Ship a typed API in five minutes.

This guide walks you from npm i to a deployed, observable backend with a real Postgres database, generated clients, and a function running at the edge. No Docker. No YAML.

Prerequisites

You'll need a few things on your machine before you start. Lambda runs on every recent runtime; we test against the matrix below in CI.

• Node.js 20.x or later (Bun, Deno, and Node 22 are also supported).
• A package manager — npm, pnpm, yarn, or bun.
• A Lambda account. The free tier includes one project, 1 GB of Postgres, and 1M function invocations.

1. Install the CLI

The Lambda CLI is a single static binary. Install it with your package manager of choice — it ships TypeScript types and the runtime SDK in the same package.

1# installs the CLI & runtime SDK
2$ npm install -g @lambda/cli
3
4# sign in — opens your browser
5$ lambda login
6→ Authorized as maya@acme.co (Acme Corp · production)

$ pnpm add -g @lambda/cli
$ lambda login

$ yarn global add @lambda/cli
$ lambda login

$ bun install -g @lambda/cli
$ lambda login

2. Initialize a project

Run lambda init in an empty directory. The CLI scaffolds a project, links it to a workspace on Lambda Cloud, and writes a lambda.config.ts you can commit.

3. Define your first model

Models are TypeScript files that compile to a typed REST API, a typed client, and a Postgres migration — automatically. Save the file and the CLI does the rest in under a second.

1import { model, id, ref, money, enumOf, timestamp } from "@lambda/schema";
2import { Customer } from "./customer";
3
4export const Order = model(({
5  id:        id("ord_"),
6  customer:  ref(Customer),
7  total:     money({ currency: "USD" }),
8  status:    enumOf("pending", "paid", "refunded"),
9  createdAt: timestamp({ default: "now()" }),
10}).index("customer", "status");

4. Endpoint reference

Every model gets the standard six endpoints. Below is the auto-generated reference for the Order model you just defined.

5. Deploy to production

When you're ready, lambda deploy ships your project to every region you've configured. Deploys are atomic — they either succeed everywhere or roll back instantly. There's no half-state.

1$ lambda deploy --env production
2
3  ✓ typecheck            0.3s
4  ✓ generate clients     0.4s · 4 langs
5  ✓ migrate database     2.1s · 1 plan applied
6  ✓ build edge bundles   3.6s · 41 functions
7  ✓ ship to 38 regions   7.8s
8
9→ https://acme.lambda.app  (commit a3f9c1d · build #1842)

That's it. Your API is live, your clients are generated, your database is migrated, and every request is being traced end-to-end. Open the dashboard to see logs streaming in real time.