When Microsoft announced the new AI Studio, it was easy to mistake it by any of the other studios available.

All AI resources in Azure use a different studio to perform their tasks, so this gets a little confusing.

The AI studio, on the other hand, is one of the most advanced and has many great new features announced during Microsoft Build.

AI Studio

AI Studio is focused on generative AI. Other AI areas, such as document intelligence, vision and others, may still work better with other studios.

AI Hub Concept

The concept of an AI Hub is new, and we can manage them from AI Studio.

The AI Hub is a central object intended to hold AI projects and be linked to AI resources. The AI resources can be either an AI Service, which by itself already aggregates many AI services, or an Azure Open AI, plus AI Search.

This is AI Studio managing AI Hubs:

This is the wizard to create a new AI Hub:

Central view of AI Projects

We create projects to build our AI work. The AI Studio and AI Hub provide a central view of the projects, and this is a great advance. It was difficult to manage AI projects before this.

AI Studio Hierarchy View

AI Studio has a hierarchy work:

• On top of the hierarchy, you can view AI Hubs available and manage them
• At an AI Hub level, we can view the projects in the AI Hub and manage them

• At the project level, we can make implementations.

Model Catalog

The model catalog allows us to choose among numerous models and deploy them to our Azure environment.

Model Benchmarks

You can compare a benchmark between models and choose the model according to their benchmark performance. In this way you can find the right model for each scenario.

AI Studio and Prompt Flow Integration

I wrote an article about Azure ML Prompt flow before.

During Microsoft Build, almost all, if not all, AI sessions had some relation with prompt flow in one way or another.

The AI Studio brings an integration with prompt flow and the new features to debug prompt flow solutions are another big announcement from Build.

Besides the integration, it seems very interesting that we can make prompt flow work using serverless deployments. I hope this feature evolves even more.

Summary

This is only the top of the iceberg, the news related to AI announced during Build requires a lot of time to be studied.