Microsoft has released its new Agent Framework into public preview, providing developers with a unified set of tools for building and managing complex AI agent systems. The release is part of a broader update to the Azure AI Foundry platform, which also includes new features for workflow orchestration, system monitoring, and responsible AI governance.
Key Takeaways
- Microsoft Agent Framework is now in public preview, combining AutoGen and Semantic Kernel into a single open-source SDK.
- New multi-agent workflow capabilities are available in private preview for orchestrating complex business processes in Azure AI Foundry.
- Voice Live API is now generally available, enabling real-time voice AI agent creation.
- Enhanced observability tools through OpenTelemetry and new responsible AI features have been introduced to improve trust and governance.
A Unified Framework for Agent Development
Microsoft has introduced the Microsoft Agent Framework, an open-source software development kit (SDK) and runtime now available in public preview. This new framework aims to simplify the creation and management of multi-agent AI systems by consolidating different toolsets into a single platform.
The framework integrates AutoGen, a project from Microsoft Research, with the enterprise-focused foundations of Semantic Kernel. This combination provides developers with a commercial-grade solution for building sophisticated AI agents that can collaborate on tasks.
Developer Productivity
An industry study indicates that 50% of developers lose more than 10 hours per week due to fragmented tools and context-switching. The unified nature of the Microsoft Agent Framework is designed to address this inefficiency.
Developers using the framework can build and test systems locally before deploying them to the Azure AI Foundry. The platform supports integration with any API through OpenAPI and allows agents to connect with various tools and runtimes using protocols like Agent2Agent (A2A) and Model Context Protocol (MCP).
New Capabilities in Azure AI Foundry
Alongside the new framework, Microsoft has announced several enhancements to its Azure AI Foundry platform, which is currently used by over 70,000 organizations. These updates are designed to provide more robust tools for managing AI systems in a cloud environment.
Multi-Agent Workflows
A new feature called multi-agent workflows is now in private preview within the Foundry Agent Service. This capability allows developers to orchestrate complex, multi-step business processes that require coordination between multiple AI agents. It provides a structured workflow layer that maintains state and context over long-running tasks.
This is particularly useful for enterprise scenarios such as automating customer onboarding, processing financial transactions, or managing supply chains. The system includes built-in features for error handling and recovery to ensure reliability.
Voice Live API Now Generally Available
Microsoft also announced the general availability of its Voice Live API. This tool offers a single interface for building real-time, voice-enabled AI agents. It combines speech-to-text, generative AI models, text-to-speech, and avatar features into one low-latency pipeline.
Companies including Capgemini, healow, and Astra Tech are already using the Voice Live API to create customer service agents, educational tools, and multilingual assistants.
Addressing Governance and Observability
As the use of agent-based AI grows, ensuring these systems are trustworthy and transparent has become a significant challenge. According to a PwC survey, eight out of ten enterprises are now using some form of agent-based AI.
The Trust Gap in AI
A 2025 Global AI Trust Survey by McKinsey identified the lack of governance and risk-management tools as the primary barrier to AI adoption for many organizations. Microsoft's new features directly target this concern.
To improve system transparency, Microsoft has contributed to the OpenTelemetry standard. These contributions help standardize tracing and telemetry for agentic systems, giving developers deeper insight into agent behavior, tool usage, and collaboration. This unified observability now supports agents built with various popular frameworks, including LangChain, LangGraph, and OpenAI's Agents SDK.
New Responsible AI Features
In the coming weeks, Microsoft will release several new responsible AI capabilities into public preview. These tools are integrated directly into Azure AI Foundry and are designed to help organizations manage risks and maintain compliance.
- Task adherence: Helps ensure agents remain focused on their assigned tasks.
- Prompt shields: Protects systems against prompt injection attacks and highlights risky inputs.
- PII detection: Identifies and helps manage personally identifiable information within the system.
Enterprise Adoption and Partner Integrations
Several major organizations are already utilizing the new Microsoft tools to build and deploy agentic AI solutions. KPMG, for example, is using the Microsoft Agent Framework to enhance its cloud-based audit platform, KPMG Clara AI.
"Foundry Agent Service and Microsoft Agent Framework connect our agents to data and each other, and the governance and observability in Azure AI Foundry provide what KPMG firms need to be successful in a regulated industry."
Other companies are also integrating these new capabilities into their operations. Commerzbank is piloting the framework for avatar-driven customer support, while Citrix is exploring its use within virtual desktop environments to boost productivity.
Tata Consultancy Services (TCS) is building a dedicated multi-agent practice on the framework for solutions in finance, IT operations, and retail. Additionally, companies like Sitecore and Elastic have announced integrations. Sitecore is using the framework to automate marketing tasks, and Elastic has released a native connector to allow developers to integrate enterprise data into their AI agents seamlessly.
"With the new Microsoft Agent Framework connector, developers can now bring the most relevant organizational context directly into intelligent agents and multi-agent workflows."
These updates position Azure AI Foundry as a comprehensive platform for developers and enterprises looking to build, manage, and scale AI agent systems responsibly.