Skip to content

ai-features.md

Latest commit

 

History

History
84 lines (61 loc) · 2.49 KB

ai-features.md

File metadata and controls

84 lines (61 loc) · 2.49 KB
Raw

AI Features Guide

QueryLab uses DeepSeek AI to help you generate, fix, and understand SQL queries.

AI-Powered SQL Generation

Generating SQL from Natural Language

  1. Click the "Ask AI" button
  2. Type your request in natural language, for example:
    • "Show me all students older than 20"
    • "Find courses with more than 50 students"
    • "Get the average age of students by course"
  3. Click "Generate" and wait for the AI to create your SQL
  4. Review the generated query and run it

Best Practices for AI Prompts

  • Be Specific: "Show students in Computer Science" is better than "show students"
  • Mention Tables: Reference table names when possible
  • Specify Conditions: Include filters, sorting, and limits
  • Ask for Explanations: Request explanations to learn from the AI

AI Suggestions

The AI can provide suggestions based on:

  • Your current database schema
  • Common SQL patterns
  • Best practices for your use case

AI-Powered Error Fixing

Using "Fix With AI"

When you encounter a SQL error:

  1. The error message appears with a "Fix With AI" button
  2. Click the button to analyze the error
  3. The AI will:
    • Analyze the error message
    • Review your SQL query
    • Consider your database schema
    • Generate a corrected query
  4. Review the fix and explanation
  5. Apply the fix or modify as needed

What Gets Fixed

The AI can fix:

  • Syntax errors (missing commas, typos)
  • Logic errors (wrong table names, invalid joins)
  • Type mismatches
  • Missing clauses
  • Incorrect function usage

Understanding AI Fixes

Each fix includes:

  • Corrected SQL: The fixed query
  • Explanation: What was wrong and why
  • Learning Points: Tips to avoid similar errors

Rate Limiting

To ensure fair usage:

  • AI requests are rate-limited (30 requests per minute)
  • Rate limit errors show when you can try again
  • Authentication is required for AI features

Tips for Best Results

  1. Provide Context: Include relevant table names and schema info
  2. Review Carefully: Always review AI-generated SQL before running
  3. Learn from Fixes: Use explanations to improve your SQL skills
  4. Iterate: Refine your prompts based on results
  5. Combine with Learning: Use AI as a learning tool, not just a crutch

Limitations

  • AI may not always generate perfect SQL
  • Complex queries might need refinement
  • Rate limits apply to prevent abuse
  • Requires internet connection and authentication