What is Prisma?
Prisma’s data platform runs two things for you: Prisma Postgres, a serverless Postgres database, and Prisma Compute, a runtime for the apps that use it. A project holds both, so an app and its database are deployed side by side, and the app finds its database without any configuration.
Take one small app: an API on Compute with a Postgres database, and a branch per preview. A workspace holds the project. Each branch of the project has its own database and app, and the app receives its database’s DATABASE_URL.
Workspaces and tokens
Section titled “Workspaces and tokens”A workspace holds your projects, members and bill. A service token belongs to a workspace, and Alchemy deploys into the workspace your token belongs to.
Connect it once, in Setup.
Projects
Section titled “Projects”A project groups a database and the apps that use it. By default, creating one creates a Prisma Postgres database too, in the region you choose:
const project = yield* Prisma.Project("app", { region: "us-east-1" });Apps in the project receive that database’s DATABASE_URL without any setup.
Prisma Postgres
Section titled “Prisma Postgres”Prisma Postgres is serverless Postgres: there’s no instance to size, and you pay for what you use. Declare a database on its own when it needs a lifecycle separate from the project’s:
const postgres = yield* Prisma.Postgres("db", { project });See Postgres.
Connections
Section titled “Connections”A connection is a set of credentials for one database. It gives you a ready-made databaseUrl, direct and pooled connection strings, and the parts of the URL separately, for poolers like Cloudflare Hyperdrive. Give each app its own.
const connection = yield* Prisma.Connection("api", { database: postgres });See Connections.
Branches
Section titled “Branches”A branch groups databases and apps under a git-style name. The project’s first branch has the production role. Every other branch is a preview, so apps on it get preview settings. Give each preview stage a branch, and its database and app are kept apart from production.
See Branches.
Compute apps
Section titled “Compute apps”Prisma Compute runs your app on Bun, next to its database. Point it at a directory, and Alchemy builds the app, uploads it, and starts it. Or define it inline, as an Effect program:
const app = yield* Prisma.Compute("api", { project, path: "./app", build: "auto" });The app listens on PORT, which Compute sets. There’s no Docker image and no registry.
See Apps.
Deployments
Section titled “Deployments”Each change to an app becomes a deployment. The new deployment starts next to the old one, must pass its health check, and only then takes traffic. If it fails after taking over, Alchemy restores the one before.
See Deployments.
Buckets
Section titled “Buckets”A bucket stores files next to the database, through an S3-compatible API.
See Buckets.
Local development
Section titled “Local development”Under alchemy dev, the same Stack runs against a local Postgres, started for you with nothing to create in the cloud, and each app runs its framework’s own dev server.
How @alchemy.run/prisma models these concepts
Section titled “How @alchemy.run/prisma models these concepts”Each of these concepts is something your Stack declares.
Your token picks the workspace, and Prisma.providers() deploys into it. A project, its databases, connections and branches are declarations. A Compute app is built and deployed from a directory or an Effect program. Inside any runtime, a Compute app, a Lambda Function or a Cloudflare Worker, Prisma.Connect turns a connection into a database URL for Prisma ORM, Effect SQL or Drizzle.
Next, build one: the tutorial puts these ideas to work, one step at a time. Or read the concepts in depth, starting with Postgres.