Blog Post

Azure SQL Blog
11 MIN READ

What's new in databases at Microsoft Ignite 2026: AI-ready data for apps and agents

Shireesh_Thota's avatar
Nov 17, 2026

This week we’re sharing how Microsoft databases help organizations modernize the data that runs their business, build the next generation of intelligent applications, and operate it all with the help of agents.

By Shireesh Thota, Corporate Vice President, Azure Databases and Fabric Applications, Microsoft

Organizations putting agents to work are learning quickly that an agent is only as good as the business context behind it. That context starts with trusted business data. Every transaction, customer interaction and operational event tells AI what is happening right now, and much of it lives in operational databases. Microsoft IQ helps bring that context together for agents. As AI becomes more capable, the database becomes even more important. 

With Azure databases and Microsoft Fabric, we're making that data estate work as one: unified, rich with business context and governed at the source. But that data is spread across today's business applications, while new intelligent apps demand more scale, speed, and context. Organizations need to do three things at once: 

  • Build what comes next on cloud-native databases designed for scale, analytics, and AI. 
  • Modernize the trusted data they already have onto AI-ready foundations. 
  • Operate a growing estate with the help of agents. 

Here's how our top announcements at Microsoft Ignite help with each. 

Build on one PostgreSQL foundation, from Azure to Fabric 

At Ignite last year, we introduced Azure HorizonDB in preview. Today, Azure HorizonDB is generally available: a fully managed, PostgreSQL-compatible database service built for mission-critical and AI-powered applications.  

With Azure HorizonDB, teams can build AI applications directly on transactional data using built-in vector search, AI Functions and AI Pipelines that invoke models from SQL queries, along with built-in AI model management in preview. With the PostgreSQL extension for VS Code, developers can manage, tune, and monitor their databases with AI assistance in the same editor where they write their application code. 

For critical workloads, multi-zone high availability and distributed storage support fast recovery to help keep applications running. Microsoft Entra ID, private endpoints, managed identities, and service- or customer-managed keys help protect sensitive data, bringing resilience and security together with HorizonDB’s AI capabilities. 

"Our AI ambitions depend on secure, reliable, high-performance API access to LSEG’s data. Azure HorizonDB meets our enterprise security requirements while providing a PostgreSQL foundation for our existing API services and the MCP layer powering new AI experiences.” 

Grzegorz Giedrojc, Director of Engineering, LSEG (London Stock Exchange Group) 

Serverless PostgreSQL in Fabric, from first query to OneLake 

In Microsoft Fabric, we're bringing the HorizonDB foundation closer to analytics with serverless PostgreSQL built for applications and agents. As AI accelerates development, databases need to start just as quickly and handle bursty workloads. Microsoft Fabric HorizonDB provisions databases in seconds, pauses after five minutes of inactivity and resumes with the next query, giving developers and agents PostgreSQL that scales with how they build, without infrastructure to manage. Fabric HorizonDB is also available as a PostgreSQL database for Fabric apps. With the Rayfin SDK, PostgreSQL joins data APIs, authentication, storage, and hosting in one managed backend, so builders can focus on application logic. Beyond serverless operations, Fabric HorizonDB supports translytical capabilities, combining transactions and analytics, on OneLake, bringing the scale and performance of Fabric’s analytical engines to applications working with live transactional data: 

  • Run translytical queries. PostgreSQL queries are transparently pushed down to Fabric Data Warehouse for distributed CPU- or GPU-powered execution. Full and partial pushdown allow a single query to join PostgreSQL tables with data in OneLake, with analytical queries running an order of magnitude faster than they would natively in PostgreSQL. 
  • Bring operational data into analytics automatically. Mirroring continuously replicates HorizonDB tables into Fabric for large-scale analytics and enrichment without separate data integration pipelines. 
  • Put enriched data back into applications. Enriched data from Fabric Data Warehouse flows back into HorizonDB synced tables, where applications and agents can access it with low-latency, high-concurrency performance. 

The result is a continuous loop. Application data flows into OneLake, analytics enrich it, and insights flow back to the applications and agents that act on them, all within Fabric. 

I'm especially excited to bring HorizonDB to both Azure and Microsoft Fabric, giving teams one PostgreSQL foundation for mission-critical apps and AI-native development. For the full story, read what’s new in HorizonDB. 

Modernize with a plan, accelerated by agents 

The applications that run your business hold your most valuable context, and modernizing them onto AI-ready foundations, with less migration friction, makes that context available to new intelligent experiences. Agents are changing how that work gets done, from discovery to code conversion.  

Every modernization starts with a plan. Azure Catalyst (formerly Azure Migrate), introduced at Ignite, brings IT teams and developers together around one plan for applications and databases, from discovery and assessment through migration and modernization. Agents help identify what's ready to move, where changes are needed and the right Azure destination for each workload, and GitHub Copilot modernization carries that plan into the code.  

