The offering provides RAG capabilities for customers that use Oracle databases. Use cases include legal research, finance and customer service support.
Oracle on Tuesday made generally available its Oracle Cloud Infrastructure GenAI Agents platform with retrieval-augmented generation.
OCI GenAI Agents with RAG was first introduced in January. RAG is a framework that helps LLMs retrieve current information better and reduce hallucinations.
OCI GenAI Agents with RAG provides Oracle customers with out-of-the-box RAG capabilities, helping them avoid manual processes like agent planning and data retrieval, according to the vendor. The platform also self-checks outputs to reduce hallucinations.
Geared for customers
While Oracle introduced OCI GenAI Agents earlier in the year, interest in AI agents, or autonomous systems that perform tasks without human intervention, is growing.
What makes OCI GenAI Agents with RAG particularly interesting is that it is geared toward Oracle customers that use Oracle databases and need to validate or complement their AI models with RAG, Constellation Research analyst Holger Mueller said.
"Oracle needs to make RAG as easy as possible for enterprises so enterprises keep using their databases," Mueller said. "If this was hard, enterprises could move the data out of the databases, something no database vendor wants."
If OCI GenAI Agents with RAG succeeds, it will likely help Oracle do well in the arena of GenAI and databases as GenAI continues to grow rapidly, he added.
AI Agents
While Oracle defines its agents as assistants that help build an application, that's not only what agents are, said Mark Beccue, an analyst with TechTarget's Enterprise Strategy Group.
Oracle needs to make RAG as easy as possible for enterprises so enterprises keep using their databases.
Holger MuellerAnalyst, Constellation Research
"An agent could be something that does a task," Beccue said, adding that an agent could be a call center assistant working to increase customer satisfaction with more accurate responses.
OCI GenAI Agents with RAG could be useful in various other scenarios, including helping researchers find answers faster, but AI agents could do even more, he said.
"They're an advanced form. They're not a building block; they are a result," Beccue continued.
Other than OCI GenAI Agents, Oracle -- at its CloudWorld 2024 conference in Las Vegas -- introduced AI-centric generative development for enterprises. With the AI development environment, developers can generate applications that use AI-powered natural language interfaces and human-centric data, Oracle said.
The AI vendor also revealed new features in some of its offerings. They include the following:
OCI Generative AI, which helps users integrate language models into different applications, now provides access to Meta Llama 3.1 models. It also supports Oracle GenAI partner Cohere's Command R, Command R+ and Embed models. Cohere also revealed that it is partnering with the Japanese consulting firm Nomura Research Institute. Together they will launch a new financial AI platform.
OCI Speech is an offering that helps users transcribe speech to text and synthesizes speech from text with natural voices. It now has a new real-time transcription capability that includes custom vocabularies support.
Oracle Code Assist helps developers boost velocity by providing suggestions to help them build and optimize applications written in different programming languages.
The new updates and offerings show ways Oracle is focusing on various vertical industries, Gartner analyst Sid Nag said.
"They are building industry-specific LLMs or SLMs [small language models] that are industry- or domain-specific … private, secure, and run in a walled garden [private] environment," Nag said.
Esther Ajao is a TechTarget Editorial news writer and podcast host covering artificial intelligence software and systems.
