AutoML feature in SageMaker Unified Studio

Regarding your questions about AutoML in SageMaker Unified Studio:

1. In the new Amazon SageMaker Unified Studio, the Autopilot UI (AutoML) has been integrated into the SageMaker Canvas application. The classic Autopilot UI that existed as a standalone feature in Studio Classic is no longer available directly in the updated Studio experience. To use AutoML capabilities via a user interface, you'll need to use the SageMaker Canvas application within Unified Studio. Canvas provides a no-code AutoML experience for tasks such as data preparation, feature engineering, algorithm selection, training, tuning, and inference.

If you prefer to use Autopilot as you did before, you have two options:

• Continue using the Autopilot UI in Studio Classic (the previous Studio experience)
• Use the AutoML API references in any supported SDK if you have coding experience

2. Regarding using AutoML with data from your PostgreSQL database without S3 permissions: The documentation provided doesn't specifically address using AutoML with direct database connections without S3. The examples shown in the documentation for AutoML jobs typically reference S3 locations for input data (training and validation). The AutoML step in pipelines, for instance, requires inputs to be specified as S3 locations.

Based on the available information, it appears that S3 is the standard storage mechanism for AutoML jobs in SageMaker. If you're using SageMaker without S3 permissions, you might face limitations when trying to use AutoML features, as the standard workflow seems to involve S3 for data storage.
Sources
Amazon SageMaker Unified Studio - Autopilot & Model Monitor | AWS re:Post
Migrate the UI from Studio Classic to Studio - Amazon SageMaker AI
Add a step - Amazon SageMaker AI
Automated ML, no-code, or low-code - Amazon SageMaker AI
AutoML - Automated Machine Learning - Amazon Web Services