Turso

Overview

Turso is a modern cloud-native distributed database built on a complete rewrite of SQLite, designed for the multi-database architecture of the AI agent era. It breaks away from the traditional single-instance shared paradigm, giving every agent, user, or tenant its own independent database. All data is stored as files, enabling "always-on, instantly available" operation that completely eliminates cold starts and wake-up latency.

Building on SQLite, Turso adds key capabilities such as concurrent writes, vector search, cloud-native access, and multi-region replication. It remains compatible with the standard SQLite protocol, so developers can migrate seamlessly and enjoy cloud elasticity. The file-based nature of the database makes it naturally suited to multi-tenant scenarios, and with the libSQL client, data can be read and written locally at edge nodes for global distribution.

Whether you're building autonomous agents, AI applications that need long-term memory and task coordination, or edge SaaS solutions that demand low latency and high availability, Turso provides a lightweight, open, SQLite-compatible experience that serves as a solid data foundation for the next generation of AI-native applications.

Core capabilities

Multi-database architecture

Assign an independent database to each agent, user, or tenant with native multi-tenant isolation

Database as a file

The database exists as a file with no cold starts or wake-up latency, always online

Concurrent writes

Supports multiple concurrent writers on top of SQLite, breaking the traditional write-lock bottleneck

Built-in vector search

Natively supports vector indexing and similarity search for building RAG and semantic search

Edge replication and distribution

Databases can be replicated and accessed locally at edge nodes worldwide for ultra-low-latency reads and writes

Full SQLite compatibility

Compatible with the standard SQLite protocol and libSQL clients, with zero-cost migration and integration

Cloud-native access

Provides managed service, API, and CLI, with databases ready on demand and elastic scaling

AI Agent friendly

Designed for AI agent scenarios, naturally suited for agent memory, assets, and task collaboration

Use cases

Assign a dedicated database to each AI agent for storing memory and task state

Build RAG-powered AI applications using built-in vector search for semantic retrieval

Provide natively isolated database instances for multi-tenant SaaS, simplifying permission management

Deploy low-latency read/write services at edge nodes to support global user access

Seamlessly migrate existing SQLite apps to the cloud with elastic scaling capabilities

Store user-uploaded files, assets, and configuration to support agent task coordination

Provide chatbots with long-term memory and persistent storage of conversation history

Achieve local-first offline data sync in IoT and mobile scenarios