Microsoft Agent Framework, now in public view, and new skills in Azure AI Foundry facilitate building, observation and management systems.
Today we announce new abilities in Azure AI Foundry This makes it easier for developers to manage, observe and more agents, while helping organizations to close a gap in confidence in AI.
As Adoption agent agent AI accelerates –Eight out of ten companies now use some form of AI -based agentsAccording to PWC1– The complexity of driving these systems is growing. Developers face fragmented instruments and organizations seek to ensure that agents behave responsibly. Our latest Azure AI Foundry updates are engaged in these challenges.
Agent Framework Presentation (Public Preview)
Tea Microsoft AgentNow in public insight, there is an open-source SDK and Runtime, which simplify the orchestration of more agents’ systems. Converge AutogenFormer Research project Microsoft and Foundation ready for business Semantic nucleus It brings top research to the united commercial framework to the outcomes.
With Agent’s Microsoft framework, developers can:
- Experiment locally and then put in the Azure AI foundry with the built -in observability, durability and matches.
- Integrate any API via OpenAPI, cooperate across the Runtimes with the agent (A2A) and dynamically connect to the tools using a model context protocol (MCP).
- Use the latest patterns with multiple agents such as Magentic One and Orchestrate Agans to workflows.
- Reduce context switching across tools and platforms.
- Build multi-agent systems connecting the Azure AI foundry, Microsoft 365 Copilot and other agents platforms.
This framework is designed to help developers remain in the flow. According to an industrial study2, 50% of developers lose more than 10 hours per week Due to inefficiency, such as fragmented tools, they emphasize the need for a solution that reduces complexity and improves developers.
One organization using the Microsoft Framework to Reduce friction is Kpmg. The KPMG transformation began KPMG Clara, its cloud platform of an intelligent audit based on every KPMG audit around the world.
KPMG Clara AI is firmly aligned with the Microsoft agent framework with an open source code based on the convergence of semantic core and autogen.
This means that KPMG Clara AI can connect specialized agents to business data and tools, while benefiting from built -in guarantees and open, expandable ecosystem developer. Framework open-source connectors allow KPMG Clara AI agents not only to be intermed with the Azure AI foundry, but also with external systems and language models, allowing you to scalance with multiple agents across the global, regulated enterprise.
Foundry and Agent Framework Agent Foundry and Agent Framework combine our agents to data and to each other, and management and observability in Azure AI foundry provides what KPMG companies must be successful in a regulated industry.
– Sebastian Stöckle, Global Audit and AI at KPMG International
We invite developers to join us in forming the future AI AI by contributing by code and feedback to Microsoft Agent Framework.
Workflows with multiple agents (private preview)
Building on Agent Microsoft, we are expanding these abilities directly into the cloud with MULTI-AGENT WORKFLOWS IN FOUNDRY Agent Service. This new function allows developers to organize sophisticated multi -stage business processes using a structured, status layer of workflow.
With more agents working procedures can teams:
- Coordinate more agents across long -term tasks with persistent status and context sharing.
- Automating comprehensive enterprise scenarios, such as customers, financial transactions and supply chain automation.
- Take advantage of the built -in errors, repetition and recovery to improve the scale reliability.
Workflows can be the author and visually tuned through the expansion of the VS code or foundry of the Azure AI, then deployed, tested and managed in founders along with existing solutions.
Several customers are currently experimenting with this ability and looking forward to expanding more customers in the coming weeks.
Observability with multiple agents across popular framework with ophememetric posts
We also indicate improvements Observability with multiple agentswith posts to Ophememetry This helps standardize monitoring and telemetry for agent systems.
This gives teams deeper visibility of the agent’s workflows, induction of instruments and cooperation – critical for tuning, optimization and compliance. We made the above improvements in cooperation with Outshift, the Outshift incubation engine.
With the above improvements, the Azure AI foundry provides united observability for agents created with multiple frameworkIncluding Microsoft Agent Framework and others, such as Langchain, Langgraph and Openai SDK agents.
Voice Live API in Azure AI Foundry is now generally available
Mutli-agent workflows are increasingly initiated through voice inputs and culmination of voice outputs. We are pleased to announce the general availability of the Voice Live, which seizes developers and businesses to build scalable AI agents ready for production. Voice Live API is a united interface of speech in real time Speech to the text (STT), Generative models AI, Text-to-Speech (TTS), Avatarand conversational improvements Function into a single low latency pipe.
Organization as Capgemini, Healiw, Astra TechandAgora They use voice live APIs to build customer service agents, educational lecturers, personnel assistants and multilingual agents.VOICE LIVE API It transforms how developers build voice AI agents by providing an integrated, scalable and efficient solution.
