Graph Memory

Knowledge Graph-based Memory System

Graph Memory manages entities and their relationships in a graph structure. Using Neo4j, it builds complex knowledge networks and retrieves information with advanced graph queries.

How to use @graph

Experience the Graph Memory visualization with real-time interaction. Drag nodes, zoom, and switch between color themes.

Loading interactive demo...

Key Features

Entity Management

Manage people, projects, concepts, tasks as structured nodes

concept, person, project, task, goal, user

Relationship Tracking

Dynamically track relationships between entities and automatically build knowledge graphs

Entity Merging

Consolidate duplicate entities and manage them as aliases

Auto Extraction

Automatically extract entities and relationships from conversations to build graphs

Graph Operations

Entity Operations

  • Add Entity: Add new entities to the graph
  • Update Entity: Update existing entity information
  • Delete Entity: Remove unnecessary entities
  • Merge Entities: Consolidate duplicate entities

Relationship Operations

  • Add Relationship: Add new relationships between entities
  • Update Relationship: Update relationship strength or type
  • Delete Relationship: Remove unnecessary relationships

Query Operations

  • Get Graph: Retrieve complete graph structure for user
  • Search Entities: Search entities with semantic search
  • Auto-extracted Graph: Get auto-extracted graph

Color Themes

Graph Memory provides three color themes to visually represent node importance. Colors are automatically assigned based on connection count following a power law distribution.

Earth Theme

Recommended

Natural, calm colors. Easy on the eyes for extended work sessions.

Hub
0-5%
High
5-20%
Medium
20-50%
Low
50-90%

Professional Theme

Modern and clean color scheme. Suitable for business and analytics use cases.

Hub
0-5%
High
5-20%
Medium
20-50%
Low
50-90%

Casual Theme

Bright and friendly colors. Perfect for casual and creative use cases.

Hub
0-5%
High
5-20%
Medium
20-50%
Low
50-90%

Power Law Distribution

Color coding based on real-world graph structures. Effectively visualizes few hub nodes and many peripheral nodes.

Top 5%
Hub
Top 20%
High
Top 50%
Medium
Top 90%
Low
Bottom 10%
Isolated

© 2026 Duzzle Inc. All rights reserved.