AI Agents for Smarter SQL Server 2025 Operations

SQL Server 2025 provides richer observability than ever before through Query Store, execution plans, wait statistics, and detailed runtime telemetry. Yet diagnosing performance problems still depends heavily on human correlation, experience, and reaction time.

This session explores how AI agents can augment SQL Server 2025 operations by continuously analysing native SQL Server signals, explaining internal engine behaviour in plain language, and safely assisting with tuning and recovery — without removing DBA control or introducing unsafe automation.

Using real SQL Server internals, we will demonstrate a practical approach to introducing autonomous behaviours across three areas:

Monitoring agents that detect anomalies and explain why SQL Server is behaving a certain way

Tuning agents that analyse execution plans, Query Store regressions, and indexing patterns to generate safe, reviewable recommendations

Self-healing agents that perform tightly scoped, pre-approved actions such as plan forcing or pausing runaway workloads

Rather than “self-driving database” promises, this session focuses on what is realistic in production environments today, the guardrails required to keep AI-driven automation safe, and the scenarios where AI should recommend — not act.

Attendees will see live demonstrations using SQL Server 2025 internals, controlled failure scenarios, and AI-assisted decision-making. You will leave with a clear understanding of where AI agents add measurable operational value, where they do not, and how to introduce autonomy into SQL Server responsibly.

1. Understand how AI agents interpret SQL Server internals such as Query Store, execution plans, waits, and blocking 2. Learn practical patterns for AI-assisted monitoring, tuning, and safe self-healing in SQL Server 2025 3. Identify guardrails and approval models that keep AI automation safe and production-ready