AI Agents-Beginner to Pro-Azure AI Foundry Agent Service

Explore the Azure AI Foundry architecture, including the model catalog and core AI services like Azure Open AI service, with observability and governance.

Explore how Azure AI Foundry uses hubs as governance containers and projects as workspaces to develop, experiment, and deploy models, a data set, and prompts with shared resources.

Compare Azure OpenAI service usage with and without a project in Azure AI Foundry, detailing model catalog, fine-tuning, deployment, SDK access, and enterprise features across multiple AI services.

Compare benchmarks in the foundry model catalog to evaluate quality, cost per million tokens, and latency. Use graphs or lists to choose among models like GPT four and llama.

Explore the assistance API as the foundation for the agent service, learn function calling with third-party APIs, and compare code between the two approaches.

Explore the Azure OpenAI assistants API architecture, detailing how the user, files, threads, and runtime environment interact, with models and tools like code interpreter and function calling driving data flow.

Explore function calling with a Python demo, extending OpenAI and Azure OpenAI by calling a third-party API to fetch weather data from openweathermap.org.

Install the latest OpenAI Python package and load dot env, then configure the Azure OpenAI endpoint and API keys in Azure OpenAI dot env for the next steps.

Instantiate the Azure OpenAI client in Python by wiring the endpoint, API key, and API version, using the latest in development and GA in production.

Define a get weather function that validates latitude and longitude, builds a weather API request using an API key from the environment, and returns latitude, longitude, weather condition, and temperature.

Learn how to define a tools list for the Azure OpenAI assistants API, using function calling to specify a get weather tool with latitude and longitude parameters and required checks.

Create an assistant named weather assistant with thread and messages, and pass a tools list via Azure OpenAI deployment name for function calls like get weather using latitude and longitude.

Demonstrate running the conversation with the assistant using the run conversation function, creating a user message, managing thread messages and runs, and processing tool outputs.

Observe a Python demo that takes user input, loops the conversation with the Azure OpenAI assistant API, and fetches weather data for London and New Delhi via open weather map.org.

Master the Azure AI agent service to build, deploy, and manage AI agents from beginner to pro levels, enabling scalable intelligent automation.

Explore the architecture and knowledge and action tools of the Azure AI Agent Service, compare it with the assistance API, and learn about quotas, pricing, and regional model availability.

Azure AI agent service is a fully managed platform that simplifies building and deploying AI-powered agents within the Azure ecosystem, enabling secure, scalable, multi-modal, and multi-vendor AI deployments.

Discover how the Azure AI agent service sits under the Azure AI Foundry, enabling bring your own file storage, bring your own search index, and on behalf of authorization.

Compare the assistance api with azure ai agents and see that the architecture remains the same, with assistant, thread, run, and messages now aligned to agent terminology and sdk usage.

Explore Azure AI agent service quotas and limits, including 10,000 files per agent per thread, 512 MB per file, 100 GB total uploads, and 2 million tokens.

Azure AI agent service is free to use, but costs accrue from the Azure OpenAI service and associated tools, including tokens, models, batch processing, file search, code interpreter, and automation.

Create an Azure AI hub and a project inside it using Azure AI Foundry, which automatically provisions the Azure Open AI service, storage, and key vault for agent deployment.

Launch studio, select an azure openai resource, and deploy a model to create and run an azure ai agent, such as gpt-4 mini, within the portal.

Explore how AI agents connect to deployments and tools, leveraging knowledge tools such as file search, Azure AI search, and Bing grounding, and action tools like code interpreters.

Submit a prompt to the agent via the agents playground and view the output. Explore thread details, tokens, and a graphical interface or code interpreter and Python codes.

Discover how to use the Azure SDK for Python to build AI agents on the Azure AI Foundry Agent Service.

Learn to use the azure sdk for python to build ai agents in azure, covering ai projects client library, authentication, and running a sample with azure cli and vscode.

Explore using the Azure AI projects client library in Python to build agents, authenticate with default credentials, and leverage tools such as file search and code interpreter.

Demonstrate setting up a VS Code environment with a Python virtual environment named agents, installing Azure CLI, logging in with default Azure credentials, and preparing to create an agent.

Explore how to instantiate an AI project client from a connection string using the default Azure credential. Create a thread and run a GPT-based agent to respond to user messages.

Learn the end-to-end workflow for agent creation in Azure AI Foundry, including creating a project and hub, deploying models, and configuring authentication with an app registration, secret, and contributor access.

Register a Microsoft Entra ID app, create a client secret, and grant contributor access to an Azure AI Foundry project by assigning the role to the app in IAM.

Explore function calling as the first action tool in the Azure AI Agent Service, install libraries, configure environment variables, and invoke a weather function to run the weather agent.

