Setting Up Your First Project in Azure AI Foundry: Step-by-Step Guide

Are you ready to get your hands dirty with Azure AI Foundry? 

In this blog, I will walk you through everything you need to get started,  from creating your Azure account to launching your first AI project. I will start of with the pre-requisites required before giving you the detailed steps.

Prerequisites

• An active Azure subscription (free tier works for getting started)

• Basic familiarity with cloud concepts

• A web browser and internet connection

Step 1: Access Azure AI Studio

Navigate to https://ai.azure.com/  and sign in with your Microsoft account or Azure credentials. You'll be greeted by the Azure AI Foundry home screen, which serves as your central hub.

Step 2: Create a Hub

A Hub is the top-level organisational unit in Azure AI Studio. Think of it as your enterprise workspace container. To create one:

• Click 'New Hub' from the home screen

• Provide a name, select your Azure subscription, and choose a resource group

• Select a region closest to your users for lower latency

• Configure networking settings (for enterprise, choose private endpoint)

• Click 'Create' and wait for deployment (typically 2-3 minutes)

Step 3: Create a Project

Within your Hub, create a Project  which is where your actual AI work lives. Projects are isolated environments that can have their own resources, team members, and configurations.

• Navigate into your Hub and click 'New Project'

• Name your project and optionally add a description

• Assign team members with appropriate roles (Owner, Contributor, Reader)

• Click 'Create Project'

Step 4: Explore the Model Catalog

With your project open, head to the Model Catalog in the left navigation. Here you'll find hundreds of models organized by category. For your first exploration:

• Search for 'GPT-35' and click on it

• Review the model card and see it includes benchmarks, pricing, and capabilities

• Click 'Deploy' to create a model endpoint

Step 5: Try the Playground

The Playground is your sandbox for testing models. Select your deployed model and start chatting. You can adjust system prompts, temperature, and max tokens to see how they affect outputs.

Step 6: Create Your First Prompt Flow

Prompt Flow is where the real power lies. 

Navigate to Prompt Flow and click 'New Flow'. 

Choose a template like 'Chat with Wikipedia' to see a pre-built example of a RAG pipeline in action.

Below video shows you how to create a project and deploy and model and try the playground

Best Practices for New Users

• Start with a well-defined use case before choosing a model

• Use the evaluation tools to benchmark model quality early

• Tag your resources consistently for cost management

• Set up budget alerts to avoid unexpected charges

Conclusion

In just a few steps, you now have a fully functional Azure AI Studio project ready for development. The platform's intuitive interface makes it easy to go from zero to experimenting with state-of-the-art AI models in minutes.