Microsoft announced Azure AI Foundry innovations on March 31st, 2025, designed to optimise AI investments and enable business differentiation in competitive landscapes. Corporate Vice President Jessica Hawk highlighted new agentic capabilities, NVIDIA integrations, and performance enhancements addressing the rapid transformation from AI excitement to strategic implementation using cutting-edge platform capabilities.
New knowledge tools with Azure AI Agent Service securely ground AI agent outputs with enterprise knowledge for accurate, relevant, and contextually aware responses across unstructured, structured, private, licensed, and public web data. Microsoft Fabric data agents, announced at the Microsoft Fabric Community Conference, enable developers using Azure AI Agent Service to connect customised conversational agents created in Microsoft Fabric for reasoning over enterprise structured and semantic data sources.
Fabric data agents retrieve, understand, and synthesise data from OneLake, determining specific data usage and combination methods for better data-driven decisions. The Fabric-Foundry pathway connects data teams with development teams on a common, secure, enterprise-ready AI platform combining Microsoft Fabric's enterprise data analysis capabilities with Azure AI Foundry's generative AI technology.
NTT DATA leverages data agents in Microsoft Fabric for conversations with HR and back office operations data to understand organisational dynamics. The Responses API enables AI-powered applications to seamlessly retrieve information, process data, and act by simplifying complex tasks for more efficient operations and cost reduction.
Computer-using agent represents a breakthrough AI model navigating software interfaces, executing tasks, and automating workflows including opening applications, clicking buttons, filling forms, and navigating multi-page workflows. CUA adapts dynamically to changes for smooth operations across web and desktop applications, integrating disparate systems without API dependencies.
Azure AI Foundry added NVIDIA NIM microservices and NVIDIA AgentIQ toolkit for efficiency, performance, and cost optimisation. NVIDIA NIM provides zero-configuration deployment, seamless Azure integration with Azure AI Agent Service and Semantic Kernel, enterprise-grade reliability through NVIDIA AI Enterprise support, scalable inference on Azure's NVIDIA accelerated infrastructure, and optimised workflows for large language models to advanced analytics.
Azure OpenAI Service Provisioned spillover, entering public preview, automatically reroutes excess traffic from provisioned deployments to maintain smooth service operation during high usage periods and unexpected traffic bursts. The feature provides flexibility for peak demand management without performance compromise, enabling adaptation to dynamic conditions while maximising AI investments.
Microsoft released "DIY GenAI: Customising generative AI for unique value" report detailing how businesses use fine-tuning, retrieval-augmented generation, and agentic specialisation for differentiation from 300 technical leaders. The findings indicate customisation capabilities create distinctive experiences helping businesses stand out through unique business data and tailored AI models. Accessibility and inclusivity integration in Azure AI Foundry development lifecycle expands reach, boosts customer satisfaction, and enhances social responsibility reputation for long-term success.