Use Gemini 3 To Generate Reports From Your SQL Server Databases

Last week we explored how to use ReportBurster for generating reports from Oracle databases. Today we extend that journey to SQL Server. And since Google's Gemini 3 Pro has been making waves as "the most advanced AI model yet" I decided to put it to work and see how it performs in building reports.

This walkthrough shows, end to end, how to create a report that fetches data from SQL Server. Along the way, we'll see how Gemini 3 Pro accelerates the process.

Starting the SQL Server Sample Database

ReportBurster includes the Northwind SQL Server database, a classic sample dataset that resembles a mini‑ERP system.

Go to Help & Support → Starter Packs / Extra Utils to start the sample database.

Note: Docker must be installed to run the sample databases. Installation instructions are available under the Docker / Extra Utilities tab.

Once Docker is installed & started, stop ReportBurster.exe, then right‑click and Run as Administrator.

ReportBurster requires admin rights to start and stop the Northwind SQL Server Starter Pack (Docker container).

Return to Help & Support → Starter Packs / Extra Utils and start the SQL Server sample.

Configuring the Database Connection

With the SQL Server database running, configure the connection details.

The Database Schema tab displays the complete structure of your Northwind database. You can see all available tables, columns, indexes, and relationships - everything you need to understand your data model. This raw schema view gives you comprehensive access to explore and select the exact tables you'll need for your reports.

At this point, we're connected and the schema is automatically fetched. But the raw schema is just a long list of tables.

To work efficiently, we need structure.

Organizing the Schema with Gemini 3

Gemini 3 can transform the unorganized schema into a Domain‑Grouped Schema. This groups related tables—HR, invoicing, orders—so you immediately know where to fetch data.

We feed Gemini the plain schema and ask: "Please return a Domain‑Grouped Schema."

Gemini 3 Pro delivered a correct Domain‑Grouped Schema on the first attempt. I pasted the JSON (over 3,000 lines) back into ReportBurster, and now the schema is neatly organized.

Refinement is possible, but for this task the result is already solid.

Building the Employee Profile Report

We'll now configure the same Employee Profile report we built for Oracle.

Create a new report and select Report Generation. Then configure the datasource.

Click Hey AI, Help Me With This SQL Query…. Because we defined a Domain‑Grouped Schema, ReportBurster suggests using it when generating the query. For an Employee Profile report, the relevant domain is Human Resources.

Select HR tables, then click Generate SQL Query with Help From AI.

Even if you know SQL, AI makes this faster.

And if you don't know SQL? Suddenly you do.

The prompt has a section for your business requirement:

<BUSINESS_REQUIREMENT>
[Insert your natural language instruction here]
</BUSINESS_REQUIREMENT>

The rest of the prompt guides the AI and should remain unchanged. Future versions of ReportBurster will automatically provide context such as "SQL Server database," so you won't need to add it manually.

Gemini quickly produced a query:

SELECT *
FROM Employees
ORDER BY HireDate DESC;

We refined it through a short chat session to focus Google Gemini only on the fields we needed:

SELECT
  EmployeeID,
  FirstName,
  LastName,
  HireDate
FROM Employees
ORDER BY HireDate DESC;

Preview your query results in the Tabulator tab to ensure you're retrieving the correct data.

Designing the Output Template

With data ready, the next step is Output Template, which defines the report's format (HTML, PDF, Excel, Word, XML, etc.).

We'll generate HTML and again use Gemini 3 Pro to assist.

Initially, Gemini produced a template for the full employee list. I clarified:

No, the report should focus on an individual employee.
Use Freemarker variables to fill employee data.

Gemini then responded with a clean, detailed HTML template for each employee.

Two prompts, and we now have a professional-looking report. Of course, you can continue refining until the template perfectly matches your needs.

Running the Report

With connection details configured, query defined, and template ready, the report is complete. Running it produces the Employee Profile report output.

What's Next?

The Employee Profile report was chosen because it's simple enough to follow, yet still relatable. It allowed us to focus on the full process—connecting to SQL Server, organizing the DB schema with help of Gemini 3 Pro, and building the report in ReportBurster.

Of course, you're not limited to employee profiles. With ReportBurster you can generate invoices, statements, or any business document, and then deliver them directly to the right people via email, web, or file sharing.

Closing Thought

In just one session, we connected to SQL Server, organized the schema with help from Gemini 3, generated queries and designed templates also with AI help, and produced a working report. ReportBurster's reporting engine, combined with AI assistance, makes the process fast, structured, and effective.

If you've followed along, you probably feel the urge to download ReportBurster and try it yourself.

When you install it, you'll find a built-in sample database ready to use, so you can immediately experiment and learn.

And once you're finished with the Northwind SQL Server sample, don't forget to stop its Docker containers to free up resources on your machine.

You might also wonder about the actual cost of setting up the Employee Profile report with Gemini 3 Pro.

Note: Gemini 3 Pro was accessed through https://openrouter.ai which provides easy access to multiple AI models from a single place.

The total cost was surprisingly low. The largest expense—$0.318—came from the initial prompt to generate the Domain-Grouped Database Schema. That's because we had to feed the AI the entire raw schema. Importantly, this is a one-time cost per database connection.

The cost of generating the Employee Profile report itself was only about $0.15.

Even better, similar results can be achieved with free AI models that don't require an account.

For example, https://copilot.microsoft.com/ works well.

It does have a limit of 10,240 characters per input query, but aside from that it's a strong option.

In a future post, we'll explore in detail how to use ReportBurster with free AI providers.

Transform Your SQL Server Data Into Actionable Reports

While we've used the Northwind sample database to demonstrate the workflow, you can now apply these same steps to your own SQL Server databases. Imagine automatically generating employee payslips with personalized filenames, or creating customer invoices that are immediately ready for distribution. As a reliable enterprise solution for report generation, automation and distribution, ReportBurster seamlessly bridges the gap between your SQL Server data and documents your business needs.

The real power emerges when you start automating your own document workflows. Whether it's HR systems for payslips, ERP systems for invoices, or CRM systems for customer statements, ReportBurster handles the complete workflow from data extraction to personalized document delivery. This automation dramatically accelerates cash flow for invoices and streamlines your document distribution processes.

Ready to transform how your organization handles reporting? Download ReportBurster and start building your automated report workflows today.