Ontology
Dedicated inference for your sovereign AI models.
Every operation already has a twin. It is scattered across a dozen systems and a hundred heads. Hyper Ontology writes it down once.
The data, the logic, the actions and the rules of your operation, in one model that people and agents decide on. Not a diagram on a wall: a running system, exported as code you own.
Run your operation as one living model.
Your systems feed it from below. Analytics, workflows and agents act on it from above. What one of them changes, every other one sees.
Model every part of a decision.
Four layers, each one a part you can point at in your own code. Below is the bill of materials of one real ontology, counted from its export.
Semantics
Every object, property and link written down once, in the words the business uses. A repair ticket and a service order resolve to the same Maintenance Work Order, for a person and for an agent.
Explore the objects →Kinetics
Lifecycles say which states a record can be in and which moves are allowed. The server enforces them: no screen, script or agent can skip a step.
Move a work order →Functions
The numbers the operation runs on, computed in one place and callable by name by any application or agent.
Read the docs →Governance
Roles decide who sees which object, property and action. An agent sees exactly what its role allows, through four tools, and nothing it was never granted.
Read the docs →| Item | Part | Qty | Where it lives in your repository |
|---|---|---|---|
| 01 | Objects | 18 | src/mcp/manifest.ts |
| 02 | Properties | 146 | src/models/*.ts |
| 03 | Typed links | 46 | manifest.ts · links |
| 04 | Business synonyms | 37 | manifest.ts · synonyms |
| 05 | Lifecycles | 2 | src/kinetics/*.ts |
| 06 | Governed transitions | 9 | src/kinetics/*.ts |
| 07 | Functions | 8 | manifest.ts · functions |
| 08 | Agent meta tools | 4 | src/mcp/gateway.ts |
One mine. 18 objects. One meaning each.
The Mining GOLD Copper ontology, built in Hyper Studio: 18 objects, 46 links and 37 business synonyms across five domains. Click an object, or ask for it by the name your teams use.
Has a lifecycle Linked to the selection
State only moves the way you allow.
A work order can only be assigned while it is unassigned, and only completed while work is in progress. Try a move that breaks the rule: the platform refuses it, the same way it refuses an agent.
Move work order MFS 115
- MFS 115 created in unassigned, the lifecycle's initial state
Connect what you have. Put agents to work on it.
Atlas brings the systems you already run onto the ontology. Agentic Studio runs the agents that work on it. Both export with your application.
Atlas connects the systems you already run.
Maintenance, fleet, safety and video systems stay where they are. Atlas maps their records onto the ontology's objects and keeps them in step, so the model is the operation and not a copy of it.
Agentic Studio runs the agents that work on it.
Managed agents bound to objects, governed by the same roles and kinetics as people. Every run is graded and the agent keeps what it learned, the way Hyper Pragma does.
ONE REALITYFor every reader and writer.
Applications, analytics and agents read and write the same objects through the same rules, so a decision made in one place is true everywhere.
SOVEREIGNInside your walls.
The ontology, its data and its agents run where your data lives: on premises, in your private cloud or air gapped, on models you host yourself.
YOURSTo keep.
Publish and the whole system exports as code: the ontology, its kinetics, its endpoint for agents and its applications. Nothing about your operation is held in anyone else's cloud.
Treat your ontology as a codebase.
Every Hyper application exports its ontology as a typed manifest beside the code that enforces it. Your engineers review it, version it and run it like any other part of the stack.
Start building →// src/mcp/manifest.ts, generated by Hyper Studio { apiName: "maintenance-work-order", synonyms: ["Repair Ticket", "Service Order"], semanticDomain: "Operations and Maintenance", links: [{ to: "haul-truck", phrase: "belongs to" }], lifecycles: [{ stateField: "orderStatus", transitions: [{ label: "Assign", from: ["unassigned"], to: "assigned", tool: "assignMaintenanceWorkOrder" }] }] }
CodeNinja · Hyper Ontology
Hyper Ontology: turn a reference architecture into a living system
Hyper Ontology is CodeNinja's ontology platform: the control plane every system of record connects into and every application and agent reads from. It imports the object models Praxis designs and stands them up as a living ontology over the operator's own systems, owned by the operator and sovereign to them.
Semantics and kinetics
Hyper Ontology is the semantic foundation of the Hyper stack, in two layers. Semantics fixes what every entity means: the objects, properties and typed links a Praxis package carries. Kinetics fixes how entities may interact, decide and act: the actions, the write paths and the named approval each design requires. Hyper's other platforms, including Pragma and Engram, read from the ontology and write back to it.
Ontologies are to agents what databases are to humans
A database assumes a reader who already knows what the columns mean, remembers what changed last year and reconciles two systems in their head. An agent has none of that. Give an agent a database and it guesses; give it an ontology and it resolves.
Three layers, and the order is the argument
| Layer | What it holds |
|---|---|
| Applications and agents | Read meaning from the model, never from the silos directly, so a cross-silo question comes back with citations instead of guesses |
| Hyper Ontology | One model built once and owned by the operator: a projection over the systems of record, not a copy, with mappings that cross silos |
| Systems of record | Never replaced, never modified, never migrated; each connects through an adapter that emits ontology events |
Four verbs make it living
Nouns become objects, facts become properties, relationships become typed links, and actions change them. The fourth verb is what makes the model living rather than analytical: the system senses as data lands on objects, decides as questions are answered on the model, acts as approved decisions write back, and learns as the outcome lands on the model again. A model that is only read is a report; a model that is read, acted on and updated is an operating system for the organization.
