Azure AI Agent Service (AI Foundry Hub, Agent Framework)

Highlight generative AI and AI agents in compound AI systems, detailing foundation models, large language models, and tools like APIs and plugins for dynamic, domain-specific orchestration with human-in-the-loop options.

Explore how Azure OpenAI delivers private, tenant-isolated models with dedicated endpoints and keys, managed via the Azure OpenAI Studio for deployment, private connectivity, and security.

Deploy an Azure OpenAI resource in portal.azure.com, create a resource group in Sweden Central, and manage primary and secondary keys and the endpoint for bearer-token API calls.

Explore the Azure AI Foundry portal, deploy a GPT 35 turbo 16 K completion model, and chat in the portal's playground, using system prompts and prompt engineering to shape responses.

Connect python to an Azure OpenAI resource and use the chat completions API to query a GPT-4 model. Instantiate the client with endpoint and key from env and adjust temperature.

Azure AI Foundry Studio Hub enables building AI agents from scratch with foundation models and an agent service. It offers orchestration, monitoring, and LM ops in a collaborative DevOps environment.

Explore the differences between hub-based and standalone Foundry projects in Azure AI Hub, including version one and two of the Foundry SDK, collaboration scope, connections, and infrastructure.

Navigate hub-based and standalone Azure Foundry projects, compare classic and new portals, and learn connection methods and model catalog essentials for building agents.

Deploy a hub-based Azure Foundry project to enable prompt flow and AI services, including multi-service account, model catalog, playground, and PromFlow within a hub and its projects.

Deploy chat completion models from the model catalog in the AI Foundry AI Studio and test them in the chat playground. Explore GPT-4.0 options and global standard deployment.

Understand the Azure AI Foundry architecture, from the hub and projects to the management center, connecting storage, Azure OpenAI, and Azure AI services with secure key vault authentication.

Explore the Azure AI Agent Service in Azure AI Foundry, comparing declarative and custom engine agents, and learn to build secure, scalable agents with the Semantic Kernel SDK.

In this hands-on lab, build an Azure AI agent via the Azure AI studio UI connected to an Azure OpenAI GPT-4 resource, and explore thread-based conversation history and knowledge sources.

Explore how the Azure AI Foundry SDK evolved from the chat completions and assistants API to an AI agent service. Access multi-provider models, enterprise data grounding, and open API tools.

Navigate choosing the Azure AI Projects SDK version for labs, and set up a Python virtual environment, install dependencies from requirements.txt, and run labs in the AI Foundry Hub ecosystem.

Learn to build a basic agent programmatically with the Azure AI Foundry SDK, configuring a GPT four model, creating an agent and thread, then running and polling for responses.

Explore thread management in Azure AI Foundry SDK and Azure Agent Service to preserve contextual conversation history across queries using GPT-4 in a terminal chat.

Deploy a Bing grounding resource in portal.azure.com and connect it to your Azure AI Foundry project to enable a Bing-based web searching agent that fetches information with URL citations.

Build an Azure AI agent with a Bing grounding tool via the Foundry SDK, deploying a GPT-4 model, and run agent.py to enable real-time citations.

Learn how function calling in Azure AI agents enables custom Python code snippets to perform math operations and call external APIs, enhancing responses with user-defined functions and LLM summarization.

Demonstrate function calling in Azure AI service by implementing two Python functions: one queries an external weather API, the other fetches user details by id, orchestrated by the agent.

Explore the OpenWeatherMap API to fetch weather conditions for a location using the Azure Agent service, and generate a free API key from your account for lab use.

Create a function calling agent that retrieves weather from Openweathermap and mock user data from local Python code. The hands-on lab covers setup, function definition, and running the agent.

Explore OpenAPI tools and the OpenAPI schema to define endpoints, request types, and parameters, enabling the Azure AI agent to call APIs for knowledge retrieval without custom wrappers.

Build an Azure AI agent that uses an OpenAPI specified tool to query a free real-time weather API from wtr.in and return structured JSON data.

This hands-on lab demonstrates building an Azure Agent Service AI agent using an OpenAPI tool to call a weather API, dynamically filling location and format parameters for live responses.

Uncover retrieval augmented generation, an architecture that retrieves documents from private data using Azure AI search, augments prompts with content, and generates answers with a language model using vector embeddings.

Learn how vector embeddings convert words, sentences, and images into high-dimensional vectors to capture semantic meaning and power retrieval augmented generation with Ada 002 and Azure storage options.

Learn to create vector embeddings in python using azure openai text embedding 002 model, configure environment variables, and generate 1536-dimensional embeddings with client.embeddings.create for semantic meaning indexing in rag architecture.

Learn to implement retrieval augmented generation with Azure AI search, leveraging vector and full-text search on unstructured data. Use indexing and AI enrichment with vector embeddings for contextual answers.

Explore the rag agent architecture using the Azure Agent Service, linking unstructured hotel PDFs to the Azure AI Search index with vector embeddings.

Upload PDF documents to an Azure storage account, generate vector embeddings with a text embedding model, and create an Azure AI search index with vector storage for a RAG agent.

Bring your rag agent to life in a hands-on lab by wiring the Azure AI Foundry SDK to an Azure AI search index, environment variables, and a GPT-4 powered agent.

Discover the code interpreter capability of an Azure AI agent by analyzing an Excel dataset and generating a Python chart of price versus product in a sandbox.