Key Concepts

OACIS treats the enterprise itself as the system under management — with version control, observability, and continuous integration applied to organizational knowledge.

The Six Principles

Every design decision in OACIS traces back to these foundational principles.

Free Format Truth

All knowledge is stored in free, open, non-proprietary formats — Markdown, RDF/OWL, Apache Parquet, Open API specs, standard SQL. If the format requires a license to read, the knowledge inside is not truly yours. 100% reconstitutability: if the system disappears tomorrow, the knowledge survives intact.

Universal Accessibility

All organizational information is accessible through standard pipelines. No information islands. No shadow IT dependencies. No tribal knowledge locked in one person's head. One graph, every source. If a system exists in the enterprise, a pipeline reaches it.

Semantic Understanding

All information is understood through knowledge graphs, ontological mapping, and semantic retrieval — not just stored and searched. The system doesn't merely hold data; it comprehends relationships, context, and meaning. Storage without understanding is a warehouse, not intelligence.

Pipeline Compilation

All data pipelines compile — they validate, cross-link, and check for inconsistencies before committing. Malformed data is quarantined, not silently ingested. Broken data never enters the graph. If you can't replay it, you can't trust it.

Tracked Predictions

All predictions and recommendations are tracked against actual outcomes. The system learns from the delta between what it predicted and what happened. Every forecast, every recommendation, every AI-generated insight carries a scorecard. No unaccountable predictions.

Ontological Mapping

All knowledge maps to a formal ontology — IEEE SUMO at the top, ISO 21838 (BFO) for structure, domain ontologies (ACORD, HL7 FHIR, ISO 20022, CDM) in the middle, and organization-specific terms at the bottom. Every term means one thing. When the knowledge graph says "policy," it knows exactly what kind.

The Seven-Level Trust Hierarchy

Not all knowledge is created equal. OACIS assigns every assertion a trust level — from natural law at the top to AI-generated inferences at the bottom. Higher-level knowledge always wins in conflicts.

1
Natural Law
Mathematics, physics, logic, fundamental constants. These don't change because a legislature convenes or a board votes. Immutable.
2
Constitutional & Foundational Law
Federal and state constitutions, corporate charter and bylaws. The foundational documents that define the entity's existence and authority.
3
Statute & Regulation
Enacted law and regulatory requirements — federal statutes, state insurance codes, NAIC model laws. Authoritative within their jurisdiction. Time-bound and versioned.
4
Industry Standards & Administrative Rules
NAIC bulletins, ACORD standards, administrative implementing guidance. Operationalizes the regulations above — binding within the industry but changeable by administrative action.
5
Corporate Policy
SOPs, policy manuals, board decisions, organizational interpretation. How the company chooses to implement higher-level requirements. Must not contradict levels above.
6
Observed Fact
System records, call logs, transaction data, monitoring metrics. Machine-generated from instrumented, trusted sources. Timestamped and immutable — but only as reliable as the system that produced them.
7
AI-Generated
LLM outputs, statistical inferences, pattern-matched results. Useful for discovery and hypothesis generation — never for decisions. Must be promoted through human validation before entering operational knowledge.

Architecture

A single database engine, extended to serve every storage mode the enterprise requires.

OACIS doesn't require a constellation of specialized databases. It runs on PostgreSQL — the world's most advanced open-source relational database — extended with purpose-built modules that transform it into a multi-modal knowledge platform:

Apache AGE

Graph queries on relational data. Property graph traversals using openCypher syntax, stored in PostgreSQL tables. The knowledge graph lives here — entities, relationships, provenance, trust levels.

pgvector

Vector similarity search. Embedding-based retrieval for semantic search, document similarity, and RAG pipelines. Nearest-neighbor queries on high-dimensional vectors without leaving PostgreSQL.

TimescaleDB

Time-series at scale. Hypertables with automatic partitioning for operational telemetry, contract timelines, regulatory change tracking, and temporal knowledge assertions.

pg_lake

Data lake integration. Query Apache Iceberg tables, Parquet files, CSV, and JSON directly from PostgreSQL. Powered by DuckDB's columnar execution engine. Cold storage, hot queries.

Citus

Horizontal scale-out. Distributed PostgreSQL for multi-tenant deployments and large-scale analytical workloads. Shard the knowledge graph across nodes without changing a query.

PostGIS

Geospatial intelligence. Location-aware queries for facility management, regional compliance mapping, service territory analysis, and geographic knowledge graph traversals.

The Ontology Stack

OACIS resolves semantic ambiguity through a layered ontology — from universal concepts down to organization-specific terminology.

IEEE SUMO UPPER ONTOLOGY

Suggested Upper Merged Ontology — 25,000+ terms covering all domains of human knowledge. The conceptual ceiling that ensures every lower ontology speaks the same language.

ISO 21838 (BFO) TOP-LEVEL BRIDGE

Basic Formal Ontology — the ISO standard for top-level ontology structure. Bridges SUMO to domain-specific ontologies with a formal, minimal upper layer.

Domain Ontologies INDUSTRY LAYER

ACORD (insurance), HL7 FHIR (healthcare), ISO 20022 (financial messaging), CDM (common domain model). Industry-standard vocabularies mapped to the upper ontology.

Organization Taxonomy LOCAL LAYER

Your organization's specific terms, roles, processes, and entities — mapped upward through the domain and upper ontologies. When your company says "case," the graph knows exactly which kind.

The Conveyor

Data doesn't just flow into the knowledge graph. It's compiled — through eight named components.

Traditional ETL moves data between systems. The OACIS Conveyor compiles raw information into structured knowledge — capturing intent, state, and behavior, then translating, cross-linking, validating, committing, and verifying. Every source, one pipeline, one graph.

1
CodeTap — Captures what someone wrote that a system should become. Terraform, CX-as-Code, Helm charts, CI/CD configs. Git-native: every change is a versioned declaration of intent.
2
ConfigTap — Discovers what a system is. Queues, skills, routing rules, IVR flows, schedules. Polls live systems, compares snapshots, detects drift between intent and reality.
3
DataTap — Captures what a system does. Interactions, call records, agent states, queue metrics. Streams continuously with batch backfill for historical data.
4
Translator — Maps raw data to the ontology's canonical concepts. Vendor-specific terms become universal vocabulary. "Genesys queue" and "Cisco skill group" both become the same ontology concept.
5
Weaver — Cross-source relationship discovery. Connects the same customer across divisions, links policies to systems, surfaces dependencies that no single source reveals.
6
Checker — Pre-ingestion validation and trust hierarchy enforcement. Verifies records for conflicts, suspicious patterns, and policy violations before they enter the graph.
7
Keeper — Writes validated data to the graph with full provenance metadata. Analogous to git commit — every assertion gets an author, timestamp, source, trust level, and confidentiality tag.
8
Tester — Closed-loop behavioral verification. Independently verifies that systems do what the graph says they should do. Analogous to pytest — the last line of defense.

Go Deeper

The full OACIS framework spans 24 chapters across six parts. Read the book, explore the blog, or bring us in to assess your organization.

Read the Blog Consulting Contact Us