DDLBuilder Docs

English

中文
日本語

English

中文
日本語

Appearance

AI-Assisted Table Design Workflow ​

This guide explains how to leverage DDLBuilder's AI suite (Master Workshop, AI Modify, AI Index Advisor, and Smart Comments) for intelligent schema modeling and refactoring.

Overview ​

Transform ambiguous product requirements into production-ready schemas, or safely refactor established legacy tables using conversational guidance, structured patch reviews, and domain-aware recommendations.

Core AI Workflows ​

1. Master Table Workshop (Conversational Green-Field Design) ​

  1. Click Master Workshop in the table header to open the AI panel.
  2. Describe Your Business Domain: Input entities, constraints, and query patterns (e.g., "Design a multi-tenant e-commerce order table with tenant ID, order code, status, price, discount amounts, optimized for date-range queries").
  3. Template Integration: Check template bundles to instruct the AI to incorporate standard audit columns (e.g., created_at, updated_by).
  4. Review Architecture Decisions: The AI generates a complete schema accompanied by rationale notes explaining data type choices and indexing strategies.
  5. Iterate Across Turns: Refine the design naturally (e.g., "Add a soft-delete column", "Change price columns to high-precision decimals").
  6. Click Apply to Table to populate your workspace with columns, indexes, and schema settings.

2. AI Modify Current Table (Structured Patch Review) ​

When iterating on existing schemas without overwriting prior work:

  1. Click AI Modify in the top navigation bar.
  2. Specify the desired changes (e.g., "Add openid column, and change phone unique index to a composite unique index on (tenant_id, phone)").
  3. Inspect the Generated Diff: The AI computes fine-grained diff items categorized into table, column, and index mutations.
  4. Approve or Reject Items: Individually accept or reject each proposed modification to maintain full manual control.
  5. Click Apply Selected Changes to update the table atomically.

3. Smart Business Comment Generation ​

  • Click Generate AI Comments on any active or imported table.
  • The AI infers semantics from naming patterns (e.g., is_deleted, pay_channel) and data types, automatically populating descriptive labels to streamline data dictionary documentation.

Verification Checklist ​

  • [ ] AI-generated columns, data types, and indexes accurately populate the workspace.
  • [ ] In AI Modify mode, all targeted changes are individually reviewed and applied.
  • [ ] Critical constraints (nullability, uniqueness, precision) are manually verified.
  • [ ] Generated comments and labels reflect clear domain terminology.

Tips and Best Practices ​

Prompt Quality and Precision

Detailed prompts specifying query filters, workload characteristics, and key constraints yield significantly higher-quality indexing and type recommendations.

  • Draft Verification: AI output represents a high-quality initial draft; always review constraints before deploying DDL to production databases.
  • Account & Credits: AI features consume account credits based on token usage. Guest users will be prompted to sign in before initiating requests.
  • Resetting Context: If a multi-turn conversation drifts off topic, click "Start Over" to clear the session context and restart with a clean prompt.