Skip to content
NEW BLOG

Using Surrealism to build your own extensions

Read blog

1/2

SurrealDB vs. traditional databases

Three generations of database architecture for the AI era. From single-model silos to the context layer.

Traditional databases support one model well, but not native graph traversal, vector search, temporal queries, or unified multi-model execution. As AI workloads grow, teams bolt on extensions, glue systems together, and manage ever-growing operational complexity.

SurrealDB was designed from the ground up as a native multi-model engine.

DATABASE EVOLUTION

From single-model to native multi-model

Single-model (Postgres, MySQL)

One data model, SQL, extensions for other models. Not designed for graphs or vectors natively.

Bolt-on multi-model (MongoDB, Neo4j)

One model done well, but still single-purpose. Other models added as extensions or separate systems.

Native multi-model (SurrealDB)

Documents, graphs, vectors, time-series, geospatial natively. One query language, one transaction boundary.

THE GAP

What traditional databases miss

No native graphs

Postgres requires joins and CTEs. Graph traversal is a workaround, not a first-class feature.

No native vectors

Extensions like pgvector are bolt-ons. Not designed for hybrid queries or agent workloads.

No temporal awareness

No bi-temporal versioning. Can't query “what did we know when?”

No agent memory

No structured memory for AI agents. No knowledge graphs, no entity extraction built in.

THE SOLUTION

Everything in one engine

Native multi-model

Documents, graphs, vectors, time-series, geospatial. All first-class, all in one engine.

SurrealQL

One query language for everything. SQL-like, familiar, powerful. No glue code between models.

ACID across models

One transaction boundary. Update a document and its graph relationships atomically.

Agent-ready

Spectron for memory. Knowledge graphs, entity extraction, temporal facts - built into the database.

GET STARTED

Move to native multi-model