterns (recurring frictions, successful frameworks)
- Supports RAG with high relevance and accuracy
- Is auditable — any claim can be traced to source