Modeling with AI

You can generate a complete data model by describing your application to an AI assistant and importing the resulting schema.

How it works

  1. Click the AI button in the top bar to open the AI dialog.
  2. Choose a tab:
    • New schema — generate a schema from scratch.
    • Edit existing — modify your current schema.
  3. Click Copy prompt to copy a pre-built prompt to your clipboard.
  4. Paste the prompt into your preferred AI assistant (ChatGPT, Claude, Gemini, etc.) along with a description of what you want to build.
  5. Copy the JSON output from the AI.
  6. Back in ScaffoldHub, click Load schema and paste the JSON.

The AI dialog also provides Import and Export buttons to save and load schema files directly.

Tips for good prompts

  • Be specific about your domain — mention the entities you expect (e.g., "a project management app with projects, tasks, and team members").
  • Describe relationships — "each project has many tasks", "users belong to one organization".
  • Mention field constraints — "email must be unique", "price is a decimal with 2 decimal places".
  • Specify enum values when relevant — "status can be draft, published, or archived".
  • Ask for validations — "title is required, max 200 characters".

After importing

Review the generated schema on the canvas. You can edit any entity, field, or relationship manually to fine-tune the result. See modeling manually for details.

Projects and modelingModeling manually