Insforge and Claude Code: build a restaurant app

This tutorial builds Cloud Kitchen, a restaurant ordering application, with Insforge and Claude Code. The presenter introduces Insforge as a Supabase alternative, then creates a Next.js project and connects the coding assistant through Insforge's CLI and skills. Prompts generate database tables for dishes and orders, plus a dashboard with a Kanban board.

The walkthrough adds live order updates across browser sessions and image uploads to a storage bucket. An upload permission error leads to policy changes. After adding authentication, the presenter restricts uploads to authenticated users. Email verification and GitHub sign-in appear in the demonstration; Google sign-in is also requested.

Data separation needs further work: accounts initially see existing dishes and orders. The presenter requests separate user data, then adds restaurants with invited members who share orders. The example demonstrates status changes between two accounts.

Insforge's Model Gateway supplies AI-generated dish descriptions. The presenter initially selects GPT-5 Mini, but the generated implementation names a different OpenAI model, which he enables before testing the button. This illustrates why generated configuration needs review.

The final steps deploy through Insforge to Vercel and inspect logs and environment variables. This is a hosted backend workflow; the tutorial does not demonstrate running the backend or models on the viewer's own hardware.