Surrealdb

Multi-model cloud database supporting document, graph, and relational queries in one engine.

Overview

SurrealDB is a multi-model database designed to unify documents, graphs, vectors, time-series, and relational data in a single ACID-compliant platform. Rather than stitching together separate systems for different data models, SurrealDB consolidates storage, context, and memory into one transaction boundary, eliminating middleware and reducing operational complexity.

The platform is positioned as "the context layer for AI agents," addressing structural failures in agent systems caused by fragmented data across multiple stores. It's built on a four-layer architecture: Spectron (agent memory), SurrealDB (unified context), and a distributed storage engine with compute-storage separation.

SurrealDB is in production at Samsung, Verizon, Tencent, and PolyAI, with deployment options ranging from cloud-managed instances to self-hosted and embedded configurations.

Key features

Multi-model querying (documents, graphs, vectors, time-series)
ACID transactions across multiple tables
Real-time live queries and changefeeds
Native graph relationships and traversal
Vector search and embeddings
Built-in authentication with scoped permissions
Spectron agent memory layer
Full-text search
Geospatial data support

Pros

Single transaction boundary eliminates middleware and glue code
Unified ACID compliance across all data models
Reduces token costs for AI agents by consolidating context
Lower latency than multi-system architectures
Flexible deployment: cloud, self-hosted, embedded, or Docker
Proven at scale (700K users, 50M graph edges, 10K+ QPS)
Comprehensive SDK support across 9+ languages

Cons

Spectron (agent memory layer) still in waitlist/early access
Learning curve for developers unfamiliar with multi-model databases
Requires understanding of SurrealQL query language
Some advanced features marked as 'coming soon' (database branching, object storage backends)

Use this if

You need a single database for documents, graphs, vectors, and time-series; you're building AI agents or RAG systems; you want ACID compliance across multiple data models; you're replacing multiple specialized databases; you need real-time live queries.

Skip this if

You need a simple key-value store; you require a fully managed serverless database with zero configuration; you're locked into a specific database ecosystem; you need extensive third-party integrations out of the box.

Best for

AI agents and RAG workflowsReal-time applications with live queriesKnowledge graphs and entity relationshipsApplications requiring multiple data modelsProduction workloads needing ACID complianceTeams replacing multiple database systems

Alternatives

PostgreSQL with PostGIS and pgvectorMongoDB with Atlas Vector SearchNeo4j for graph-focused workloadsPinecone or Weaviate for vector-only use casesElasticsearch for full-text searchDuckDB for analytical workloads

Compare Surrealdb alternatives

More Database

Compare all

PocketBaseFeatured

Open-source backend in a single file: database, auth, file storage, and admin UI.

MariaDB

Community-developed SQL database fork of MySQL with compatible APIs and enterprise features.

Microsoft Access

Desktop database and app builder for teams that need forms, reports, and relational data.

Microsoft SQL Server

Enterprise relational database for transactional workloads, analytics, and hybrid cloud.