Skip to main content

​
Contract status

Canonical interface contract The aiConnectedOS docs define personas, instances, roles, and lifecycle behavior in enough detail to serve as a service contract, even though final HTTP paths are not fixed in the repo.

​
Source documents

​
Core concepts

​
Persona

A persona is a long-lived AI entity with:
  • Stable identity
  • Role and responsibility assignment
  • Persistent memory
  • Emotional and behavioral state
  • Configurable communication style
  • Integration access inside an instance

​
Instance

An instance is the bounded workspace or business context that contains:
  • One or more personas
  • Its own data and integrations
  • Its own workflows and access boundaries
  • Shared contextual memory scoped to that environment

​
Required persona resources

ResourcePurpose
personasIdentity, role, tone, model, and lifecycle state
persona_profilesDisplay identity, wake settings, voice, and public metadata
persona_assignmentsPersona-to-instance responsibility mapping
persona_statesMood, mode, activation, and runtime state
instancesWorkspace containers for data, permissions, and integrations
instance_integrationsConnected services available to personas in an instance

​
Required persona operations

The OS docs require operations for:
  • Creating personas
  • Updating identity, role, and voice settings
  • Assigning personas to instances
  • Switching active persona context
  • Loading a persona’s persistent state before interaction
  • Exporting and importing persona definitions where appropriate

​
Persona profile fields

The developer docs and related feature docs imply these fields:
  • id
  • name
  • description
  • role
  • communication_style
  • wakeword
  • tts_voice
  • preferred_llm
  • default_tools
  • schedule
  • avatar
  • status

​
Instance rules

  • Instances are isolation boundaries.
  • Persona access is explicit, not global.
  • Data access must follow instance permissions.
  • Personas can collaborate inside an instance, but only through governed orchestration.
  • Instance context should flow into memory retrieval and downstream integrations.

​
Multi-persona collaboration

The OS docs describe collaborative personas and meeting-like coordination. That implies APIs or service operations for:
  • Enumerating personas in an instance
  • Declaring which persona is primary for a task
  • Invoking supporting personas for a shared context
  • Preserving persona-specific authorship in messages, tasks, and memory writes
{
  "persona_id": "prs_001",
  "instance_id": "inst_001",
  "name": "Orion",
  "role": "operator",
  "wakeword": "Hey Orion",
  "tts_voice": "com.apple.voice.Alex",
  "preferred_llm": {
    "provider": "openai",
    "model": "gpt-4o"
  },
  "status": "active"
}

​
Implementation note

The repo consistently treats personas as persistent team members, not disposable assistants. Model your contracts around lifecycle and continuity, not single-request chat sessions.
Last modified on April 18, 2026