“Transforming a legacy mainframe archive into a modern cloud solution is not just a technical upgrade; it's an opportunity to innovate, streamline, and future-proof your environment. Embrace the challenge, and you'll unlock new efficiencies and capabilities that were once unimaginable.”  

- Daniel Tanner, Stream Technical Lead, Client Document & Records Management (CDRM), UBS 

AI-assisted migrations from Oracle to managed PostgreSQL on Azure 

Some workloads ask more of that plan than others. For many organizations, Oracle workloads can be the hardest to move, because migrating the data is only half the job. The application code that depends on it must move, too. Two new previews address both halves. 

Oracle to PostgreSQL migration in Azure Database Migration Service now supports offline data migrations to Azure Database for PostgreSQL. Teams can configure connections, select schemas, and start and track the transfer in one experience, without building custom data-movement processes. 

For application code, AI-assisted Oracle application conversion provides a plug-in that converts Oracle code to work with Azure Database for PostgreSQL. Available through GitHub Copilot CLI or Visual Studio Code, and directly within existing modernization workflows such as Modernize CLI, the skills identify Oracle-specific dependencies and propose code changes for PostgreSQL compatibility. Developers review and validate the changes before accepting them. 

Together, they bring agentic assistance to the whole migration, not just the data. There's much more to the PostgreSQL story at Ignite – check it out.  

Run your apps where your data lives with DocumentDB 

Last year, we introduced Azure DocumentDB, our MongoDB-compatible document database powered by the open-source DocumentDB engine governed by the Linux Foundation. Now we’re expanding how organizations can deploy DocumentDB across hybrid and multicloud environments. DocumentDB enabled by Azure Arc and DocumentDB on Azure Local, both in preview, let organizations run MongoDB compatible workloads on supported Kubernetes clusters or customer-owned infrastructure, including on-premises and edge environments. Learn more about these new announcements here. 

Build intelligent apps faster 

Cloud-native applications need cloud-native data services designed for elastic scale, resilience, and security, with diverse data models and direct connections to analytics and AI. AI is raising the bar on all of it as apps and agents are built faster, grow less predictably and need context in real time. Our news this week meets developers where they build, from deploying complete apps in Fabric to scaling SQL, giving agents memory and serving AI experiences in at speed. 

Build and deploy intelligent apps with Fabric apps 

Fabric apps are now generally available, giving developers and coding agents a simpler path to building and deploying complete business applications. They build with Rayfin, the open-source SDK and CLI that provides databases, authentication, functions, storage, and observability, and deploy to Fabric, connected to enterprise data and governance. In just five months of preview, builders created more than 100,000 apps with Rayfin and deployed them on Fabric. Fabric apps connect to existing semantic models, warehouses, and lakehouses, so teams can build on trusted data without recreating data models or access controls. Teams can also turn analytics into applications that move users from insight to action. Content Security Policy settings and private sites add more protection. 

Builders can also start in familiar tools. With Microsoft Copilot, they can go from a simple prompt to a working application, using Fabric as a ready-made backend for databases, storage, and data APIs. Integrations with Power BI Pro and Power BI Desktop, both in preview, help teams turn trusted semantic models into interactive business applications.  

At Thomson Reuters, the Monthly Business Review now runs as a Fabric app, built on the same certified Power BI semantic models that ground the company's DeepVista AI framework. 

"This is a breakthrough in how Finance works. Certified semantic models give us numbers we can trust. Thomson Reuters DeepVista's AI agents let analysts and leaders ask questions in natural language and get insights they can act on. The monthly business hub solution automates the business review, so we spend our time on what to do next."  

Michael Goddard, Head of Corporate Finance, Thomson Reuters

Fabric apps are just one part of a big week for Microsoft Fabric. Catch up on all the Fabric news from Ignite to see what else is new. 

Build locally, scale confidently with Azure SQL Database Hyperscale 

The SQL applications organizations trust most are ready for their next chapter: cloud-native scale, better economics, and AI. Azure SQL Database Hyperscale runs the SQL engine teams already know, on a cloud native distributed architecture that separates compute from storage, enabling applications to scale with demand. With relational, JSON, graph and vector data unified under familiar SQL security and governance, developers can build AI powered experiences grounded in real-time business data without a separate vector database. That makes Hyperscale the destination for modernizing SQL applications and the foundation for new AI applications. 

Great applications start on a laptop, not in a provisioning queue. Azure SQL Database container, now in public preview, brings the SQL engine to local development without requiring an Azure subscription. Developers get familiar T-SQL, system views, and driver behavior, along with vector support and AI_GENERATE_EMBEDDINGS to prototype AI-powered applications. When ready to deploy, the new local-to-cloud deployment wizard in the MSSQL extension for VS Code, also in public preview, provides a streamlined path from the developer's laptop to Azure SQL Database Hyperscale. 

