Skip to content

Build an agent with Azure AI Agent Service

Imagine you are a sales manager at Contoso, a multinational retail company that sells outdoor equipment. You need to analyze sales data to find trends, understand customer preferences, and make informed business decisions. To help you, Contoso has developed a conversational agent that can answer questions about your sales data.

What is an LLM-Powered AI Agent

An AI Agent is semi-autonomous software designed to achieve a given goal without requiring predefined steps or processes. Instead of following explicitly programmed instructions, the agent determines how to accomplish the task dynamically.

For example, if a user asks, "Show the total sales by region as a pie chart", the app doesn't rely on predefined logic for this specific request. Instead, a Large Language Model (LLM) interprets the request, manages the conversation flow and context, and orchestrates the necessary actions to produce the desired pie chart based on the regional sales data.

Unlike traditional applications, where developers define the logic and workflows to support business processes, AI Agents shift this responsibility to the LLM. In these systems, prompt engineering, clear instructions, and tool development are critical to ensuring the app performs as intended.

What is the Azure AI Agent Service

The Azure AI Agent Service is now in public preview, offering a fully managed cloud service with SDKs for Python and C#. It simplifies AI agent development, reducing complex tasks like function calling to just a few lines of code.

Function calling allows you to connect LLMs to external tools and systems. This is useful for many things such as empowering AI agents with capabilities, or building deep integrations between your applications and LLMs.

The Azure AI Agent Service offers several advantages over traditional agent platforms:

  • Rapid Deployment: Optimized SDK for fast deployment, letting developers focus on building agents.
  • Scalability: Designed to handle varying user loads without performance issues.
  • Custom Integrations: Supports Function Calling for extending agent capabilities.
  • Built-in Tools: Includes Fabric, SharePoint, Azure AI Search, and Azure Storage for quick development.
  • RAG-Style Search: Features a built-in vector store for efficient file and semantic search.
  • Conversation State Management: Maintains context across multiple interactions.
  • AI Model Compatibility: Works with various AI models.

Learn more about the Azure AI Agent Service in the Azure AI Agent Service documentation. In particular, read about the components of agents.

AI Agent Frameworks

Popular agent frameworks include LangChain, Semantic Kernel, and CrewAI. What distinguishes the Azure AI Agent Service is its seamless integration capabilities and an SDK optimized for rapid deployment. In complex multi-agent scenarios, solutions will combine SDKs like Semantic Kernel and AutoGen with the Azure AI Agent Service to build robust and scalable systems.