Fully Autonomous AI Agent in Azure Logic Apps (Here's How It Works)

Discover how to build your own autonomous AI agent using Logic Apps Agent Loop, even if you're new to AI development! This video provides a comprehensive walkthrough of creating an AI agent designed to play card games, offering a practical and engaging way to learn about agent loops in public preview. While playing card games might not be an enterprise scenario, it's an excellent method for understanding how to build and refine your agent instructions, setting you up for more complex AI solutions. In this demo, you'll learn: - How to configure an autonomous Logic App that runs without user interaction. - Strategies for refining agent instructions to improve AI performance and reliability, including handling service call retries and interpreting responses. - How to integrate various tools and models (like OpenAI and DeepSeek) within your agent to enhance its capabilities, such as determining game rules and optimizing player decisions. - Practical tips for debugging and monitoring your AI agent's execution using run history and AI Foundry. - Insights into the cost implications of running AI agents based on token usage and how different models can impact expenses. Don't miss out on these valuable insights into building intelligent, self-sufficient AI agents! Referenced Links: - Blog Post with Agent Instructions - https://www.stephenwthomas.com/azure-integration-thoughts/first-look-building-an-autonomous-ai-agent-with-logic-apps-agent-loop-demo-included/ - Getting Started Step by Step with Logic Apps Agent Loops - https://youtu.be/KPsssW2w6iM Key Highlights: 00:00 Introduction 01:05 Logic App Agent Loop Overview 03:34 Review Agent Instructions 05:57 Agent Tools Setup in Agent Loop 07:56 Using OpenAI Reasoning Model API as a Tool 09:22 Using OpenAI Web Search API as a Tool 10:37 Using DeepSeep R1 MaaS as a Tool 13:18 Card Game AI Agent Overview 13:36 Testing the AI Agent with High Card Wins 15:33 Testing the AI Agent with High Card Wins with 60 players 16:46 Testing the AI Agent with Five Card Draw 18:59 Checking Token Usage in AI Foundry 21:35 Reviewing the Five Card Draw Results 22:55 Reviewing the Agent Chat and Run History Don't forget to like and subscribe to @StephenWThomas!for more tutorials on Azure Integration Services, Logic Apps, and Artificial Intelligence (AI)! Check out my website and blog at https://www.StephenWThomas.com