Announcing the public preview of Azure AI Agent Service
AI agents are transforming the way businesses automate workflows. By providing agents access to the same apps and services your employees have, AI agents can automate manual, time-intensive processes and drive significant productivity gains in the process.
Deploying reliable AI agents in real-world environments, however, remains a challenge. Many existing agent services lack 1) the secure, integrated tools necessary for AI agents to perform real work, like updating a database or sending an email; 2) crucial context necessary for an agent to complete tasks is also often missing; and once the AI agent is running, 3) it's difficult to identify and diagnose issues.
To address these challenges, we announced Azure AI Agent Service at Microsoft Ignite 2024. This service is purpose-built for designing, customizing, and deploying AI agents that natively connect with the rest of the Microsoft ecosystem:
- With the OpenAPI defined tools and Azure Functions, businesses can create support agents that automate mundane tasks like sending emails, schedule meetings, and updating databases.
- With Azure AI Search and Grounding with Bing Search, agents can pull the information needed to make informed actions.
- And with integrated OpenTelemetry tracing with Application Insights, you can observe everything your agent is doing.
In addition to these integrations, Azure AI Agent Service can also be used with two of the most popular multi-agent frameworks. With AutoGen and Semantic Kernel, businesses can orchestrate multi-agent processes which allows AI agents to dynamically interact with each other to coordinate complex tasks, refine responses, and improve decision-making.
Today, we’re excited to announce that Azure AI Agent Service is now publicly available in the Azure AI Foundry SDK and the Azure AI Foundry portal, bringing the power of enterprise-ready AI agents to businesses worldwide.
Transforming Automation for Businesses
Our private preview customers, have already shown how Azure AI Agent Service can transform industries by enabling organizations to automate workflows, enhance decision-making, and improve efficiency.
Fujitsu, for example, is leveraging Azure AI Agent Service to increase the efficiency of their sales team and push what’s possible with retrieval augmented generation (RAG). Instead of having a single agent powered by Azure AI Search, Fujitsu created several specialized sales AI agents that work together to answer user questions.
In addition to Fujitsu, we’ve seen Azure AI Agent Service used in many other industries. In healthcare, AI agents are automating administrative workflows, streamlining access to clinical research, and assisting with patient data management. Energy companies are using AI-powered monitoring and predictive maintenance to optimize grid performance and drive sustainability efforts. In travel and hospitality, businesses can build AI assistants that enhance itinerary planning, personalize recommendations, and automate customer inquiries using proprietary data from Tripadvisor.
Consulting and professional services firms are deploying AI agents to analyze financial reports, generate insights, and improve decision-making, while software and technology companies are leveraging AI-powered developer agents for code generation, validation, and debugging. In retail and automotive, AI agents are automating customer support, optimizing supply chain operations, and enhancing real-time inventory tracking, improving both operational efficiency and customer experiences.
What makes Azure AI Agent Service special?
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 the knowledge and action tools you provide them.
Creating agents with Azure AI Agent Service can easily be done either through the code-first Azure AI Foundry SDK or the Azure AI Foundry portal experience. Regardless of the route, developers can be confident that the agents they deploy can benefit from Azure’s enterprise-grade security and performance.
What makes Azure AI Agent unique is the development speed it enables, the amount of flexibility it offers, and the enterprise assurances it provides.
1. Rapid Development and Automation
With Azure AI Agent Service, all the compute, networking, and storage needed to run an agent micro-service is already managed for you. As a developer, you merely need to declaratively define the agent’s model, instructions, and tools within the Azure AI Foundry portal or with the Azure AI Foundry SDK.
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,
)Afterwards, running an agent is as simple as creating a thread and running the agent on top of it.
# Create a thread with messages
thread = project_client.agents.create_thread()
message = project_client.agents.create_message(
thread_id=thread.id,
role="user",
content="Could you please create a bar chart for the operating profit using the following data and provide the file to me? Company A: $1.2 million, Company B: $2.5 million, Company C: $3.0 million, Company D: $1.8 million",
)
# Ask the agent to perform work on the thread
run = project_client.agents.create_and_process_run(thread_id=thread.id, agent_id=agent.id)
# Fetch and log all messages to see the agent's response
messages = project_client.agents.list_messages(thread_id=thread.id)
print(f"Messages: {messages}")For more complex scenarios requiring multi-agent orchestration, you can use Semantic Kernel’s new connection to Azure AI Agent service or AutoGen’s connection to agents that use the Assistants API. By combining Azure AI Agent Service with Semantic Kernel or AutoGen, you get the best of both worlds: a fully managed enterprise agent service with the latest in multi-agent research.
2. Extensive Data Connectivity
The service supports integrations with a variety of data sources, such as files, Microsoft Bing, Azure AI Search, and actions, such as OpenAPI defined tools for existing APIs and Azure Functions for custom logic.
"We are leveraging Azure AI Agent Service to revolutionize our sales journey and go-to-market strategy. By deploying data-driven AI agents, we gain deeper insights into our clients' needs, allowing us to enhance and customize our offerings. The service also facilitates seamless integration, enabling rapid deployment and scaling, thereby accelerating our speed-to-value." - Maureen Tan, Head of AI CoE Global Units, NTT Data Inc.
In the coming weeks, you can also bring in licensed data from providers like Tripadvisor. With Tripadvisor’s vast repository of unique data across the entire spectrum of travel journey inclusive of hotels, experiences and restaurants, plus verified travel insights, businesses can build AI agents that can offer personalized recommendations, generate dynamic itineraries, and enhance customer interactions with real-time, trusted travel data.
With these connections, your agent can retrieve contextual data from both public and private data sources when it’s asked to perform actions.
3. Flexible Model Options
In addition to the latest OpenAI models, the agent service also supports other agentic models from key partners like Meta, Mistral, and Cohere. With these function-calling enabled models, AI agents can process tasks, create a plan, and use tools to automatically complete tasks.
4. Built-in Enterprise Readiness
Lastly, as an enterprise developer, you may bring your own Azure resources, like Azure AI Search, Azure Storage, Azure Key Vault, private virtual networks (coming soon), and thread storage (coming soon). This gives you complete control over the data used by the Azure AI Agent Service to orchestrate agents.
Additionally, Azure AI Agent Service provides several other enterprise features out of the box, including:
- No public egress for strict data privacy compliance
- Monitor agent performance with OpenTelemetry-based tracing
These capabilities help organizations protect sensitive information and support their ongoing journey to align to regulatory standards.
"YoungWilliams' AI agent, 'Priya,' harnesses advanced Azure AI technologies to elevate customer service experiences for State Health and Human Services Agencies. We are actively exploring Microsoft's Azure AI Agent Service to further enhance these capabilities. With its seamless integration and secure access to private data, this service presents exciting opportunities to deliver highly personalized and efficient support." – Rick Broxterman, Director of Emerging Technology, YoungWilliams.
What’s Next?
- Explore the documentation to learn more about Azure AI Agent Service
- Start building your agents today in Azure AI Foundry
- Visit our learning repo to build your first agent with Azure AI Agent Service
- Join Microsoft product leaders on 2/10 at 8:45 am PT on Discord for an AMA on Azure AI Agent Service
- Join our webinar on 2/27 to gain knowledge on best practices for implementing multi-agent systems with Azure AI Agent Service
We are excited to see how you will use Azure AI Agent Service for innovative AI-powered business solutions.
Microsoft