Responsible abilities of AI public insight
Based on progress in the agent and integration of the framework, it is equally important to ensure that AI systems work responsibly and safely – especially because they become deeper rooted into critical business procedures.
According to McKinsey’s survey Global AI Trust from 20253teaNumber one barrier in adoption AI is a lack of tool management and control tools. Therefore, in the coming weeks we present the following responsible AI functions in public preview:
These abilities are built into the Azure AI foundry that help organizations with confidence and adhere to internal and external standards.
Momentum of the customer
Azure Ai Foundry’s solution helps70,000 organizations worldwide– from digital natives to corporate companies – transform AI innovations into practical results. For example:
Commerzbank: Commerzbank The Microsoft Agent Pilot Framework to power the Avatar Customer Support, which allows for more natural, more affordable and more comprehensive interactions with customers.
The new Microsoft Framework Agent Framework simplifies coding, reduces efforts and fully supports MCP for agent solutions. We are really looking forward to the productive use of Ai AI foundry agents based on containers, which significantly reduces workload in IT operations.
– Gerald Ertl, CEO/Digital Banking Solution, Commerzbank AG
Citrix: Citrix examines how they can use AI AI in virtual desktop infrastructure (VDI) to improve the productivity and efficiency of the company.
Citrix has always accepted flexible ways of working as a leader in safe work. If we move to the world where AI side side by side is working with us, we are excited that we can also do it in the workspaces that our customers already use every day. Microsoft’s framework brings a modern, development approach to building agents. With the support of key APIs and languages and native acceptance of newly emerging protocols for calling and visibility of tools, the intuitive development of agents on the Azure AI founders allows the developers to control. We are eager to use the framework for providing our vision-up-to-line and AI agents ready for production for our customers.
—Gorge tsolis, distinguished engineer, citrix
TCS: Tata Consultancy Services actively creates more agency practice in the technical framework of Microsoft Agent, with several initiatives that show their strategic investment and depth, including agency solutions for finance, IT operations and retail.
The acceptance of the Microsoft Agent framework is not only technological advances, but a bold step towards reimagining the value chains of industry. By using AI and Frontier AI and Frontier models, we allow our teams to create flexible, scalable corporate level solutions that transform workflows and give value across platforms. The actual management is about authorizing innovation, accepting changes and support of the environment where agility and cooperation lead to perfection.
—Girish Phadke, Head, Microsoft Azure Practice, TCS
Sitecore: Sitecore is developing a solution that allows traders to interact smoothly with the platform of automation of tasks throughout the content chain of content – from creating and managing web experiences to handling digital assets – using intelligent agents.
Working with Microsoft to use its new Microsoft Framework agent can combine Sitecore the best of both worlds: flexibility to power fully nodeterminist agency orchestrations and reliability for launching determinist and repeatable substances. At the same time, we benefit from Microsoft’s observability and telemetry, which ensures that these orchestrations are not only strong but also safe, measurable and ready for production.
– mo cherif, vp ai, sitecore
Elastic: Elasticsearch supports the native connector to Microsoft Agent and allows developers to smoothly integrate business data into intelligent agents and workflows.
Elasticsearch is a context -engineering platform and a vector database for organizations that organizations store and look for their most valuable operating data. With the new agent Microsoft Agent Agent, developers can now bring the most important organizational context directly to intelligent agents and more agents. This makes it easier than ever to create AI solutions ready for production that combines a thorough power of agents with the speed and range of elasticsearch.
– Steve Kearns, Solution of Search CEO, Elastic
A trusted agent for developers
The Azure AI foundry is more than the platform – it is aA trusted agent factory for developers and businesses. Whether you are a CIO who is trying to responsibly the AI scale, the security architect focused on the management of public affairs or the developer who has built the next generation of intelligent agents, Azure AI Foundry provides the tools, framework and trust you need.
Microsoft stands out in the AI landscape with a commitment to open standards, interoperability and responsible AI. Microsoft’s framework, now in public preview, is a unified business level that integrates top research and allows developers to smoothly organize more agent systems with built -in observability, durability and compliance.
Unlike other solutions, our framework supports integration with any API via OpenAPI, cooperation across AGET2AGENT (A2A) and dynamic tools using MCP. This ensures that developers can reduce context switching and remain in the flow and speed up innovation.
Open-Source The nature of the framework invites developers to contribute and shape the future of agent artificial intelligence, making it truly cooperation and forward. With Microsoft, organizations may believe that their AI systems will be powerful, efficient, responsible and safe, which will solve the highest AI adoption barriers identified in Global AI Trust in McKinsey 2025.
1 Agent survey AI PWC.
2 AII adoption is growing, but friction persists.
3 Knowledge about the responsible AI from the Global AI Trust Resting survey.