Skip to main content
The Ontology is a structured model of your organization’s real-world entities and the relationships between them. Rather than working with disconnected files and databases, the Ontology lets you define what things exist in your world — deals, clients, candidates, companies — and how they relate to each other. Once defined, Athena’s agents can query, create, and update ontology data directly within conversations, turning unstructured workflows into structured, actionable knowledge.

​
Why Use an Ontology?

Most organizations store critical information across dozens of tools — CRMs, spreadsheets, documents, emails. The Ontology brings this scattered data into a single, coherent model that both humans and AI agents can work with.

Structured Knowledge

Define the entities and relationships that matter to your business with typed properties and validation rules.

AI-Native

Athena agents can read, create, and update ontology data in conversations — no manual data entry required.

Connected Data

Link related entities together to capture how your data connects, enabling richer queries and insights.

​
Core Concepts

The Ontology is built from three primitives. If you’re familiar with databases, the analogy is straightforward:
Ontology ConceptDatabase EquivalentDescription
Object TypeTableA template defining a category of real-world entity (e.g., “Deal”, “Client”)
InstanceRowA single record of an object type (e.g., the deal “Acme Series B”)
RelationshipForeign key / JoinA named, typed link between two object types (e.g., Deal → assigned_to → Person)
Each object type has properties (columns) with types, validation rules, and descriptions. One required text property is designated the title property — this is the display name shown in tabs, search results, and mentions.

​
Object Types

An object type is the schema definition for a category of entity. For example, a “Company” object type might have properties like name, industry, employee_count, and annual_revenue. Every instance of the “Company” type will have these same properties.
Athena provides built-in templates for common object types like Person, Company, Deal, Client, and Recruit to get you started quickly.

​
Properties

Properties define the attributes of an object type. Each property has a name, a type, and optional constraints:
TypeDescriptionExample
TextFree-form stringName, email, notes
NumberDecimal valueRevenue, price, latitude
IntegerWhole numberEmployee count, quantity
BooleanTrue/falseIs active, is verified
DateCalendar dateStart date, founded year
EnumFixed set of optionsStage, status, tier
Properties can be marked as required, given default values, and constrained with rules like minimum/maximum values or string patterns.

​
Relationships

Relationships define how object types connect to each other. Each relationship has:
  • Source and target object types
  • A name describing the relationship (e.g., “assigned_to”, “works_at”, “owns”)
  • Cardinality: one-to-one, one-to-many, many-to-one, or many-to-many
  • Directionality: whether the relationship goes one way or both
For example, a “Deal” might have a many-to-one relationship called “owned_by” linking it to a “Person”, meaning each deal has one owner but each person can own many deals.

​
How Agents Use the Ontology

Once you’ve defined your ontology, Athena agents can work with it directly:
  • Create instances — “Log a new deal: Acme Series B, $5M, closing next quarter”
  • Query data — “Show me all deals in the negotiation stage worth over $1M”
  • Update records — “Move the Acme deal to closed-won and update the value to $5.5M”
  • Traverse relationships — “Who owns the deals connected to Acme Corp?”
The ontology gives agents the structured context they need to take precise, data-driven actions instead of relying solely on unstructured conversation.

​
Getting Started

1

Open the Ontology Builder

Navigate to Experimental → Ontology Builder from the sidebar. This is the visual workspace for defining your data model.
2

Create your first object type

Click New Object Type or double-click the canvas. Choose a template or start from scratch. Define properties and set a title property.
3

Add relationships

Drag from the handle on one object type node to another to create a relationship. Configure the name, cardinality, and direction.
4

Start using it

Once your ontology is defined, agents with the ontology tools enabled can create and query instances in any conversation.
Read on to learn about Object Types, Relationships, and the Ontology Builder in detail.