Skip to main content
Start with ERD
Products -> ERDPricingDocsBlogSign in한국어Start with ERD

Getting Started

AI-ERD is the source of truth for your database schema in AI-assisted coding workflows.

Stop letting AI guess your database schema.

Describe your database in plain language or paste SQL — AI builds the ERD instantly. Connect via MCP and let Claude Code, Codex, Copilot, Cursor, Gemini, or Windsurf read and modify the exact schema instead of guessing.

Why AI-ERD?​

When you ask an AI assistant to write code that touches your database, it has to guess the schema unless you paste it in every time. That guess is wrong often enough to cause subtle bugs — wrong column names, wrong types, missing foreign keys, hallucinated tables.

AI-ERD solves this by becoming the single source of truth that your AI coding tools can read directly:

  1. Design once — Build the schema visually, by natural language, or by importing existing SQL.
  2. Connect via MCP — Your AI coding assistant calls AI-ERD's MCP server to read or modify the real schema.
  3. Generate accurate code — The AI works against actual table names, column types, and relationships — no guessing.

Quick Start​

1. Create an ERD​

Open AI-ERD, sign in, and create a new ERD document from the home screen. You can start three ways:

  • Empty canvas — Add tables manually with the visual editor.
  • Natural language — Open the AI panel, describe your domain ("a SaaS app with users, projects, and tasks"), and let AI generate the initial schema.
  • SQL import — Paste a CREATE TABLE script (MySQL, PostgreSQL, Oracle, or SQL Server) and AI-ERD converts it.

2. Refine the schema​

Use the visual canvas to fine-tune:

  • Drag tables to organize the layout
  • Click a table to edit columns, indexes, and foreign keys in the side panel
  • Draw relationships by point-and-click between columns
  • Group related tables with colored containers

See Core Concepts for the data model and Visual Canvas for editor details.

3. Connect your AI coding tool via MCP​

The Model Context Protocol (MCP) lets your AI assistant call AI-ERD directly. One integration works across:

Once connected, your assistant can list tables, read columns, create new entities, modify relationships, and snapshot every change — all against the real schema.

See MCP Overview for how the protocol works, or jump straight to your client's setup page.

What's next​

Need help?​

If anything in these docs is wrong, missing, or unclear, please open an issue or contact support. The product changes quickly — feedback keeps the docs honest.