No more excuses: AI-powered assistants are in SSMS, VS Code, and Fabric
You like writing T-SQL. You’re good at it. Or maybe you’re not. But let’s be honest—there are days when you’d rather not write that same GROUP BY clause for the hundredth time or spend twenty minutes deciphering a stored procedure someone wrote in 2014.
Good news: AI-powered assistants are now available wherever you write SQL for every deployment option (SQL Server, Azure SQL, SQL database in Fabric). Whether you’re in SQL Server Management Studio (SSMS), Visual Studio Code, or the Fabric portal, there’s an AI assistant ready to help based on the context of your database(s). And it’s not just autocompleted—it’s chat, code actions, query optimization, and more.
Here’s what you can do.
GitHub Copilot in SSMS
SSMS 22 introduced GitHub Copilot as a preview feature, and it keeps getting better. You get inline code completions as you type, helping you write T-SQL faster without breaking your flow.
But the real power is in the chat window, editor, and in menus. You can have a full conversation with GitHub Copilot in the chat or let it complete your code and queries—ask it to generate queries, explain code, or help you understand what’s happening in your database.
Got a stored procedure you inherited and can’t make sense of? Use GitHub Copilot. Need to add documentation before you forget what your own code does? Use GitHub Copilot. Query running slower than it should. Ask GitHub Copilot to optimize it—you can even attach an execution plan to your chat and ask, “why is this query slow?” GitHub Copilot reads the plan and gives you actionable recommendations.
It’s the kind of help that used to require pinging a senior DBA or digging through Stack Overflow. Now it’s a prompt away.
Watch the demos in the GitHub Copilot in SSMS playlist to see it in action.
GitHub Copilot in the MSSQL Extension for VS Code
The MSSQL extension first introduced GitHub Copilot through chat and inline assistance, grounding suggestions in the connected database. It then expanded into Agent Mode, enabling multistep, tool-driven database workflows. Slash commands and deeper integrations followed, making common tasks faster, more structured, and more predictable.
With agent mode, GitHub Copilot can handle routine database operations while you focus on building your application. Need to scaffold a schema? Generate seed data? Debug a query that’s not returning what you expect? GitHub Copilot can do it—connecting to your database, running queries, and returning results directly in chat. Every action requires your approval, so you stay in control without getting buried in the details.
For developers who know their app but don’t want to stress about the database backend, this is the point. You describe what you need, GitHub Copilot figures out how to get it done.
Explore the features in the MSSQL extension GitHub Copilot demo playlist.
Yes, SQL database in Fabric too
GitHub Copilot in SSMS and VS Code work with SQL database in Fabric. Connect to your Fabric SQL database from either tool and you get the same Copilot capabilities—chat, code actions, query optimization, all of it.
If you prefer to stay in the Fabric portal, the built-in Copilot in the Query Editor has you covered. It’s a separate experience from GitHub Copilot, but it uses the same underlying tools that power the GitHub Copilot integrations in VS Code and SSMS. You’re getting consistent, high-quality AI assistance whether you’re in your desktop tools or the browser.
One set of capabilities, multiple ways to access them. By the way, Azure Copilot also works with your Azure SQL resources in the Azure portal.
Common myths, busted
A few questions come up often. Let’s address them:
“Does Copilot use my prompts to train models?”
No. This is clearly stated in the documentation. Prompts and responses are not used for training.
“Can Copilot run destructive commands against my database?”
Not without your explicit approval. In SSMS, Copilot currently operates in Ask mode—all queries it executes are read-only SELECT statements. Agent mode (coming soon) will allow write operations, but every action requires your confirmation. In VS Code, Agent mode already exists with the same approval requirement, and in SQL database in Fabric, though not full agent mode, write with approval is supported.
“Are my prompts encrypted?”
Yes. Visit the GitHub Copilot Trust Center for details on privacy and security.
Get started
Ready to try it?
For SSMS, download SQL Server Management Studio 22 and enable the GitHub Copilot workload in the Visual Studio Installer.
For VS Code, install the MSSQL extension and sign in with your GitHub Copilot subscription.
For Fabric, connect to your SQL database in Fabric from SSMS or VS Code, or use Copilot directly in the Fabric portal Query Editor.
For deep dives and hands-on learning, join us at SQLCon, co-located with FabCon Atlanta, March 16–20, 2026. And subscribe to the Azure SQL YouTube channel for weekly demos, walkthroughs, and community content.
Catch all videos from the team by subscribing to the Azure SQL YouTube channel
