Answers before you install

SQL Performance Intelligence is a Windows desktop application for read-only SQL Server diagnostics. It combines deterministic analysis, runtime signals, Query Store evidence, execution plans, and optional AI interpretation to help DBAs investigate performance issues and produce audit-ready reports.

It is designed for DBAs, database developers, engineering teams, and consultants who need to diagnose SQL Server performance, review security signals, or communicate evidence-backed recommendations without deploying an agent or automatically changing production systems.

The application includes Dashboard, Query Statistics, Index Advisor, Blocking Analysis, Wait Statistics, Security Audit, Scheduled Jobs, and Object Explorer workflows. Each module focuses on a specific operational question and can export evidence for review.

No. The product is a decision-support and investigation tool. It surfaces evidence, recommendations, and scripts for review; a qualified operator remains responsible for validating and applying any change.

No. SQL Performance Intelligence is an interactive diagnostic workbench, not a persistent monitoring service. It does not continuously collect data while the desktop application is closed, provide a centralized multi-server operations console, measure SLA or uptime, or replace an enterprise alerting platform. Historical investigation can use evidence already retained by SQL Server Query Store and snapshots created during product use.

The desktop application runs on Windows 10 and Windows 11 and targets Microsoft SQL Server environments. See the installation guide for current prerequisites, connection preparation, and first-run setup.

Use a dedicated read-only diagnostic account. Typical access includes CONNECT, VIEW SERVER STATE, VIEW DATABASE STATE, VIEW DEFINITION, and limited access to system metadata. Exact requirements vary by module, so use the documented permission script and have your DBA review it before production use.

Query Store is strongly recommended because it provides richer historical evidence, plan history, and workload trends. Some workflows can fall back to live DMV data when Query Store is unavailable, but historical depth and confidence may be more limited.

Follow Quickstart to configure a SQL connection and AI provider, then open Dashboard to confirm that the active server, database, and top-level health metrics load correctly. From there, choose the module that matches the issue you are investigating.

No automatic schema, index, job, or data changes are applied. The application performs read-only diagnostics and may generate recommendations or scripts, but those outputs stay under operator control.

Yes. Local AI through Ollama is the default path and can run without sending prompts or database context to a cloud model. Core analysis, filtering, refresh, and export workflows do not depend on a cloud AI provider.

You can configure local Ollama or supported hosted providers such as OpenAI, Anthropic, Azure OpenAI, and DeepSeek. Provider availability and settings are managed in the application, and cloud use is optional.

Cloud-directed analysis is an explicit user action. Send-policy and redaction controls govern the context prepared for the provider, while showing or sending sensitive values requires additional operator choice. Review your organization’s data policy before enabling any hosted model.

No. SQL Performance Intelligence is a desktop application and does not require background agents or persistent server-side collectors. It reads diagnostic signals when you connect and run or refresh a workflow.

Every plan includes a 30-day full-feature trial, and no credit card is required to start. Download the application, complete first-run setup, and start the trial from the licensing flow.

Licensing is per user, not per SQL Server. A licensed user can connect to multiple SQL Server environments. Team and enterprise options provide multiple user licenses for broader rollout.

Yes. Use the trial for technical validation and contact us for procurement, multi-user rollout, or enterprise requirements. We recommend validating permissions, AI policy, and representative workloads as part of the evaluation.

First check the Query Store health indicator, permissions, selected server and database, and whether the workload has executed during the available retention period. If Query Store is unavailable, the module may use live DMV data with less historical depth.

Confirm that Ollama is running, the selected model is installed, and the configured host URL is reachable from the desktop application. Then use the provider test in Settings before retrying the analysis.

Use the documentation for setup and workflow guidance. If the issue is not covered, contact us with the product version, affected module, and a concise description of what you expected and observed. Avoid including credentials or sensitive database content.