If you haven’t already, check out Arun Ulag’s hero blog “Microsoft Build 2026: Building Agentic Apps with Microsoft Fabric and Microsoft Databases” for a complete look at all of our Microsoft Build announcements across our Fabric and database offerings.

_______________________________

Anna Hoffman, Principal Group Product Manager, Microsoft SQL

Whether you start with code or start with data, the SQL developer workflow has been predictable: design a schema, write queries, tune performance, deploy. Maybe you’re wiring up an ORM and building app layers on top of your database. Maybe you’re writing T-SQL directly to explore datasets and surface insights. Either way, AI changed the game. AI shifted the complexity in database development. Developers still connect tools, manage context, and translate intent into execution. That breaks down at scale.

At Microsoft Build 2026, that model changes. AI doesn’t just assist, it executes. Multi-step reasoning. Cross-object awareness. End-to-end workflows from schema design to production API with ORM awareness, orchestrated by GitHub Copilot, validated by you. This is what we mean by the agentic database developer: you set the intent, AI executes the workflow, and you stay in control of what ships.

And it’s not just about your tools and workflows, but also the database and infrastructure that supports them. The best infrastructure is the kind you don’t worry about. Azure SQL Database Hyperscale is the database for the agentic database developer. Azure SQL Database Hyperscale’s cloud-native architecture provides independently scalable compute and storage to support the widest variety of traditional and modern applications. Hyperscale has support for vector and vector indexes, the ability to talk to any model, and extensive enterprise capabilities. And a reminder: Hyperscale is license-free.

Build apps with Hyperscale and the MSSQL Extension for VS Code

The VS Code SQL experience now covers the full developer inner loop, from empty project to running endpoint, without switching tools.

The MSSQL extension in VS Code is releasing two major integrations (Generally Available). Schema Designer lets you design and evolve database schemas using natural language directly within a visual canvas. Describe what you need, and GitHub Copilot generates tables, columns, data types, and relationships in real time, with the changes available as T-SQL or migration-ready ORM code for Prisma, Sequelize, TypeORM, Drizzle, SQLAlchemy, and Entity Framework Core. You can also feed in JSON files, documents, or even images and let GitHub Copilot generate matching schema elements, which are especially useful for migrations and reverse engineering. Every change gets Accept/Undo controls and a consolidated diff view before anything touches your database.

Then Data API builder integration takes that schema and generates fully configured REST, GraphQL, and MCP endpoints (now Generally Available as well) with granular role-based access control per entity. Deploy locally with one-click Docker containerization, and test with the built-in Swagger UI or GraphQL playground inside VS Code's built-in browser so you can iterate without leaving the editor. You can also configure entities via GitHub Copilot chat: “Add all HR tables with read-only access” and it handles the rest.

SQL Notebooks (Generally Available) bring Jupyter .ipynb support directly into VS Code. Interactive SQL execution with rich result grids, IntelliSense from your active connection, markdown documentation cells, and multi-kernel support so you can combine SQL and Python in the same notebook for end-to-end data workflows.

And if you don’t have a database yet? Azure SQL Database provisioning (Preview) lets you spin up an instance directly from VS Code with free options—without ever having to leave the editor.

Figure: The provisioning experience now includes Azure SQL Database as an option.

The whole flow: idea → schema → API → production database, all in one tool, all AI-assisted.

SSMS goes Agentic

GitHub Copilot Agent Mode in SQL Server Management Studio (Preview) brings multi-step reasoning to SSMS. This isn’t autocomplete. Agent Mode reads context across your database objects, reasons through multi-step problems, and executes workflows. Ask it to find a performance bottleneck. It reads the execution plan, identifies missing indexes, evaluates the trade-offs, and generates the DDL. Ask it to refactor a stored procedure. It traces dependencies, rewrites the logic, and presents the difference.

Figure: An example of GitHub Copilot Agent Mode in SSMS, showing how an initial prompt can be in Agent mode and reference other tools, and then showing a preview of the results..

But agentic capabilities need disciplined tooling to match:

• SQL Formatter (Preview) applies consistent formatting at keystroke speed. Crucial when AI-generated code needs to meet your team’s style standards.

• Schema Compare (Preview) brings drift detection into your CI/CD workflow. Compare, diff, and generate migration scripts directly from your database project.

SSMS is now the tool where Agent Mode handles the heavy cognitive lifting, and formatter + schema compare + SQL Projects provide the deployment discipline.

We recently shipped a What’s New page in SSMS.

Learn the latest in the Release notes for SQL Server Management Studio (SSMS).

Develop securely from the start

Agentic workflows move faster. That means your security posture needs to be higher from the start, not bolted on after the fact.

Microsoft Entra server-level logins and fixed server roles (Generally Available). Entra-backed server principals now offer parity with SQL authentication—without passwords to manage or rotate. Fixed server roles let you delegate server-wide metadata access and selected management tasks without granting full admin rights, supporting least privilege and separation of duties. If you’re still using SQL authentication in 2026, this is your path to a modern identity model without the operational overhead of managing credentials.

Transparent Data Encryption (TDE) in Azure SQL Database with AES key support (Preview) expands customer-managed TDE to support AES keys. Previously, customer‑managed TDE relied on RSA keys, powerful today, but increasingly vulnerable in a post‑quantum world. While quantum computing evolves, this could dramatically reduce the effort required to break RSA keys.

To stay ahead, we’re introducing support for symmetric AES keys, which are considered quantum resistant and better suited for high‑performance data encryption.

This move aligns with Microsoft’s security guidance and the NSA’s CNSA 2.0 recommendations to transition away from RSA for key wrapping. With AES keys, Azure SQL helps ensure your data remains protected, today and as cryptography evolves.

Regex-powered Dynamic Data Masking (Preview coming soon) gives you pattern-based control over what gets masked and how. Define masking rules using regular expressions for more precision than the existing default/partial/email presets, especially for variable length data.

Enterprise-grade, developer-ready

Local time zone support (Preview coming soon) lands for Hyperscale (as well as Azure SQL Managed Instance, SQL DB in Fabric). Set your database time zone and stop doing UTC conversion math in every query. Finally.

Change Event Streaming (CES) (Preview) across Hyperscale, SQL Managed Instance. Support for SQL database in Fabric is coming soon. CES brings near real-time streaming of row-level inserts, updates, and deletes directly into Azure Event Hubs as CloudEvents (JSON or Avro). Build event-driven architectures, sync data across microservices, or feed real-time analytics without custom CDC pipelines or polling.

From data to apps with Fabric Apps

Fabric Apps (Preview) extends the foundation from data to applications. With native integration with SQL database in Fabric, Fabric Apps let teams build and run end-to-end applications without standing up any infrastructure. For enterprises, this means faster time to build, fewer moving parts to operate, and a more predictable cost and security model. All aligned with the same platform primitives you already trust.

What next: level up

Get certified

The new SQL AI Developer certification validates the skills at the center of this post: agentic workflows, AI-assisted development, modern SQL tooling. If you’re building with these tools, prove it. Learn about what’s new and best practices from our YouTube channel.

Join us at SQLCon Barcelona, September 2026

Sessions, hands-on labs, and the SQL community in person. Deep dives on everything announced here: Agent Mode, App Quickstarts, Hyperscale, and more. Register now.

Check out the roadmaps for SSMS and the MSSQL extension for VS Code, we have a lot planned for both to help you achieve more with AI and our tools.

The agentic database developer era is here. The tools are shipping and the products are evolving accordingly. Go build!