Enhancing Agents with Azure AI Services

Agentic AI goes beyond single-turn chat: it plans against a goal, decides which tools to call, and keeps context for later steps. You can use this pattern with or without a formal agent construct; here we use Azure AI Foundry’s Agent plus Azure AI Translator and Document Intelligence to show a Reason → Act → Observe loop.

In this hands-on lab, you’ll build a Logic App workflow that handles a multilingual customer complaint payload, translate the text, and analyze an invoice document; then you’ll connect that workflow as a tool to an Azure AI Foundry Agent. The agent uses a deployed Large Language Model, calls the Logic App to orchestrate translation and document extraction, and returns a bilingual summary. You’ll run and trace the flow in the Agent Playground to see the tool calls and results end to end.

Learning objectives

Upon completion of this beginner-level lab, you will be able to:

• Explain how agentic AI differs from single-turn LLM prompts and why tool use improves reliability.
• Build a multimodal agent that processes text and documents using Azure AI Translator, Document Intelligence, and a Logic App workflow as a tool.
• Observe and reason about the Reason → Act → Observe loop using the Agent trace view in Azure AI Foundry.

Intended audience

• Cloud Solution Architects​
• AI/ML Engineers and Developers​
• Technical Consultants exploring Azure AI​
• Innovation and R&D Teams evaluating AI use cases

Prerequisites

• Basics of Azure AI Foundry (Agents & Tools)
• Awareness of Azure AI Language, Translator, and Document Intelligence