Microsoft Foundry expands AI agent development

Introduction

Microsoft is positioning Foundry as a flexible foundation for enterprise AI agents that can evolve as models and workloads change. For Azure customers, this matters because it reduces the need to rebuild agent architectures every time a new model or interaction pattern becomes available.

What’s new in Microsoft Foundry

Broader model choice

Microsoft Foundry now includes additional frontier models, including the full GPT-6 family from OpenAI and Claude Opus 5.5 from Anthropic. This gives teams more options to compare models based on:

• Quality
• Latency
• Cost
• Workload fit

Instead of locking into one provider, organizations can evaluate models against their own data and production scenarios.

Voice agents in public preview

Voice agents are now available in public preview through Foundry Agent Service. Rather than layering speech on top of text-first agents, voice is built into the same platform, APIs, and SDKs used for other Foundry agents.

Key capabilities include:

• Support for prompt and hosted agents
• Deployment to web, Microsoft Teams, Teams Phone, and Twilio
• Support for 80+ languages and 140+ locales
• Custom speech, custom voices, and avatars for branded experiences
• Voice observability with tracing, transcripts, and evaluation

Long-running hosted agents

Foundry Agent Service also adds long-running resilience for hosted agents, now in public preview. This allows workflows to continue after a request disconnects and recover from unexpected hosting interruptions.

This is useful for enterprise scenarios such as:

• Multi-step research tasks
• Report generation
• Approval-based workflows
• Integrations with external systems

Continuous optimization and smarter tool access

Microsoft is also emphasizing continuous improvement using production traces to refine prompts, tools, skills, and model selection. In addition, Toolboxes now provide on-demand access to tools and procedures, helping agents discover only what they need when they need it.

Impact on IT administrators and developers

For Azure architects, platform engineers, and AI developers, these updates simplify agent lifecycle management. Teams can standardize on one agent foundation while still adapting to rapid changes in the AI model landscape.

The native voice capabilities may also reduce integration complexity for organizations planning customer service, employee support, or conversational business workflows across Teams and telephony channels.

Next steps

IT teams and developers should:

• Review new model options in Microsoft Foundry
• Test voice agents in public preview for multilingual use cases
• Evaluate long-running hosted agents for complex workflows
• Use production telemetry to optimize quality, latency, and cost over time

For organizations building enterprise AI on Azure, these updates make Foundry a more complete platform for deploying, operating, and continuously improving agents.