Oracle AI Database Enhances with New Agentic AI Features
Oracle AI Database Enhances with New Agentic AI Features
It is well-known that AI works well only when powered by a good data foundation: many enterprise artificial intelligence projects have failed to meet expectations precisely due to information silos or lack of well-structured data. The solution to this problem proposed by Oracle is as simple as it is functional: why not integrate AI where the data resides? For this reason, Larry Ellison's company has enhanced the Oracle AI Database by incorporating agentic AI capabilities.
AI Agents Enter the Oracle Database
According to Oracle experts, there are numerous benefits to having AI agents capable of operating natively within the database. Among these is the fact that it is no longer necessary to create and manage data movement pipelines. This consequently reduces the complexity of AI infrastructures, cutting costs and, last but not least, enhancing security. The data, in fact, does not need to be copied into data lakes and repurposed to allow AI to operate on it, thus reducing both the required storage space and avoiding transfers to other systems that could inevitably introduce critical cybersecurity issues.
New Features
The Autonomous AI Vector Database introduces native support for vector data while maintaining the typical Oracle platform settings: a single environment that combines ease of use for developers and data scientists with enterprise-level requirements for security and scalability. Access is initially limited, featuring freemium models and low-cost development environments, extendable with direct upgrades.
On the agentic AI front, the AI Database Private Agent Factory is a no-code environment that allows users to create and deploy agents directly where data resides, whether on public cloud or on-premise. This approach eliminates the orchestration of external systems and maintains data control within the organization. Preconfigured agents for structured analysis, knowledge management, and advanced search are included.
At the architectural level, Oracle also introduces the Unified Memory Core, a layer that centralizes the context of the agents and enables low-latency processing across multiple data formats, from vectors to JSON, from graphs to relational data. The goal is to reduce fragmentation and bring together functionalities that usually require multiple systems into a single engine.
A Database Designed with Security in Mind
Given that AI infrastructures have quickly become a favored target for cybercriminals, Oracle has focused heavily on data protection. An example of this is Oracle Deep Data Security, a feature that introduces a granular access model based on the end-user. Each user or AI agent operating on their own can access only the data for which they are authorized. Rules can be defined in detail by role and function, for example, separating information visible to sales, finance, logistics, or customer care. Security is shifted directly into the database and no longer resides in the application code. This approach allows for the application of controls that can be updated over time and for responding to new AI-specific threats, such as prompt injection, while maintaining consistency and centralized control.
Also introduced is the Private AI Services Container, an environment that allows AI models to run in isolation without sharing data with external vendors. The container can be deployed on public, private, or on-premise clouds, even in air-gapped contexts, and enables intensive workloads, such as embedding generation, without moving data outside the corporate perimeter.
Finally, with Trusted Answer Search, Oracle aims to reduce the uncertainty typical of generative models. Instead of directly relying on a LLM for an answer, the system uses vector search to link the user's query to verified and previously produced content. The result is a more deterministic approach, designed for contexts where accuracy and traceability are critical requirements.
"The next phase of enterprise AI will be defined by customers' ability to use AI in mission-critical production systems to safely deliver revolutionary innovations, insights, and productivity," says Juan Loaiza, Executive Vice President, Oracle Database Technologies at Oracle. "With Oracle AI Database, customers are not just managing data; they are activating it for AI. By integrating AI and data together within the same architecture, we help customers rapidly create and manage agentic AI applications capable of querying and securely operating on business data in real time, with securities robust enough for the stock market, across all major cloud and on-premise environments."