Skip to main content

​
Overview

Entities represent named entities — people, companies, products, places — automatically extracted from your project’s content. Each Entity is a uniquely identified record that connects related information across all your data types. For why structured entity management matters, see the concept overview.

​
Fields

FieldTypeDescription
idstringUnique identifier
namestringPrimary entity name
typeEntityTypeEntity type classification
aliasesstring[]Alternative names / spellings
metadataobjectAdditional metadata
createdAtstringCreation timestamp (ISO 8601)
updatedAtstringLast update timestamp (ISO 8601)

​
Entity Types

TypeDescriptionExamples
personPeopleJohn Smith, Taro Yamada
companyCompanies / OrganizationsGoogle, Acme Corp
productProducts / ServicesiPhone, Slack
placePlacesTokyo, Silicon Valley
otherOthersProject names, etc.

​
Entity Relations

Entities can have typed relationships with each other. The full list of relation types:
RelationDescriptionExample
CEO_OFCEO of ~John Smith -> Acme Corp
CTO_OFCTO of ~Jane Doe -> Acme Corp
CFO_OFCFO of ~Bob -> Acme Corp
FOUNDER_OFFounder of ~John Smith -> Acme Corp
MEMBER_OFMember of ~Alice -> Engineering Team
WORKS_ATWorks at ~Alice -> Acme Corp
PARENT_COMPANY_OFParent company of ~XYZ Holdings -> Acme Corp
SUBSIDIARY_OFSubsidiary of ~Acme Corp -> XYZ Holdings
LOCATED_INLocated in ~Acme Corp -> New York
PARTNER_OFPartner of ~Acme Corp -> Beta Inc
COMPETITOR_OFCompetitor of ~Acme Corp -> Rival Inc
ACQUIRED_BYAcquired by ~Startup -> Big Corp
INVESTED_INInvested in ~VC Fund -> Startup

​
Aliases

Multiple names for the same Entity are managed as aliases. Aliases are learned automatically and can also be added manually. 
Entity: John Smith
├── Alias: John
├── Alias: Mr. Smith
└── Alias: J. Smith
All aliases are matched during search, so searching “Mr. Smith” finds the same Entity as “John Smith”.

​
Unresolved Mentions

When automatic extraction has low confidence, items are held as unresolved mentions instead of being immediately linked. You can resolve them by choosing one of the following options:
  • Link to an existing Entity — associate the mention with an Entity that already exists
  • Create as a new Entity — register it as a brand-new Entity
  • Ignore — dismiss the mention
This prevents incorrect automatic matching while maintaining data quality.

​
Duplicate Detection and Merging

Similar Entities are automatically detected (e.g., “Acme Corp” and “Acme Corporation”). When two Entities are merged:
  • All Knowledge links are migrated
  • All Episode links are migrated
  • All relations are preserved
  • Aliases are combined

Entity Concept

Concept overview and use cases

Entities API

Manage entities with the SDK