Use Oracle data next to Microsoft AI

Microsoft’s Oracle-on-Azure team said on April 14 that companies already running Oracle databases beside Azure services should start wiring that data into artificial intelligence tools now, not treat artificial intelligence as a separate pilot. (techcommunity.microsoft.com) The post came from Microsoft’s Oracle on Azure blog and described six “agentic” patterns, including plain-language querying, workflow automation, and retrieval systems that pull live answers from Oracle data instead of copied snapshots. Microsoft said those patterns are already being deployed by enterprise customers. (techcommunity.microsoft.com) The pitch depends on where the systems sit. Microsoft said Oracle AI Database@Azure and Microsoft artificial intelligence services can run in the same datacenter with sub-millisecond latency, which is the delay between a request and a response measured in thousandths of a second. (techcommunity.microsoft.com) Oracle Database@Azure is the joint service behind that message. Microsoft says customers can provision Oracle database services through the Azure portal and Azure application programming interfaces, while Oracle manages the underlying Oracle Cloud Infrastructure operations. (learn.microsoft.com) Oracle has framed the arrangement as a multicloud product, not a migration away from Oracle. Oracle says Azure customers can procure, deploy, and use Oracle database services running on Oracle Cloud Infrastructure from the native Azure portal and apply those purchases through Azure commitments. (oracle.com; blogs.oracle.com) The practical issue is that many companies keep their most valuable records in Oracle systems used for finance, supply chains, and customer operations. Microsoft’s argument is that if those databases already sit next to Azure services, the hard part is connecting the data safely and in real time, not starting a new artificial intelligence program from scratch. (techcommunity.microsoft.com; learn.microsoft.com) Microsoft has been adding the plumbing for that approach across its data stack. Microsoft Fabric, the company’s analytics platform, supports mirroring Oracle databases into OneLake in near real time, including Oracle Database@Azure, Oracle Cloud Infrastructure, Exadata, and on-premises Oracle versions 10 and above with LogMiner enabled. (learn.microsoft.com; learn.microsoft.com) Microsoft also published Oracle connectors for Fabric Data Factory, which lets teams set up Oracle database connections inside data pipelines and dataflows. That gives customers a second route: move selected Oracle data into analytics systems, or leave it in place and query it where it already runs. (learn.microsoft.com; learn.microsoft.com) The sales push lands as Microsoft and Oracle broaden the product itself. Microsoft’s Oracle-on-Azure blog said in March that Oracle Database@Azure had expanded and was renamed Oracle AI Database@Azure, and that Oracle Autonomous AI Database on Dedicated Exadata Infrastructure became generally available on the service. (techcommunity.microsoft.com) Microsoft is also trying to answer a common customer concern: what happens to the data once a model sees it. Microsoft says Azure OpenAI does not use customer data to retrain models, and its data privacy documentation says Azure Direct Models process and store data to provide the service and monitor abuse under Microsoft’s product terms. (learn.microsoft.com; learn.microsoft.com) The message in this week’s post was narrower than a product launch. For companies that already run Oracle on Azure and already pay for Microsoft Foundry or Azure OpenAI, Microsoft said the next step is to treat Oracle data as a live source for agents and analytics, not as a system that has to be copied out before artificial intelligence can use it. (techcommunity.microsoft.com; learn.microsoft.com)