Create a new deployment inside an Azure AI project using Launch Studio, select a base model such as GPT-4 or mini, and configure deployment details for LLM access in code.

Demonstrate the weather agent workflow by calling a third-party Openweathermap API through function calling, converting London to latitude and longitude, and generating a friendly weather report.

Install the Azure AI projects SDK and Azure Identity library, and load environment variables from a .env file. Fetch client id and tenant id from the environment to initialize authentication.

Define a get weather function that calls OpenWeatherMap with latitude, longitude, and an API key, validates inputs, builds the request, and returns weather condition and temperature in JSON.

Create and call an azure ai agent using a function tool and tool set to fetch weather data from OpenWeatherMap, authenticate with client secret credentials, and run interactive weather query.

Learn how to use a code interpreter within AI agents using Azure AI Foundry Agent Service, guiding beginners to pro users through practical implementation and workflows.

Explore the Azure OpenAI code interpreter in a sandbox that runs Python for secure data analysis and automation in chat, demonstrated in the agent playground with sales data graphs.

Demonstrate the Azure AI agent workflow using the code interpreter to upload the sales data CSV to Google Colab, set credentials, and generate and save graphs as PNGs.

Demonstrates a while true loop driving a thread and message conversation in Azure AI Foundry Agent Service, including breaking on end, creating messages, running and fetching responses, and saving images.

Execute the program to upload data, create the agent and thread, and generate a graph of total revenue and profit by product, with the agent replying from the uploaded data.

Discover OpenAPI 3.0 concepts for designing clear, interoperable APIs. Apply these principles to Azure AI Foundry Agent Service workflows.

Describe Open API 3.0 as standard for describing RESTful APIs in YAML or JSON, detailing endpoints, parameters, and authentication, formerly swagger, used as a tool in Azure AI Agent Service.

Explore how an Azure AI agent uses an open API tool connected to Rapid API and Yahoo Finance to fetch stock data, with connectors, JSON schema, and API key authentication.

Create a RapidAPI key in the console, then wire your azure ai agent service to Yahoo Finance data using the Yahoo Finance collection and api endpoint.

Authenticate with Azure using a client secret credential, initialize the Azure AI project client with the project connect string, and load the OpenAPI specification to resolve refs and describe endpoints.

Extract the Yahoo Finance connection ID, create the authentication object for the open API tool, and initialize the Get Yahoo Finance data tool to fetch real-time stock market and financial data.

Execute and test a yahoo finance assistant by attaching the get yahoo finance data tool and querying tickers like apple and tesla, using azure openai agents.

Utilize Bing search as a knowledge tool to empower AI agents with fast, accurate information retrieval and improved decision making.

Explore grounding with Bing search as a knowledge tool for AI agents. Learn supported models and build a Bing search agent in the playground, with no-code and Python options.

Explore grounding with Bing search, a knowledge tool in the Azure AI agent service. Learn pricing, setup, and GPT model availability for grounding with Bing search.

Set up your environment and install the Azure AI Projects SDK, Azure Identity, and dotenv to manage environment variables, then configure the Bing connection name for correct execution.

Configure an Azure AI project client with client secret credentials and an environment variable connection string, and integrate a Bing grounding tool via a Bing search resource connection ID.

Execute the code to create and manage an agent, reveal the connection ID, fetch top news with Bing, and delete conversations and agents for cleanup.

Explore knowledge tools and file search techniques for building AI agents with the Azure AI Foundry Agent Service, bridging beginner concepts to professional capabilities.

Explore file search as a knowledge tool in the Azure AI agent service, enabling retrieval from blob storage to power Rag, with both no-code Foundry and code-based demonstrations.

Learn how file search augments an assistant with external private data using embeddings and a vector store, enabling semantic and keyword search with automatic updates.

Demonstrate file search with Azure AI agents using a vector store for semantic search, uploading a document, and using retrieval-augmented generation to summarize content and extract cybersecurity tips.

Demonstrate how the file search tool links to an Azure AI agent via Python to perform semantic searches on uploaded files using a vector store and generate bullet points.

Install azure ai projects sdk and azure identity library, configure environment variables, upload api settings dot env and vector file for semantic search, then create vector stores and AI agents.

Import modules, configure azure credentials, initialize azure ai project client, and upload GPT four system card to create a vector store for AI agent interaction and knowledge via file search.

Create a vector store for embeddings, linking uploaded files to enable semantic search. Build a file search tool and an Azure OpenAI agent that replies only from uploaded documents.

Create a conversation thread, upload a file as a message attachment, send user messages to the AI agent, process and retrieve responses, and optionally clean up resources to avoid costs.

Demonstrates running the code to create a vector store, an agent, and a thread, then summarizes a document into five lines and tests context awareness in azure ai agents.