Ship & Sleep: Production for Vibe Coders

9 lessons · ₪99

Practice shipping, watching and recovering a live app. Topics that Core also covers start with a short plain explanation. You still need Git, npm and a terminal — from Core or equivalent — and Lab 2 also needs Docker Desktop. Lab 1 builds Pulse on Vercel (logs, charts, errors, speed, Blob, alerts). Lab 2 runs the same observe-break-fix loop locally with Docker, Postgres, Prometheus and Grafana. A real production release stays a separate, owned decision.

Production course overview — from a working demo to a live app you can run safely

You built an app that works. Now real people are about to depend on it.

0/57 topics

0/8 lab steps

Try the free lesson

Buy · ₪99

Who it's for

Your app is live, or about to be, and real people or real money depend on it.
You're responsible for keeping the app running and responding when something goes wrong.
You can already ask an AI agent for a change and read what it did: commits, branches and a database change. The lessons explain the rest as they go, and work for any app.

Not for: Complete beginners who have never built an app. Core teaches building one from scratch.

What you'll walk away with

A rollback process you've practiced, and a habit of releasing small changes.
Alerts that help you recognize failures and decide which ones need attention.
Backups you've tested by restoring them, database access rules, and spending controls to help manage costs.
Launch preparations that cover terms, privacy, consent and accessibility.
A runbook and the exact prompts to give your AI in the middle of an incident.

Who made this course, and why →

Prerequisites

Node.js (LTS): installed by hand, together with Git below; the AI installs the rest. nodejs.org
Git: the save-point tool every Lab uses from its first steps. On a Mac, typing git in the terminal offers to install it. git-scm.com
An AI coding agent: one that works in your project folder and runs commands. A browser chat can't run your tests. Most of them need a paid plan or API credit: budget roughly $20 a month at the entry tier (prices change). The course examples use Claude Code, and every step works in any of these. When a step names a Claude Code command or folder, ask your agent for its equivalent. Claude Code · Cursor · Codex · Copilot
A code editor (or IDE): to read what changed. An IDE is an editor plus helpers. Cursor is an editor with an AI agent; Claude Code is an agent in the terminal — different tools. VS Code
Accounts you'll need, when a step asks: Lab 1 needs free GitHub and Vercel Hobby accounts, plus free tiers of an error tracker and an uptime monitor (stop if a step asks for a card). Lab 2 needs Docker Desktop (or another Compose engine) on your machine — no cloud account. Set spending limits on every account that bills by usage.

The Labs: what you'll build with AI

Two labs — pick whichever fits your stack (or do both). Finish the Production lessons first. Lab 1 builds Pulse from an empty folder and deploys it to Vercel. Lab 2 builds a local Docker Compose stack with Postgres, Prometheus and Grafana. Both need Git, npm and a terminal — from Core or equivalent practice. Lab 2 also needs Docker Desktop installed before you start it.

Lab 1 · Pulse on Vercel

Build Pulse, a tiny status board, from an empty folder: harden the GitHub repo (including adding a collaborator), deploy to Vercel, add CI and preview cleanup, read logs and charts, cause and fix errors, improve load time, use Blob and alerts, then rehearse a staging-only incident.

Lab 2 · Local stack with Docker and Grafana

Build a small Pulse-shaped app from an empty folder and run it locally with Docker Compose: app, Postgres, Prometheus and Grafana. Cause a failure, see it on charts, fire a local alert, then tear the stack down. No cloud bill.

By the end, you'll have shipped a practice app to Vercel with logs, charts, error tracking and alerts, improved its load time, and run the same observe-break-fix loop locally with Docker, Prometheus and Grafana.