Get Started with Generative AI on Azure

Understand Generative AI Applications

• GenAI applications are built with language models that powers the 'app logic' component of the interaction between users and GenAI.

Understand Assistants

• GenAI often appears as chat-based assistants that are integrated into applications to help users find information and perform tasks efficiently.

note

Microsoft Copilot is a GenAI based assistant that is integrated into a wide range of Microsoft applications and user experiences.

Understand Agents

• GenAI that can execute tasks such as filing taxes or coordinating shipping arrangements, just as a few examples, are known as agents.
• Agents are applications that can respond to user input or assess situations autonomously, and take appropriate actions.
• These actions could help with a series of tasks.
• Agents contain three main components:
  • A language model that powers reasoning and language understanding.
  • Instructions that define the agent’s goals, behavior, and constraints.
  • Tools, or functions, that enable the agent to complete tasks.

note

• Today's AI solutions often contain a combination of assistant, agentic, and other AI capabilities.
• The process of coordinating and managing multiple AI components such as models, data sources, tools, and workflows to work together efficiently in a unified solution is known as orchestration.

Use a framework for understanding Generative AI applications

• GenAI applications can be categorized into the following categories.
  1. Ready-to-use: No work required from the user's end. User can just start asking questions.
  2. Extendable: Same as ready-to-use application but with the extension to use own data.
  3. Applications you build from the foundation: You can build your own assistants and assistants with agentic capabilities starting from a language model.
• Often, you will use services to extend or build generative AI applications.
• These services provide the infrastructure, tools, and frameworks necessary to develop, train, and deploy generative AI models.

Understand tools to develop generative AI

• Microsoft offers a powerful ecosystem of tools and services for building GenAI solutions, designed to support developers, data scientists, and enterprises at every stage of the AI lifecycle.
• Azure AI Foundry is a PaaS that gives developers control over the customization of language models used for building applications.
• These models can be deployed in the cloud and consumed from custom-developed apps and services.
• You can use Azure AI Foundry portal, a user interface for building, customizing, and managing AI applications and agents, especially those powered by generative AI.
• Components of Azure AI Foundry include:

Component	Description
Azure AI Foundry model catalog	A centralized hub for discovering, comparing, and deploying a wide range of models for generative AI development.
Playgrounds	Ready-to-use environments for quickly testing ideas, trying out models, and exploring Azure AI services.
Azure AI services	In Azure AI Foundry portal, you can build, test, see demos, and deploy Azure AI services.
Solutions	You can build agents and customize models in Azure AI Foundry portal.
Observability	Ability to monitor usage and performance of your application's models.

note

Microsoft's Copilot Studio is another generative AI development tool. It is designed to work well for low-code development scenarios in which technically proficient business users or developers can create conversational AI experiences.

Understand Azure AI Foundry's model catalog

• Azure AI Foundry provides a comprehensive and dynamic marketplace containing models sold directly by Microsoft and models from its partners and community.

• Azure OpenAI in Foundry models make up Microsoft's first-party model family and are considered foundation models.
• Foundation models are pretrained on large texts and can be fine-tuned for specific tasks with a relatively small dataset.
• You can deploy the models from Azure AI Foundry model catalog to an endpoint without any extra training.
• If you want the model to be specialized in a task, or perform better on domain-specific knowledge, you can also choose to customize a foundation model.
• To choose the model that best fits your needs, you can test out different models in a playground setting and utilize model leaderboards (preview).
• Model leaderboards provide a way to see what models are performing best in different criteria such as quality, cost, and throughput.
• You can also see graphical comparisons of models based on specific metrics.

Understand Azure AI Foundry capabilities

• Azure AI Foundry portal provides a user interface based around hubs and projects.
• In general, creating a hub provides more comprehensive access to Azure AI and Azure Machine Learning.
• Within a hub, you can create projects.
• Projects provide more specific access to models and agent development.
• You can manage your projects from Azure AI Foundry portal's overview page.

• When you create an Azure AI Hub, several other resources are created simultaniously, including an Azure AI services resource.
• In Azure AI Foundry portal, you can test all kinds of Azure AI services, including Azure AI Speech, Azure AI Language, Azure AI Vision, and Azure AI Foundry Content Safety.

• In addition to demos, Azure AI Foundry portal provides playgrounds to test Azure AI services and other models from the model catalog.

Customizing models

• There are many ways to customize the models in GenAI applications.
• The purpose of customizing the model is to improve its performance, including quality and safety of the responses.
• Following are four main ways to customize models in Azure AI Foundry:

Using grounding data

• It refers the process to ensure that the systems output are aligned with reliable data source.
• It can be done in various ways such as linking the model to a database, search engine or providing domain-specific knowledge bases.

Implementing Retrieval-Augmented Generation (RAG)

• RAG augments the language model by connecting it to an organization's proprietary database.
• This technique involves retrieving relevant information and using it to generate contextually accurate responses.

Fine-tuning

• It involves taking a pretrained model and further training it on a smaller, task-specific dataset to make it more suitable for a particular application.
• Fine-tuning is useful for adapting models to domain-specific requirements, improving accuracy, and reducing the likelihood of generating irrelevant or inaccurate responses.

Managing security and governance controls

• Security and governance controls are needed to manage access, authentication, and data usage.
• These controls help prevent the publication of incorrect or unauthorized information.

Understand observability

• Although there are many ways to measure GenAI quality; in general, there can be three dimensions for evaluating and monitoring GenAI.
  • Performance and quality evaluators: assess the accuracy, groundedness, and relevance of generated content.
  • Risk and safety evaluators: assess potential risks associated with AI-generated content to safeguard against content risks. This includes evaluating an AI system's predisposition towards generating harmful or inappropriate content.
  • Custom evaluators: industry-specific metrics to meet specific needs and goals.
• Azure AI Foundry supports observability features that improve the performance and trustworthiness of GenAI responses.
• Evaluators are specialized tools in Azure AI Foundry that measure the quality, safety, and reliability of AI responses.
• Some evaluators include:
  • Groundedness: measures how consistent the response is with respect to the retrieved context.
  • Relevance: measures how relevant the response is with respect to the query.
  • Fluency: measures natural language quality and readability.
  • Coherence: measures logical consistency and flow of responses.
  • Content safety: comprehensive assessment of various safety concerns.

Explore generative AI in Azure AI Foundry portal

• Complete the lab to Explore generative AI in Azure AI Foundry portal.