Building a Practical AI Agent to Help SQL Server DBAs Triage Performance Problems

Every experienced SQL Server DBA has a set of trusted scripts, queries, and diagnostic techniques they rely on when performance problems strike.

Over time, those techniques become institutional knowledge — powerful, but often locked away in notebooks, documents, and muscle memory.

In this session, you’ll learn how to turn your own DBA expertise into a practical AI powered assistant that helps triage and diagnose SQL Server performance issues — without replacing your judgement or existing processes.

We’ll walk through building an AI agent that:

• Uses your documentation and investigation notes to understand how you diagnose issues
• Runs real diagnostics using DMVs and system data
• Explains why it asked for specific information
• Produces evidence based findings and recommended next steps

The solution can be driven directly from SQL Server via REST endpoints, allowing the agent to be invoked from a stored procedure, application, or external tool — wherever it best fits your workflow.

Rather than a generic “AI for DBAs” talk, this session focuses on practical patterns that allow you to build a DBA assistant that works the way you already work, helping you investigate performance problems faster and more consistently.


Intended Audience

SQL Server DBAs
Data Platform Engineers
Developers responsible for SQL Server performance and stability

• Understand how to translate existing DBA processes into agent friendly instructions
• See how AI agents can safely and usefully work with SQL Server diagnostics
• Learn a repeatable pattern for building their own DBA helper agent