​
Overview

The AI Chat feature enables you to:

  • Query your database using natural language instead of SQL
  • Get insights and explanations about your database structure and performance
  • Diagnose issues through conversational troubleshooting
  • Generate and execute SQL based on natural language descriptions
  • Receive recommendations for database optimizations

​
Getting Started

​
Accessing the Chat Interface

  1. Log in to your dba.ai account
  2. Navigate to the Chat section in the main navigation
  3. Select the database connection you want to interact with
  4. Start asking questions in natural language

​
Sample Questions

Here are examples of what you can ask:

​
Database Structure Queries

"What tables do I have in my database?"
"Show me the schema for the users table"
"What foreign key relationships exist in my database?"

​
Performance Queries

"What are my slowest queries from the last 24 hours?"
"Why is my database CPU usage high right now?"
"How has my query performance changed this week?"

​
Data Analysis Queries

"How many users registered in the last month?"
"What's the average order value by product category?"
"Find transactions over $1000 from new customers"

​
Troubleshooting Queries

"Why am I seeing connection timeouts?"
"What's causing these deadlock errors?"
"How can I reduce my database's memory usage?"

​
How It Works

The AI Chat interface:

  1. Processes your natural language query
  2. Determines the appropriate SQL to answer your question
  3. Safely executes the SQL against your database
  4. Formats and explains the results in an easy-to-understand way
  5. Provides context and insights beyond the raw data

​
SQL Visibility and Safety

For transparency and learning:

  • The AI shows you the SQL it generates before executing it
  • You can review and approve SQL before execution
  • You can edit the generated SQL if needed
  • All queries are run with read-only permissions by default

​
Advanced Features

​
Conversation Memory

The chat maintains context throughout your conversation, allowing for follow-up questions:

You: "How many users do we have?"
AI: "You have 10,532 users in your database."
You: "How many of them signed up this month?"
AI: "1,246 users signed up this month (May 2025)."

​
Data Visualization

For appropriate queries, the AI can generate visualizations:

  1. Ask a question that would benefit from a chart
  2. The AI will generate SQL and execute it
  3. Results will be displayed as both a table and a relevant chart
  4. You can download data or export charts

​
SQL Explanation

When you ask about complex SQL queries:

"Explain what this query does: SELECT u.name, COUNT(o.id) FROM users u LEFT JOIN orders o ON u.id = o.user_id GROUP BY u.name HAVING COUNT(o.id) > 5"

The AI will break down the query step by step, explaining each component in plain language.

​
Best Practices

​
Optimizing Your Questions

For best results:

  • Be specific about the time period for performance questions
  • Mention table names when asking about specific data
  • Specify columns of interest for more focused responses
  • Use precise terms for technical issues

​
Handling Large Results

When working with large datasets:

  • Specify limits: “Show me the top 10 customers by revenue”
  • Ask for aggregations: “What’s the average order value by month?” instead of “Show me all orders”
  • Request specific filtering: “Show active users who haven’t logged in for 30 days”

​
Security Considerations

The AI Chat feature is designed with security in mind:

  • All database connections use TLS encryption
  • Chat sessions use read-only database users by default
  • SQL injection protection is built into the query processor
  • Query history is stored securely and accessible only to authorized users
  • No sensitive data is shared outside your organization

​
Limitations

Current limitations to be aware of:

  • Complex statistical analysis may require traditional SQL
  • Visualization options are currently limited to basic chart types
  • Some advanced PostgreSQL features may not be fully supported
  • Very large result sets may be truncated for performance reasons

​
Feedback and Improvement

The AI Chat system learns from interactions:

  • Use the thumbs up/down buttons to rate responses
  • Provide specific feedback when a response isn’t helpful
  • Suggest improvements in the feedback form

​
Next Steps

Was this page helpful?