Agent Development Options
This managed Azure service, based on the OpenAI Assistants API but with increased model choice, data integration, and enterprise security, provides a framework for creating, managing, and using AI agents.
Azure AI Foundry Agent Service
These AI agents assist individuals with daily tasks like scheduling meetings, sending emails, and managing to‑do lists; Microsoft 365 Copilot is an example.
Personal productivity agents
Interpreting user input, selecting the appropriate connected agent, forwarding context and instructions, and aggregating results are responsibilities of this central agent in a connected agent solution.
What is the main agent?
This VisualFeatures option identifies tags about an image, including objects, scenery, and actions.
What is VisualFeatures.TAGS?
This Azure Cognitive Service can detect human faces in an image and return attributes like age, emotion, and head pose.
What is the Face service?
This API provides a subset of the features in Foundry Agent Service and can only be used with OpenAI models.
OpenAI Assistants API
These agents continuously monitor market trends, gather data, and generate reports; they can be used in financial services, healthcare, or marketing.
Research agents
Completing a specific action based on a clear prompt, using tools if needed, and returning results to the main agent are responsibilities of these specialized agents.
What are connected agents?
This feature returns the bounding box for each detected object in an image.
What is VisualFeatures.OBJECTS?
This feature of the Face service allows you to determine if two detected faces belong to the same person.
What is face verification?
This lightweight open-source development kit is designed to build AI agents and orchestrate multi-agent solutions, with a core SDK and an Agent Framework.
Semantic Kernel
These agents automate lead generation and qualification, research potential leads, send personalized follow‑up messages, and even schedule sales calls.
Sales agents
The first step in setting up a multi-agent solution using Azure AI Foundry Agent Service is to create this client to connect to your Azure AI Foundry project.
What is initializing the Agents client?
This feature generates a natural language description of the entire image.
What is VisualFeatures.CAPTION?
When comparing a detected face against a set of stored faces, this operation identifies the most similar faces.
What is face identification?
This open‑source framework is useful as a research and ideation tool when experimenting with agents.
AutoGen
These agents handle routine inquiries, provide information, and resolve common issues via chatbots; Cineplex uses one to process refund requests and improve customer satisfaction.
Customer service agents
After defining a connected agent using the Agents client, create this tool with a name and description so the main agent can call the connected agent.
What is a ConnectedAgentTool?
This feature extracts readable text from an image.
What is VisualFeatures.READ?
The Face service can return these two types of attributes for detected faces: one related to physical features and one related to emotional state.
What are facial landmarks and emotion?
Using this SDK, developers can create self‑hosted agents for delivery through channels like Slack or Messenger, even though the SDK’s name references Microsoft 365.
Microsoft 365 Agents SDK
These agents assist in software development tasks like code review, bug fixing, and repository management, and can automatically update codebases; GitHub Copilot is an example.
Developer agents
After creating the main agent, these final steps complete the multi-agent workflow: create a thread and send a message, run the agent workflow, and handle the results.
What are creating a thread and message, running the agent workflow, and handling the results?
This feature returns the bounding box of the specified aspect ratio for the area of interest in an image.
What is VisualFeatures.SMART_CROPS?
Before using the Face service in production, Microsoft recommends following these two key principles to ensure responsible AI use.
What are fairness and transparency?