Microsoft recently announced the public preview of Azure AI Agent Service, now available in the Azure AI Foundry SDK and the Azure AI Foundry portal.
At Ignite 2024, the company first introduced the service to developers for designing, customizing, and deploying AI agents that natively connect with the rest of the Microsoft ecosystem:
Mads Bolaris, group program manager at Microsoft for Azure Assistants, writes:
Azure AI Agent Service is a flexible, use-case-agnostic platform for building, deploying, and managing AI agents as micro-services. These agents can operate independently or side-by-side with a human by leveraging your knowledge and action tools. Creating agents with Azure AI Agent Service is easy, either through the code-first Azure AI Foundry SDK or the Azure AI Foundry portal experience.
With Azure AI Agent Service, the compute, networking, and storage for an agent micro-service are managed, and developers need to define the agent’s model, instructions, and tools in the Azure AI Foundry portal or using the SDK. Developers can use Semantic Kernel with Azure AI Agent service or AutoGen’s Assistants API for complex scenarios needing multi-agent orchestration.
(Source: Tech Community blog post)
Developers can choose the latest OpenAI models and other agentic models from key partners like Meta, Mistral, and Cohere when defining the AI model. A definition can look like:
agent = project_client.agents.create_agent(
model="gpt-4o-mini",
name="my-agent",
instructions="You are helpful agent",
tools=code_interpreter.definitions,
tool_resources=code_interpreter.resources,
)
Other hyperscalers, like AWS and Google, offer similar services to Microsoft’s. AWS has Amazon Bedrock Agents, a fully-managed service that allows developers to build and configure autonomous agents for applications – and Google offers Vertex AI Agent Builder, an AI development tool that simplifies building AI agents.
The Azure AI Agent Service can be viewed as an Agent as a Service (AaaS). AI expert Animesh Bokil posted on LinkedIn:
Agent-as-a-Service (AaaS) redefines how businesses leverage AI, moving beyond static software to autonomous AI-driven agents. Unlike traditional SaaS, which provides software tools, AaaS enables intelligent agents that can automate workflows, interact with systems, and make real-time decisions on behalf of users. From customer service chatbots to autonomous financial advisors and digital operations managers, AI agents are transforming industries by enhancing efficiency, reducing manual workload, and driving intelligent automation.
Lastly, developers can find guidance for building an agent through a GitHub repo and course.
