Skip to content

What's new in AI

Analytics

Amazon S3 Tables is a new storage class that stores your data in a tabular format using Apache Iceberg. This improves query performance and offers 10x more transactions per second. The thing integrates pretty well with other data services, like Glue and Athena.

S3 Queryable Object Metadata (preview) stores metadata in Iceberg tables. In that table, all metadata about S3 objects is stored and allows you to easily query the data using for example Athena.

spark.sql("SELECT * FROM bucket.aws_s3_metadata.my_first_table")

SageMaker

The Next Generation of SageMaker wants to make SageMaker the center of your data, analytics and AI. The Studio is unified heyooo. SageMaker Lakehouse unifies all your data across S3 and Amazon Redshift. This thing is powered by Glue Data Catalog (I still need to research this thing).

This dude is talking so fast, I can't follow, but he mentions Glue ETL, so figure that out, seems cool. Also do some proof of concept with SageMaker Studio, seems easy enough to use.

Generative AI

Bedrock Data Automation helps developer to process and extract information from unstructured data. It has some built-in safeguards such as visual grounding with confidence scores. After the data is extracted, the results are stored in S3. There's also Bedrock Knowledge Bases, the fully managed service to create RAG applications, whatever that may be. More on RAG applications

Amazon Nova is the new generation foundation model. This model is built to rank it with the top tier models available on the market, but at a competitive price.