Memory
Memory
Where your system keeps state. Streams for structured workflow output, disks for files, a managed database Swirls provisions, and the Postgres you already have.
What it is. The places a Swirls system keeps state, each shaped for a different kind of data.
Use it when a workflow produces data another run needs, an agent needs a working file system, or a relational database (yours or one Swirls provisions) holds the records a workflow should read and write.
- Streams: Swirls-managed persistent storage for workflow output. Structured, typed records you can filter and reuse in other workflows or later runs of the same one.
- Views: a spreadsheet over your streams. Map each stored row into a table and add computed columns that run a graph per row (for example, an AI enrichment).
- Disks: a platform-provisioned shared file system for unstructured files. Workflows run commands on it; agents mount it in their sandboxes. Agents are very good at navigating file systems.
- Database: a PostgreSQL database Swirls provisions and manages for you, declared with a Prisma schema and queried through a generated typed client.
- Postgres: your external PostgreSQL database, declared with its table schemas so workflows can run parameterized SQL against it.
A useful rule of thumb: workflow output goes to a stream, working files go to a disk, a new relational store goes to a managed database, and records in a database you already run stay in Postgres.