Core Concepts

Storage

RiceDB - the storage and memory engine for AI agents.

Rice's storage layer is a high-performance database designed for Multi-Agent AI Systems.

Challenge	Solution
Knowledge has relationships	Integrated semantic linking
Multi-tenant environments	Bitmap-based ACL for zero-latency permission checks
High-frequency updates	LSM-tree storage with Write-Ahead Log (WAL)
Real-time notifications	Pub/Sub with semantic subscriptions

Features

High Performance: Supports both HTTP and gRPC transports
Multi-User ACL: Bitmap-based access control with zero-latency permission checks
Real-time Pub/Sub: Server-side streaming with semantic subscriptions (gRPC)

# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# Or via Homebrew
brew install uv

Add to your pyproject.toml:

[project]
dependencies = [
    "ricedb[grpc,embeddings] @ git+https://github.com/rice-ai-hq/ricedb-python.git",
]

Then sync:

uv sync

git clone https://github.com/rice-ai-hq/ricedb-python.git
cd ricedb-python
uv sync --extra dev --extra grpc --extra embeddings

RiceDB Python Client

Full source code and examples.

Core Concepts

Understanding the Rice architecture.

Slate

State and runtime context management for AI agents.