Microsoft Foundry Agent Service is now generally available,
empowering developers to design, deploy, and scale enterprise-grade AI agents.
It supports multi-agent workflows, allowing specialized agents to handle
complex tasks, accelerate decisions, and boost operational efficiency. With
open protocols like Agent-to-Agent (A2A) and Model Context Protocol (MCP), it
enables seamless interoperability while maintaining control over orchestration.
The service integrates with knowledge sources like Bing,
SharePoint, Databricks, and Microsoft Fabric, and includes a centralized
catalog of tools and pre-built agent templates for easy customization. It also
leverages a unified runtime that combines Semantic Kernel and AutoGen, offering
a consistent experience for building, testing, and deploying agents locally or
in the cloud.
To ensure reliable
performance, Microsoft Foundry Agent Service introduces Bring Your Own (BYO)
thread storage with Azure Cosmos DB and robust AgentOps capabilities like
tracing, evaluation, and monitoring, helping developers optimize agent behavior
with confidence.