Knowledge Architecture & Ontology

What It Is

The capability to design, build, and operate knowledge systems that encode organizational truth as computable structure — not document stores, but searchable knowledge surfaces where entities have typed relationships, formal invariants govern validity, and mathematical discovery operations surface what must exist but hasn't been cataloged. This spans four layers: ontology design (what can exist), entity graph architecture (what does exist), domain discovery methodology (what should exist), and research knowledge management (what will exist).

Why It Matters

Storing information and encoding truth produce different outcomes. One compounds intelligence. The other accumulates noise.

Most enterprise knowledge management stores documents. Documents are opaque to machines, degrade through staleness, and resist composition. Search finds documents. It does not navigate structure, validate relationships, or discover gaps. The alternative is computable structure: entities with typed relationships, formal ontologies with invariant validation, and discovery methods that surface what must exist but hasn't been found.

Four production systems address the four layers. The Practice OS entity graph holds 253 entities across 13 types with 248 typed links, full-text search via FTS5, and 38+ MCP tools for programmatic access — not a document store but a knowledge surface accessible to both humans and agents, with 112 tests validating graph integrity. The BrigitteOS ontology defines atomic capabilities as JSON-LD (capability.jsonld), making them machine-readable, composable, and interoperable with linked data standards. The Simiya methodology discovers new entities through 10 mathematically grounded operations — FCA lattice supremum, Galois connection closure, Mendeleev gap prediction — implemented as a swarm architecture (Haiku specialists + Sonnet synthesis) at 4-8x cost reduction. The Paper Factory manages 29+ research papers through a 7-phase pipeline with parallel subagent architecture for literature review and review dimensions.

Each component addresses a different layer: ontology defines what can exist, the graph tracks what does exist, discovery finds what should exist, and the pipeline creates what will exist. The layers compose — discovered entities feed the graph, the graph informs ontology evolution, ontology constraints validate discovery, and research creates the formal foundations for all three.

Proof Points

Practice OS entity graph: 253 entities, 248 typed links, 13 entity types (capability, competency, utility, product, client, paper, framework, spec, reference, project, topic, surface), 13 link types, FTS5 full-text search, 38+ MCP tools, 112 tests passing, dual transport (stdio + HTTP)
BrigitteOS ontology: Atomic capability ontology in JSON-LD (capability.jsonld) — machine-readable, linked data compliant, composable with external ontologies
Simiya Platform: 73 tests passing, 5 Canvas visualizations, client handoff ZIP export, deployed at simiya-platform.vercel.app — sole production implementation of the 10x10 methodology
SimiyaID: 26 MCP tools, 113 tests, Being/Doing/Becoming enrichments, materialization pipeline, slug utilities — the domain identification MCP server
Simiya Discovery Hub: 9 MCP tools, 10 adapters (DNS, WHOIS, web scraping, social, technology detection), 78 tests — multi-source connected discovery
Paper Factory: 29+ papers, 7-phase pipeline (intake → thesis → literature review → outline → writing guide → draft → review), 6 parallel review dimensions, swarm architecture for literature review
55 skills: Claude Code extensions codifying domain knowledge as reusable utilities, searchable by the Skill Library Index
Mathematical foundations: FCA (Ganter & Wille), UFO (Guizzardi), Category Theory databases (Spivak), autopoiesis (Maturana & Varela), Ashby’s requisite variety, Thagard coherence
Patent: USPTO 19/418,922

Market Position and IP

Patent-protected methodology (USPTO 19/418,922). The combination of formal ontology, computable entity graph, mathematical discovery operations, and research pipeline operating as an integrated system is unique in the market. No competing practice or product operates across all four knowledge layers.

Enterprise knowledge management vendors (Confluence, Notion, SharePoint) provide document stores with search. Enterprise architecture tools (TOGAF, Zachman implementations) provide static taxonomies with manual curation. Ontology platforms (TopBraid, Protégé) provide formal ontology editing without discovery methods. Research management tools (Zotero, Mendeley) manage citations without production pipelines. Each addresses one layer. None addresses the generative discovery layer — the Becoming layer where the 10x10 methodology operates.

Production evidence: three deployed Simiya products (Platform, SimiyaID, Discovery Hub) with 264 combined tests, Practice OS running live with 38+ MCP tools and 112 tests, Paper Factory with 29+ papers in pipeline. The competitive moat is the mathematical discovery methodology — the 10 operations grounded in FCA, Galois connections, and category theory require original mathematical work, not engineering effort, to replicate.

Novel Research Contribution

The Being/Doing/Becoming ontological framework provides the first formalized generative layer for organizational ontology. Porter (1985) gives the value chain structure. APQC gives the process classification. Guizzardi (UFO/OntoUML) gives foundational semantics. None provides a discovery method for the Becoming layer — what entities, capabilities, and structures must emerge from the intersection of Being and Doing.

The 10 discovery operations and 10 validation invariants are novel contributions. Each operation is grounded in established mathematics — FCA lattice supremum (Ganter & Wille), Galois connection closure, natural transformations, community detection, autopoietic closure (Maturana & Varela) — but their composition into a systematic discovery methodology for organizational ontology has no precedent. The invariants (compression, sovereignty, emergence, regeneration, anticipation, revelation, prediction, coherence, resonance, recursion) provide formal validation criteria that distinguish genuine ontological discovery from confirmation bias.

Target venue: ISR (Information Systems Research) or MIS Quarterly. No established literature exists at the junction of formal ontology, enterprise architecture, and mathematical discovery methods.

Implementation and Impact

Clients receive a domain identification engagement that produces a complete Being/Doing/Becoming ontology of their organization. The three-phase pipeline — Frame (align to Porter + APQC scaffolding), Triplets (extract Being/Doing/Becoming triples using AI swarm), Magic (10 parallel specialist agents, each running one discovery operation) — runs in days, not the months typical of enterprise architecture assessments.

The deliverable includes: entity graph with typed relationships mapping the full organizational ontology, gap analysis (Mendeleev predictions — structural positions where entities must exist but haven't been identified), materialization templates for converting discovered entities into operational form, and a periodic table of domain entities visualizing the complete ontological structure.

For organizations that have invested in EA assessments and received inventories of what exists, this engagement reveals what their inventories missed — not by finding errors in what was mapped, but by discovering what was never in the mapping's scope. Engagement model: 1-2 week domain identification using the Simiya Platform, with client handoff ZIP export containing the complete model. Measurable outcome: organizational intelligence that reveals structural truth invisible to inventory-based methods.

Connections

Builds: Practice OS, Simiya Platform, SimiyaID, Simiya Discovery Hub, BrigitteOS, Paper Factory
Frameworks: Simiya Methodology, Content Engine Pipeline
Papers: being-doing-becoming
Imperatives: 10x10 Domain Intelligence, Fractal Design
Capabilities: Agentic System of Systems