Skip to main content

​
Overview

AI DBA Assistant AI DBA Assistant transforms how you interact with your databases. Instead of writing complex queries or navigating multiple monitoring dashboards, simply ask questions in plain English. The AI analyzes your fleet’s monitoring data and returns insights with actionable recommendations. Key capability: The Assistant doesn’t just identify problems—it provides complete query analysis with remediation plans to fix them.

​
Getting Started

Access the Assistant from the left navigation panel under AI DBA → Assistant. The interface is straightforward: type your question in the input field and press enter or click the send button. Conversation Controls:
  • New conversation — Starts a fresh conversation, clearing the current context
  • Clear history — Removes all messages from the current view (disabled when no messages exist)
Important: The Assistant maintains conversation history within a session using a conversation ID. However, each new conversation starts fresh with no memory of previous sessions. Include all relevant context when starting a new conversation or asking follow-up questions that reference earlier messages. Example first question:
“Show me the top 3 longest queries from the last day”

​
Key Features

​
Natural Language Queries

Ask questions the way you would ask a senior DBA. The AI understands context and database terminology across both MySQL and PostgreSQL environments. What you can ask:
  • “Are there any performance issues in my databases over the last 24 hours?”
  • “How many locks do we have today on the fleet?”
  • “Which instances have the highest connection count right now?”
  • “Show me slow queries affecting the payments table”
  • “What’s causing high CPU on our production database?”
  • “Analyze the queries running longer than 5 minutes”
  • “Why is this query slow: SELECT * FROM orders WHERE created_at > ‘2024-01-01‘“

​
Rich Analysis Results

Every response includes multiple components designed to give you complete context:

​
Findings Summary

A clear statement of what the AI discovered, including:
  • Affected database instances
  • Timeframes and frequency
  • Severity indicators
  • Impact on performance

​
The Query

When relevant, the AI displays the actual SQL being analyzed, formatted for readability. This includes:
  • Full query text with syntax highlighting
  • Execution time and resource usage
  • Database and user information

​
Query Analysis

The AI doesn’t just show you what’s slow—it explains why and how to fix it: Root Cause Identification:
  • Missing indexes
  • Inefficient joins or subqueries
  • Full table scans
  • Locking or blocking issues
  • Resource constraints (CPU, memory, I/O)
Remediation Plans: Each analysis includes a step-by-step remediation plan with:
  • Specific actions to implement
  • Ready-to-run SQL for creating indexes or modifying queries
  • Estimated performance impact (e.g., “Expected 80% reduction in execution time”)
  • Implementation guidance with considerations and best practices
Example Remediation Plan: 
Problem: Full table scan on `orders` table (5M rows)
Root Cause: Missing index on `created_at` column

Recommended Action:
CREATE INDEX idx_orders_created_at ON orders(created_at);

Expected Impact:
- Reduce query time from 45s to ~2s (95% improvement)
- Reduce CPU usage during query execution

Considerations:
- Index creation will lock the table briefly
- Recommended to run during low-traffic period

​
Recommendations

Prioritized action items beyond the immediate query fix:
  • Related optimizations for similar queries
  • Schema design improvements
  • Configuration adjustments
  • Monitoring suggestions

​
Follow-up Questions

Suggested next steps to continue your investigation. Click any suggestion to drill deeper into the issue. Examples:
  • “Show me all queries using the orders table”
  • “What other indexes might help this workload?”
  • “Are there any blocking queries right now?”

​
Results & Visualizations Panel

AI DBA Assistant Chat For queries that return data, a side panel displays interactive results:

​
Data Tables

Searchable result sets showing:
  • Query details and full SQL text
  • Execution metrics (duration, CPU, rows examined)
  • User information and connection details
  • Query state and status
Actions: Export data to CSV or copy directly from the view

​
Charts & Graphs

Visual representations of fleet-wide metrics:
  • Lock counts by database instance
  • Average wait times across the fleet
  • Distribution breakdowns (lock types, query types, etc.)
  • Time-series trends (CPU, connections, query duration)

​
Quick Stats

At-a-glance metrics:
  • Total fleet locks
  • Instance counts
  • Aggregate values (total queries, average duration, etc.)

