Objects and relations
Objects (nodes)
An object is a type of element of your domain (e.g. BRAND, PART, CONTRACT). Each object carries attributes: the information stored for each of its instances.
Create an object:
- Click + in the Objects section of the side list
- Give it a name and a description
- Add its attributes in the detail panel
Example: the CONTRACT object
| Attribute | Type |
|---|---|
unit_price | FLOAT |
currency | STRING |
annual_volume | INT |
starting_date | DATE |
For each object and each attribute, you can set a sensitivity level (e.g. C1).
You can also change an object’s icon, to recognize it at a glance on the diagram (e.g. a document for CONTRACT, a silhouette for SUPPLIER).

Relations (edges)
A relation connects two objects and describes their link (e.g. PART -[purchased_via]-> CONTRACT). Like an object, a relation can carry attributes.
Create a relation: connect two objects directly on the diagram, or click + in the Relations section of the side list.
Example:
BRAND -[produces]-> MODELMODEL -[based_on]-> PLATFORMMODEL -[contains]-> ASSEMBLYASSEMBLY -[consists_of]-> PARTPART -[purchased_via]-> CONTRACTCONTRACT -[negotiated_with]-> SUPPLIERPART -[equivalent_to]-> PART
Semantic layer
Available in Basic and Expert modes.
Each object and each relation can be enriched to help the AI assistant understand it properly:
- Short description (objects only) — a short sentence that sums up the object
- Long description — what the element represents in your business
- Synonyms — the other words users use
- Business rules — the rules to know to interpret the data correctly
Example: the PART object
Short description: Vehicle component.Description: A physical component mounted on a vehicle, identified by its reference.Synonyms: component, piece, item, spare partBusiness rules:- Two parts with different references can be functionally identical; they are then linked by equivalent_to.- The price of a part is never stored on the part itself: read unit_price on its CONTRACT (PART -[purchased_via]-> CONTRACT).The more the ontology is enriched, the better the quality score (see Semantic).
