Appearance
Data Models
Iris uses Eloquent models to manage persistent data. Models are in app/Models/.
Overview
| Model | Purpose |
|---|---|
User | User accounts and integration settings |
Memory | Persistent semantic memories with vector embeddings |
Conversation | Chat messages (user and assistant) |
ConversationSummary | Compressed conversation history |
TokenUsage | API usage tracking |
Attachment | Uploaded and generated files |
Memory
The core data model for persistent memories.
| Field | Description |
|---|---|
content | Memory content |
memory_type | fact, preference, goal, event, skill, relationship, habit, context |
importance | Priority score (0.0-1.0) |
embedding | 1536-dimensional vector for semantic search |
tags | Organization tags |
category | personal, professional, hobbies, health, etc. |
ConversationSummary
Rich summaries of conversation history.
| Field | Description |
|---|---|
summary | Narrative summary text |
emotional_markers | Key emotional moments |
evolving_themes | Recurring topics |
unresolved_threads | Open discussion topics |
relationship_dynamics | Trust, rapport tracking |
Database
Uses PostgreSQL with pgvector extension for efficient semantic search on memory embeddings.