Asha Sharma introduces the latest updates to Microsoft Foundry, highlighting its modular agent stack, enhanced security, expanded AI model support, and powerful tools empowering developers and enterprises to build scalable, secure, and production-ready AI solutions.

Microsoft Foundry: Empowering Developers with a Modular and Secure AI Agent Platform

By: Asha Sharma

Microsoft Foundry has quickly evolved into a modular, interoperable, and secure agent platform, reshaping how organizations and developers build, deploy, and manage intelligent AI-driven systems at scale. This update from Microsoft Ignite 2025 presents major advancements in modular agent stacks, integration of new model partners, orchestration tooling, robust security controls, and cloud-edge deployment.

Highlights and Key Announcements

1. Expanded Model Ecosystem

Foundry Models now includes models from Anthropic (Claude family), Cohere, NVIDIA, and more.
Features a generally available model router, letting developers dynamically select the best-fit AI model per task.
Developer tier enables accessible model fine-tuning using idle GPU resources.

2. Foundry IQ: Advanced Retrieval-Augmented Generation

Foundry IQ revolutionizes RAG by offering iterative dynamic reasoning, centralizing workflows, and simplifying orchestration for agents.
Integrates closely with Azure AI Search and Microsoft Fabric IQ, providing high-quality information retrieval with respect for permissions/data classifications.

3. Foundry Agent Service: Production-Grade Agent Management

Supports hosted single and multi-agent workflows with built-in memory and state management.
Enables deployment to Microsoft 365 apps and Agent 365 for enterprise-grade scenarios.
Visual and code-first workflow designers empower developers with flexibility and scalability.

4. Foundry Tools: Business Systems and Multimodal Integration

Provides real-time, secure API and connector access (over 1,400 connectors to business systems like SAP, Salesforce).
Multimodal capabilities such as document processing, translation, and transcription.
MCP tools API management allows integration of any function or workflow for agent use.

5. Foundry Control Plane: Unified Security, Identity, and Operations

Centralizes policy, observability, and security with durable identity (Entra Agent ID), integrated Microsoft Defender and Purview protection, and OpenTelemetry support for monitoring.
Enables continuous red teaming and unified alerting for risks across the agent lifecycle.
Defender for Cloud and GitHub Advanced Security integration surfaces AI-driven code security improvements directly in developer workflows.

6. Foundry Local: Edge AI on Android

Foundry Local, already present on Windows and Mac, is now previewing on Android.
Supports offline/edge agent scenarios—real-time inference, privacy-aware computation, and global scalability.

7. Managed Instance on Azure App Service

Lets organizations migrate .NET web apps to scalable, AI-augmented Azure infrastructure with minimal code changes and streamlined operations.

For Developers: Productivity and Collaboration

GitHub Agent HQ introduces unified agent management within GitHub and Visual Studio Code.
Brings together tools for rapid design, testing, monitoring, and improvement of AI-powered systems.
Copilot Metrics empowers data-driven insights for teams.

Security, Governance, and Operations

Unified guardrails across models, agent behaviors, and tool interactions (input/output).
Real-time policy enforcement and fleet-wide observability.
Seamless collaboration between development and security teams thanks to new integrations.

Get Started

Conclusion

Microsoft Foundry delivers a unified, open platform for AI agent development, combining rich model choice, adaptive orchestration, robust security, and developer-first tooling. With these updates, developers can confidently build and scale production-ready AI and agent solutions for the next generation of intelligent enterprise software.

This post appeared first on “The Azure Blog”. Read the entire article here