What an AI agent can do for your operations
Running SQL Server means a lot of repetitive checks and a few critical moments when you need to understand what is going on, fast. A well-scoped AI agent takes care of the first part and prepares the second. For example:
- checking backups, SQL Server Agent jobs, disk space and alerts every day, and only flagging what needs a decision;
- spotting a performance regression in Query Store or the wait statistics, and explaining which query changed plans and why;
- preparing the file for an incident (blocking, deadlocks, saturation) before a DBA even opens it;
- drafting a fix script, an index or a configuration change, together with its rollback script, and submitting it for approval.
The same principles apply to the infrastructure around your databases: Windows and Linux servers, storage, backups, virtual machines and Kubernetes clusters.
Four non-negotiable safeguards
1. Access limited to performance data and metadata
The agent uses a dedicated login with read permissions on only the system views it needs (such as VIEW SERVER STATE and VIEW ANY DEFINITION). It sees how your databases work, not what they contain.
2. Human approval before any action
The agent proposes, an authorized person decides. No production change goes out without explicit approval, and the approval rules are defined with you according to how critical each server is.
3. A complete audit trail
Every read, every recommendation, every approval and every action is timestamped and kept. At any time you know who approved what, when, and on which evidence.
4. A rollback planned in advance
Every change ships with the procedure to undo it, checked before it is applied. If the result is not what was expected, you go back to the previous state.
Why DBAs should build it
An AI agent is no more careful than the rules it is given. Ours come from more than 15 years of running SQL Server in production and from the diagnostic rules we apply in our SQL Server audits. We know which alerts matter, which fixes are safe and which ones must wait for a maintenance window. The agents complement our remote DBA services or the work of your own team.
How we get started
- A scoping workshop: which servers, which risks, which tasks the agent takes on, and the approval rules.
- A read-only pilot on one instance, with no automatic actions, to judge the quality of the analysis.
- A gradual rollout, server by server, with regular reviews of the audit trail.
Let's talk about your operations
Tell us about your environment and what you would like to automate: we reply within one business day. You can also use our contact page.