Growing applications often split data across databases. Cross-database querying for Hyperscale, now in limited preview, lets developers use familiar T-SQL three-part names to query across databases in the same elastic pool from one connection, while each database enforces its own permissions. From laptop to cloud scale, it's the same SQL the whole way.  

Ready to start building? Read more about these and other Azure SQL announcements at Ignite to see how teams are modernizing and building what’s next on Hyperscale. 

Modernize and build AI apps that remember 

Azure Cosmos DB powers some of the world's most demanding AI applications, and more organizations want to bring their existing applications onto the same foundation. With Relational-to-CosmosDB modernization assistant, now in preview, London Stock Exchange Group (LSEG) modernized several legacy Oracle databases of record, including real-time pricing and instrument data, to Azure Cosmos DB, improving the latency and scale at which LSEG serves that data to traders.  

“Thanks to Azure Cosmos DB scale and performance, we have centered on Cosmos DB as the database strategy going forward for our database of records, the authoritative datasets used for downstream services and products.” 

Claire Armstrong, Head of Engineering Instrument & Quote, LSEG Data & Analytics 

For new applications, the AI-assisted data modeler in Azure Cosmos DB extension for Visual Studio Code, now in preview, helps developers get the data model right from the start. And as these workloads become business-critical, Azure Backup for Azure Cosmos DB, now generally available, adds immutable backups and long-term retention in a vault isolated from the source account, a key requirement in regulated industries. 

Modernizing the data layer matters even more as organizations put agents to work, because an agent is only as useful as what it remembers. That memory lives in the database, and Cosmos DB is a natural home for it. To keep that memory relevant as it grows, automatic embedding generation for Azure Cosmos DB, now in public preview, lets developers generate vector embeddings right where their data lives.  

We are excited to announce Microsoft.Embedding-1, a new multimodal embedding model with single- and multi-vector support for more granular matching, developed by the Cosmos DB team in partnership with Microsoft Research. The model is available via the Cosmos DB embedding generation service, and can be used with and without Cosmos DB. At the time of this publication, the Microsoft.Embedding-1 tops the ViDoRe embedding models leaderboard.  

And semantic reranker, now generally available, reorders retrieved results by meaning and query intent, sharpening relevance for search, RAG, and agents without custom ranking pipelines. 

Agent Memory Toolkit takes advantage of these core database capabilities to provide end-to-end serverless agent memory solution for facts, episodic, and procedural memories with built-in memory workflows for summation, reconciliation, and promotion. 

Explore these and other Azure Cosmos DB announcements at Ignite to see how teams are modernizing and building smarter AI apps. 

Relevant context is only half the equation. AI experiences also need to respond in real time. That's where Azure Managed Redis comes in. Teams are moving beyond traditional caching to use Redis as a real-time data layer for AI-driven applications, serving the session state, cached results and vector lookups that keep experiences fast. Redis 8 is now available in Azure Managed Redis, bringing the latest Redis innovation to Azure's fully managed service. With double the throughput, richer queries, atomic JSON updates, field-level expiration and fine-grained security, Redis 8 simplifies development and helps you build sophisticated applications faster.  

Scale operations with confidence 

As estates grow and as agents create more applications and more databases, operations can’t scale on human effort alone. The next step is bringing agents into the database lifecycle itself, with people defining goals, boundaries and trust, and agents providing the scale. 

Gain visibility and control across your database estate 

Operations start with knowing what you have. Database Hub, now generally available, brings your entire database estate together in one place, with unified visibility, observability, and centralized governance. It gives teams a single control plane for the Microsoft databases portfolio, and a place where people and agents work side by side to understand the estate, prioritize what matters and act. 

Manage your databases more intelligently with AI 

Database agent, now in preview, brings AI-assisted operations into the tools database professionals and developers already use. Database agent for SQL combines native in-database intelligence with AI-powered reasoning to continuously monitor operational signals across the SQL estate, proactively detects issues, explains their impact, and recommends actionable remediation. Database agent in VS Code for PostgreSQL brings AI-powered reasoning into the developer’s editor and recommends remediation steps or executes approved actions through Copilot agent mode. 

Rather than replacing human judgment, database agent works within your existing identity, permissions, approval boundaries, and audit trail. It helps teams move from observing databases to intelligently managing them, from a single database in VS Code to the full estate in Database Hub. For database teams, that means less time chasing alerts and more time on the work that moves the business forward. Learn more about Database Hub, database agent, and other exciting news. 

Build what’s next without leaving what matters behind 

Every intelligent experience starts with data. The opportunity in front of every organization is to modernize the applications that run the business while creating what comes next, on one connected foundation. The organizations that lead with AI won’t be the ones that start over. They’ll be the ones that turn the trusted data they already have into context their agents can act on. With Azure databases and Microsoft Fabric shipped together, that’s exactly what we’re building. 

To keep going: 

Updated Nov 17, 2026
Version 1.0
No CommentsBe the first to comment