Magneteco
Getting Started

Getting Started

Get started with Magneteco memory infrastructure

Getting Started

Magneteco provides persistent memory infrastructure for AI agents. This guide will help you integrate Magneteco into your application.

Overview

Magneteco works by:

  1. Ingesting content - Conversations and events are sent to Magneteco
  2. Extracting facts - LLM extracts atomic facts and entities
  3. Building knowledge - Facts are stored with embeddings and relationships
  4. Retrieving context - Query memory to get relevant context for your agent

Memory Layers

LayerPurposeStorage
ResourcesRaw immutable logsS3
ItemsAtomic extracted factsPostgreSQL + pgvector
CategoriesEvolving summariesPostgreSQL
GraphEntity relationshipsNeo4j

Architecture Overview

┌─────────────────────────────────────────────────────────────────┐
│                        YOUR APPS                                 │
├─────────────┬─────────────┬─────────────┬───────────────────────┤
│   App A     │   App B     │   App C     │   Future Apps         │
└──────┬──────┴──────┬──────┴──────┬──────┴─────────┬─────────────┘
       │             │             │                │
       └─────────────┴─────────────┴────────────────┘


              ┌─────────────────────────────┐
              │     MAGNETECO SERVICE       │
              │                             │
              │  • Memorize (conversations) │
              │  • Events (system events)   │
              │  • Webhooks (external)      │
              │  • Retrieve (queries)       │
              │  • Tools (agent interface)  │
              └─────────────┬───────────────┘

       ┌────────────────────┼────────────────────┐
       │                    │                    │
       ▼                    ▼                    ▼
   PostgreSQL            Neo4j                 S3
   + pgvector            (graph)            (raw logs)

Next Steps

On this page