​
Hystersis

Give your AI agents persistent memory that compounds over time. Hystersis is an open-source memory infrastructure for AI agents. It provides semantic search, knowledge graphs, and procedural memory that grows with every interaction. Hystersis Architecture

​
Why Hystersis?

  • Memory That Adapts - Agents learn from past interactions
  • Knowledge Graphs - Understand relationships between entities
  • Semantic Search - Find information using natural language
  • Multi-Agent Support - Shared memory across agent teams
  • 85% Compression - Efficient storage without losing context

​
Quick Start

from hystersis import Hystersis

client = Hystersis("https://api.hystersis.ai", api_key="your-key")

# Create a memory
memory = client.create_memory(
    content="User loves machine learning",
    user_id="user-123"
)

# Search semantically
results = client.search("deep learning AI")

​
Features

FeatureDescription
Conversational MemorySession-based message history
Semantic SearchVector-based similarity search
Knowledge GraphEntities with typed relationships
Procedural MemoryExtract and reuse skills
Multi-Agent PoolReal-time shared memory

​
Status

Current status: Operational Better Stack

​
License

MIT License - See GitHub for details.