​
Action Buttons

Each analysis includes action buttons to move from insight to resolution: Result Table View the raw data behind the analysis in a sortable, filterable table format. Copy Generated Query Get the SQL that Assistant used internally, ready to paste into your own tools or save for future reference. Execute Query Opens Rapydo’s Execute Query tool with the generated SQL pre-loaded. From there you can:
  • Review and modify the query
  • Run it against RapydoDB for further investigation
  • Save it as a template for future use
  • Share it with your team

​
Best Practices

​
Be Specific About Timeframes

“Last 24 hours” or “since yesterday” helps Assistant focus on relevant data. Examples:
  • ❌ “Show me slow queries” (too vague)
  • ✅ “Show me queries running longer than 10 seconds in the last 6 hours”

​
Name Your Instances

If you’re investigating a specific server, include its identifier in your question. Examples:
  • ❌ “Why is the database slow?”
  • ✅ “Why is production-db-01 experiencing high CPU?”

​
Include Full Context

Since each question is independent, provide all necessary details in a single message rather than relying on follow-ups. Examples:
  • ❌ “What about yesterday?” (no context from previous question)
  • ✅ “Show me CPU usage on production-db-01 yesterday between 2pm and 4pm”

​
Ask for Analysis When Needed

If you want recommendations and remediation plans, explicitly ask for analysis. Examples:
  • “Analyze this query and suggest optimizations: SELECT * FROM…”
  • “What’s causing slow performance on database X and how can I fix it?”

​
How It Works

Assistant uses two data sources depending on the type of question asked:
  • RapydoDB — Rapydo’s dedicated monitoring database, queried for performance metrics, process history, slow queries, CPU/memory trends, locks, and fleet-wide activity data.
  • Your database instances — Queried directly for schema information and data retrieval intents (e.g., listing tables, retrieving schema details).

​
Fast Responses

Performance and monitoring data is already aggregated in RapydoDB, so those queries return quickly regardless of your fleet size.

​
Historical Analysis

All monitoring data is retained in RapydoDB, enabling you to investigate past incidents and identify long-term trends. 💡 Important: The Assistant cannot modify your production databases or execute DML/DDL commands. It is read-only in all cases.

​
Supported Databases

Assistant analyzes monitoring data from all database engines supported by Rapydo:
  • MySQL (all versions)
  • PostgreSQL (all versions)
The AI automatically adjusts its analysis and recommendations based on the database engine, ensuring relevant and accurate guidance for your specific environment.

​
Common Use Cases

​
Performance Investigation

“Show me queries running longer than 30 seconds on production-db-01 in the last hour” → Get full analysis with root causes and remediation plans

​
Fleet-Wide Monitoring

“Which instances have the highest CPU usage right now?” → View visual charts ranking your instances by resource usage

​
Lock and Blocking Analysis

“Are there any blocking queries affecting the orders table?” → Identify lock chains and get recommendations to resolve contention

​
Query Optimization

“Analyze this query: SELECT * FROM users WHERE email = ‘test@example.com’” → Receive index recommendations and query rewrite suggestions

​
Capacity Planning

“Show me connection count trends across all instances for the last 7 days” → Visual time-series data to identify growth patterns

​
Tips for Better Results

✅ Use specific table or database names when investigating schema-specific issues ✅ Include query text when asking for optimization help ✅ Specify instance names when troubleshooting specific servers ✅ Combine multiple conditions for precise results: “Queries longer than 10s on production AND using the payments table”

​
What’s Next?

  • Query Analysis - Learn about Rapydo’s comprehensive query analysis and remediation features
  • Execute Query - Explore the SQL editor for running custom queries against RapydoDB
  • Queries - Understand real-time query monitoring