Oracle offers Generative AI RAG Agent

Oracle has announced the general availability of Oracle Cloud Infrastructure (OCI) Generative AI (GenAI) Agents with retrieval-augmented generation (RAG) capabilities and enhanced Oracle AI innovations that help customers turn their data into a competitive advantage by making it easier to apply AI to real-world business operations.

The first of a series of OCI GenAI Agents, the RAG Agent provides out-of-the-box RAG capabilities, enabling customers to get started while avoiding manual processes such as agent planning, retrieval, reranking, generation, and integration. It also provides a self-check to reduce hallucinations, enabling customers to adopt RAG technologies to streamline their business processes without spending cycles on research and development. OCI GenAI Agents enable customers to access Oracle Database 23ai AI Vector Search and run fast similarity queries on enterprise data stored in the database. For customers who have a subscription to Oracle Database 23ai on OCI, the GenAI Agents service adds an automation layer to execute RAG and similarity search functions without having to move data to a separate vector database.

For organizations looking for open source solutions for their generative AI workloads, OCI GenAI Agents service also supports OCI Search with OpenSearch.

“AI is driving breakthroughs and efficiencies at an unprecedented pace, leading to new business models, applications, and innovation across industries,” said Greg Pavlik, executive vice president, AI and Data Management Services, Oracle Cloud Infrastructure. “With the OCI Generative AI Agents service, Oracle Database 23ai customers on OCI can subscribe to a fully managed RAG service that enables them to talk to their data. It’s AI designed for companies looking to put AI into their real-world production environments quickly—by bringing AI to their data.”

Key use cases for OCI GenAI Agents include:

Call center optimization: Helps operators increase customer satisfaction through more accurate responses and a higher volume of query resolution.

Expedite legal research: Helps researchers find answers faster by conversing with AI rather than manually searching court record databases.

Revenue intelligence: Helps finance teams understand customer purchase history and trends by asking natural language questions instead of running reports.

Recruit qualified job candidates: Helps recruiters source potential new hires more easily by typing in natural language rather than constructing a database query.

To provide versatile and accessible AI solutions for developers and business users, Oracle is also introducing AI innovations to the following:

OCI Generative AI: Helps users seamlessly integrate versatile language models into a wide range of use cases, including writing assistance, summarization, analysis, and chat. To enable customers to harness the latest open source AI technology, OCI Generative AI now provides access to Meta’s Llama 3.1 models in sizes ranging from 405 billion parameters for AI use cases that require state-of-the-art capabilities to 70 billion parameters for more targeted workloads at a lower price. In addition, OCI Generative AI supports the Cohere Command R, Command R+, and Embed models.

OCI Data Science: Helps teams of data scientists access open source tools to build, train, deploy, and manage AI models. OCI Data Science provides several new updates to the no-code AI Quick Actions feature, including OCI Ampere A1 shape support and the ability to bring any model from Hugging Face into OCI Data Science with just a few clicks.

OCI Language: Helps developers perform sophisticated text analysis and machine translation at scale. New multilingual models are designed to seamlessly support over 100 languages, while a new Protected Health Information (PHI) feature extends the current Personally Identifying Information (PII) service to detect PHI entities and redact them from textual data.

OCI Document Understanding: Helps developers automate manual business processing tasks with pre-built AI models. With newly trained models for the healthcare field, developers are able to identify and extract important data on medical identification cards while increasing efficiency and reducing manual errors.

OCI Vision: Helps developers analyze videos frame-by-frame to identify objects, labels, text, and faces. It provides data about detection timestamps and bounding box coordinates for objects to support use cases, including digital asset management, visual anomaly detection, safety and surveillance, and ad tracking and placement.

OCI Speech: Helps users transcribe speech to text and synthesizes speech from text with natural voices and a new real-time transcription capability that includes custom vocabularies support.

Oracle Code Assist: Will help developers boost velocity by providing intelligent suggestions to help them build and optimize applications written in modern programming languages, including Java, Python, JavaScript, SuiteScript, Rust, Ruby, Go, PL/SQL, C#, and C.