Agentic AI Development using Microsoft Foundry Agent Service

Discover how ai agents—generative models with contextual knowledge and autonomous tools—act on your behalf, using vector stores and integrated apps to research, decide, and execute complex tasks.

Identify Azure agent development options, from Azure AI Foundry to Semantic Kernel, OpenAI API, Autogen, and Copilot Studio, and choose the right tool for multi-agent solutions.

Create an Azure Foundry hub and project in the Azure portal, configure your subscription, and deploy a GPT-4 agent. Explore model catalog, endpoints, and API keys, test deployments in playground.

Build an expense agent in Azure AI Foundry, use a knowledge base from expense policy data, and enable code interpreter to generate a downloadable claim text file.

Explore how the Azure Foundry Agent Service enables building AI agents as a fully managed, scalable solution with custom instructions, Python functions, tool connectivity, and real-world data.

Harness Azure AI Foundry Agent Service to enable automatic tool calling and secure thread management across sessions. Integrate built-in and custom tools, data, and diverse models for flexible, cost-conscious deployments.

Develop a data analysis agent on Azure AI Foundry, connect data, knowledge base, and code execution to generate charts via the web portal, and prepare for Python SDK integration.

Build and deploy a data analysis agent using the Python SDK for Azure, configure environments, upload knowledge base, enable code interpreter, and manage agent sessions with threads.

Explore integrating custom tools with Azure AI Foundry, a no-code platform, using built-in tools like Python code executor and vector store access to empower agents and boost efficiency.

Use custom tools to empower the agentic framework, automate repetitive tasks, and deliver streamlined, accurate outputs and workflows, with fixed input and fixed output for tailored solutions.

Explore options for implementing custom tools in the Azure Agent Foundry service, including custom function calling, Azure Functions, OpenAPI tools, and Azure Logic Apps.

Create an agent with a custom tool to convert natural language issues into formatted tickets, using the Azure Foundry service for automatic ticket creation.

Extend an Azure agent with a second tool to generate a support ticket and automatically email the details to the support team using SMTP and a custom email sender.

Delete your Azure resources by removing the resource group to stop costly foundation services and avoid charges up to ₹1,000 by deleting the agent foundry and project resources.

Learn to design a modular multi-agent system with a master agent coordinating connected agents to automate complex tasks using the agent foundry service.

Understand why connected agents matter in a multi-agent framework, with a manager agent orchestrating task flows and enabling modular, specialized per-agent capabilities, scalable and reliable traceability across tasks.

Design steps in a multi-agent framework begin with agent initialization, define a manager agent and connected agents, and manage tools and run workflow using the master diagram on Azure.

Design and deploy a four-agent workflow on Azure Foundry agent service to triage IT tickets with issue classifier, priority, and team agents under a manager agent.

Explore model context protocol tools integrated with Azure AI agents, wrapping MCP tools as asynchronous functions and revealing client-server discovery and dynamic access to external APIs and databases.

Explore how the model context protocol uses MCP as a bridge between agents and dynamic tools, enabling queries to external APIs from suppliers like L'Oreal, Unilever, Walmart, and Amazon.

Explore how dynamic tool discovery with MCP boosts scalability, modularity, maintainability, and flexibility, enabling centralized tool management and focused task delegation for agents.

Register MCP tools with an Azure AI agent by defining tool details, creating an MCP server and client, and describing capabilities for discovery and invocation.

Explore how an agent uses a GPT-4 LM and the MCP tool, bridged by the MCP client, to perform Microsoft document search via semantic search for Azure containerized applications.

Build and connect MCP tools to Azure AI agents, deploy a GPT-4 model, and enable semantic search of Microsoft docs to dynamically drive agent actions.

Learn how Semantic Kernel offers an open source sdk to build agentic AI frameworks, connect to Azure agent foundry, create agents, and integrate plugins or functions for customized tasks.

Leverage semantic kernel, an open source sdk from Microsoft, to connect application logic with large language model powered agents, bridging ai models and code through plugins, functions, and Azure Foundry.

Discover the core components of semantic kernel, including the AI service connector, memory and vector stores, functions and plugins, prompt templates, and filters, enabling connections to Azure Foundry agent services.

Build an expense claim agent using semantic kernel with Azure AI client and an email plugin to generate emails in the corporate format from expense details.

Build an expense claim agent using semantic kernel with Azure AI Foundry, create the agent, integrate an email plugin, and process claims end-to-end.

Build a multi-agent solution with semantic kernel and its agent framework, enabling group chat collaboration and task decomposition, guided by custom selection and termination strategies to manage costs.

Discover why a multi-agent framework overcomes single-agent limits for complex tasks, by separating monitoring logs, root-cause analysis, automated deployment, and insightful reporting within a scalable DevOps pipeline.

Explore the core components of the semantic kernel agent framework, including agents, agent collaboration, kernel, tools and plugins, and history management for robust multi-agent tasks.

Design an agent selection strategy to route queries to the right agents, improving accuracy, efficiency, and scalability, using predefined, kernel, and custom strategies in single and multi-agent frameworks.

Learn to manage multi-agent conversations with semantic kernel by applying termination strategies, including default, kernel function, and custom, to save tokens, prevent infinite loops, and improve accuracy.

Build a multi-agent DevOps solution using semantic kernel, with an incident manager and a DevOps assistant coordinated through a group chat and plugins for restart, rollback, redeploy, quota, and logging.

Implement the Azure Foundry service, document all course use cases, and showcase the knowledge on your resume while engaging with future topics and open-source agentic frameworks.