The latest version of Oracle’s groundbreaking database is Database 23ai, a next-generation converged database that aligns with Oracle’s commitment to spearheading developments in AI.
What’s New
Mendelsohn cites Database 23ai as being up there with the top database releases ever launched at Oracle. The launch focuses on three core areas: AI for data, features for developers and analysts, and further enhancing mission-critical features.
Database 23ai — which is now generally available — delivers 300 plus features and thousands of enhancements but, for the purposes of this analysis, we’ll focus on the specific AI developments. The goal of Database 23ai is to simplify AI for data for all personas, from developers to business users, for all applications, and all workloads.
Six Core AI Approaches
Database 23ai incorporates six key AI domains, or approaches, to enable what Oracle describes as the “AI for Data revolution.” These include Algorithmic AI, AI Vector Search, Augmented Generative AI, Distributed AI, Storage AI, and AI Developer Tools. Let’s run through these elements one by one.
Algorithmic AI: This technology has been built into the Oracle Database for decades, with the company continually improving it year on year. The trouble is, that algorithmic AI is complicated to build. That’s why the focus has been on introducing AutoML to simplify the model-building process. With Database 23ai, the direction is to introduce GenAI to further simplify algorithmic AI.
AI Vector Search: Database 23ai introduces a new data type called vectors that take the features of a data asset and encodes them numerically. These vectors are used for AI Vector Search on various data formats such as images and documents but the most value is released when you combine vector search with relational search on business data.
With Database 23ai, you can carry out this action in the same query with just a few lines of code or via natural language; no AI knowledge is needed, and basic coding experience is all that’s required to carry out the function. Ultimately, Oracle is providing a fully combined solution by enabling users to store vectors alongside other data types.
Augmented Generative AI: AI Vector Search also has an impact on GenAI in the 23ai database, improving the output from queries by augmenting prompts with private database content through Retrieval Augmented Generation (RAG). For example, a user could generate personalized financial advice reports using relevant financial data and trends specific to their transaction history and current financial market conditions. This would provide them with actionable insights tailored to their spending patterns, saving goals, and investment preferences by combining their historical transactional (structured) data with up-to-date, general financial market information.
Distributed AI: GoldenGate 23ai enables vector search from Oracle data sources or third-party vendors without moving the data or upgrading existing data stores. The technology enables users to replicate data from anywhere and run it on the 23ai database, helping to eliminate AI silos and enabling users to operationalize AI Vector Search across their data ecosystem.
Storage AI: Exadata 24ai Software in Oracle Database 23ai enables offloading of vector search to Exadata storage, leveraging the technology for faster search.
AI Developer Tools: Database 23ai includes a range of developer tools designed to make it easier for developers to embed AI into applications. Most notably, this includes an interface with LangChain, an open-source project for the development, observation, and deployment of LLM-driven applications.
Closing Thoughts
The fine-tuned focus on AI in Oracle Database 23ai is impressive. And this advance is set to continue with more capabilities for Database 23ai on the horizon.
Enchant Apps has a team of Oracle Cloud experts across all modules. If you are interested in learning more or have a project where you need Oracle cloud expertise, feel free to contact us!