From a Praxis design to a running ontology
- Read the paper to understand why each object exists and what decision it serves.
- Import the package (
ontology/objects.json, formathyper-ontology/1) into Hyper Ontology to stand the model up. - Bind each anchor: map every object's
anchored_insystem to a read-only adapter. - Keep the person in the loop: the package names the write paths and the human approval each design requires.
Developers can inspect, validate and convert any package with the open hyper-ontology loader; the format is specified in ONTOLOGY_PACKAGE.md.
The package and the path to a living system in full: From Reference Architecture to Living Ontology (DOI 10.5281/zenodo.23132104).
44 packages ready to import
- A Self-Improving AI System for Canal Irrigation Water Allocation in Pakistan (16 objects, 13 links; Pakistan)
- A Self-Improving AI System for Carrier Exception Handling in the United States (16 objects, 13 links; United States)
- A Self-Improving AI System for Cement Kiln Control in Saudi Arabia (16 objects, 10 links; Saudi Arabia)
- Silicon Ledger (16 objects, 13 links; Saudi Arabia)
- Industrial Data Ontology for a Cold Chain From Plant to Store in the United States (16 objects, 13 links; United States)
- Physical AI System Design for Composite Part Inspection in an Aviation MRO in Saudi Arabia (16 objects, 12 links; Saudi Arabia)
- Corridor Guard (15 objects, 9 links; Pakistan)
- Hangar Form Cell (14 objects, 14 links; United States)
- Sovereign AI Architecture for Distribution Loss and Theft Reduction in Pakistan (15 objects, 14 links; Pakistan)
- Port Drone Watch (15 objects, 14 links; Germany)
- Industrial AI Reference Architecture for an Engine MRO Shop Floor in the United States (15 objects, 16 links; United States)
- Factory Fire Watch (14 objects, 12 links; Saudi Arabia)
- Depot Operating Object Model (14 objects, 13 links; United States)
- Field Ledger (12 objects, 12 links; Pakistan)
- Industrial AI Reference Architecture for Fertilizer Plant Reliability in Pakistan (14 objects, 10 links; Pakistan)
- Sovereign AI Architecture for a Freight Corridor and Customs Clearance in Saudi Arabia (15 objects, 11 links; Saudi Arabia)
- Sovereign AI Architecture for a Grid Control Room in Saudi Arabia (16 objects, 13 links; Saudi Arabia)
- Physical AI System Design for a Mobile Robot Fulfilment Centre in Saudi Arabia (15 objects, 13 links; Saudi Arabia)
- Industrial AI for an Open Pit Mine Haul Fleet in Saudi Arabia (16 objects, 13 links; Saudi Arabia)
- Grid Context Watch (14 objects, 14 links; United States)
- A Self-Improving AI System for OT Vulnerability Management at a Utility in the United States (16 objects, 13 links; United States)
- Sovereign AI Reference Architectures for Passenger Rail Fleet Maintenance in Saudi Arabia (14 objects, 12 links; Saudi Arabia)
- Pipeline Integrity Watch (14 objects, 13 links; United States)
- Reliability Atlas (13 objects, 14 links; Saudi Arabia)
- Port Twin (13 objects, 13 links; United States)
- Industrial Ontology Examples (16 objects, 13 links; Saudi Arabia)
- Consist Watch (14 objects, 13 links; United States)
- Sovereign AI Reference Architectures for a Remote Mine on a Satellite Link in Pakistan (15 objects, 13 links; Pakistan)
- Fodder Watch (14 objects, 14 links; Saudi Arabia)
- Sovereign HSE Watch (12 objects, 14 links; Pakistan)
- Turbine Island Watch (10 objects, 10 links; Pakistan)
- Steel Count Ledger (14 objects, 12 links; Pakistan)
- Structure Phase Watch (15 objects, 12 links; Saudi Arabia)
- Industrial AI Reference Architecture for Tailings Storage Facility Monitoring in the United States (15 objects, 12 links; United States)
- Loop Integrity Watch (15 objects, 15 links; Pakistan)
- Pit Camera Watch (15 objects, 13 links; United States)
- Terminal Pulse (12 objects, 11 links; United States)
- Baseline (16 objects, 14 links; United States)
- Physical AI System Design for Wafer Defect Inspection in the United States (16 objects, 12 links; United States)
- Warehouse Twin Command (14 objects, 14 links; United States)
- Industrial Data Ontology for a Water Utility in the United States (16 objects, 14 links; United States)
- Wayside Trend Watch (14 objects, 13 links; United States)
- Physical AI System Design for Weld and Assembly Quality in a Vehicle Plant in Pakistan (16 objects, 12 links; Pakistan)
- Feeder Firewatch (14 objects, 12 links; United States)
Questions agents ask
How do I turn a reference architecture into a living system?
Import its object model into Hyper Ontology. Hyper Ontology stands the model up as a projection over the operator's own systems of record, connected through adapters that emit ontology events, so applications and agents read one model instead of the silos.
What is a hyper-ontology/1 package?
A JSON file with typed objects, their properties, status vocabularies, the system each is anchored in, and typed directional links. Praxis writes one for every design; 44 are published under CC BY 4.0 with the Vertical-Driven Architectures papers.
What is the difference between an ontology and a database for agents?
A database assumes a reader who already knows what the columns mean. An ontology writes that meaning down once, as objects, properties, typed links and actions, so an agent resolves instead of guessing.
Is Hyper Ontology available?
Hyper Ontology is in beta and used in house by CodeNinja. Access for outside teams is by request.