Oracle has announced new AI capabilities for its cloud-based database, analytics and development platform, with a particular focus on the MySQL database managed by HeatWave. This was revealed during the Oracle CloudWorld conference.

At its CloudWorld conference, Oracle announces new AI capabilities within HeatWave and Analytics. The HeatWave service, which runs on MySQL, will have improved options to bring AI applications into production more easily and quickly. Built-in large language models (LLMs) reduce the need for specific AI expertise and limit manual integration and security complications. According to Oracle, HeatWave on Amazon Web Services (AWS) enables customers to replace up to six AWS services with a single HeatWave instance.

Additionally, users can automatically create vector stores and vector embeddings. HeatWave offers the ability to use LLMs directly in the database, both on CPUs and with Amazon Bedrock models. Users can also access documents in Amazon S3 through conversations, which Oracle claims is faster than alternative services such as Snowflake, Databricks, and Google.

Extending Lakehouse functionality

A key addition is the Lakehouse architecture on HeatWave, which stores and processes data from Amazon S3. This makes it possible to search structured, semi-structured and unstructured data at high speed without copying the data to a database. Object storage support has been improved so that query results can be shared and stored in object storage. HeatWave can also be used for MapReduce applications.

In addition, new AI functionalities have been added. For example, HeatWave now supports documents in 27 languages ​​in the Vector Store, and users can search for similar documents across scanned files using optical character recognition (OCR). The batch process for LLM inference has been improved to handle multiple queries at once.

Finally, through AutoML, Oracle is introducing additional machine learning capabilities available at no additional cost, including four times larger training datasets and the ability to identify topics in large text files through topic modeling. It also added semi-supervised log anomaly detection to help better detect anomalies and refine predictions.

With these updates, Oracle is committed to simplifying AI development and integration for its customers. HeatWave is now also available as part of Oracle Cloud Infrastructure (OCI)’s Always Free service.