23 November 2024, Sat |
3:10 AM
Oracle has announced the broad availability of its Oracle Cloud Infrastructure (OCI) Generative AI service, unveiling new features designed to streamline the utilization of cutting-edge generative AI technology by enterprises. The OCI Generative AI service, a fully managed and comprehensive offering, now incorporates extensive language models (LLMs) from Cohere and Meta Llama 2, catering to a diverse range of business applications. Notable enhancements to the service comprise support for over 100 languages, an enhanced GPU cluster management interface, and flexible customization options. This service is accessible both in the Oracle Cloud and on-site through OCI Dedicated Region.
Greg Pavlik, Oracle’s Senior Vice President for AI and Data Management, emphasized Oracle’s commitment to practical AI applications in the business sector. The company’s strategy involves embedding AI throughout its technological framework, integrating generative AI into applications and databases, and providing new LLMs and managed services, all underpinned by a fast, cost-effective AI infrastructure. Oracle aims to deliver a cohesive suite of ready-to-use generative AI services and functionalities, streamlining the resolution of business challenges for its customers.
Oracle is also enhancing the adaptability of generative AI models. The latest models from Cohere and Meta Llama 2, available as a managed service, are designed to address text generation, summarization, and semantic similarity tasks through simple API integrations. This allows for seamless, secure incorporation of generative AI into existing technology infrastructures, with robust data security and governance measures in place.
Customers can further tailor these models to their specific needs using retrieval augmented generation (RAG) techniques, ensuring the models are attuned to their unique business contexts. The OCI Generative AI Agents service, currently in beta with a RAG agent, combines LLMs with enterprise search capabilities built on OCI OpenSearch. This enables users to interact with a variety of enterprise data sources through natural language queries, providing up-to-date, referenced results.
The initial beta supports OCI OpenSearch, with future releases set to expand compatibility to additional data search and aggregation tools, and integration with Oracle Database 23c and MySQL HeatWave. Oracle also plans to introduce pre-built agent actions for its suite of SaaS applications, including Oracle Fusion Cloud Applications Suite, Oracle NetSuite, and sector-specific solutions like Oracle Health.
Ritu Jyoti from IDC lauds Oracle’s approach to bringing generative AI directly to customer workloads and data, bypassing the need for separate vector databases. She highlights Oracle’s unified generative AI architecture, which is integrated across its ecosystem, from Autonomous Databases to Fusion SaaS applications. This integration allows customers to utilize generative AI with existing data, whether in cloud data centers or on-premises, streamlining deployment within current business operations.
Oracle’s comprehensive cloud application suite, combined with its top-tier AI infrastructure, establishes a robust foundation for customer trust. By incorporating generative AI across its cloud applications (ERP, HCM, SCM, CX), Oracle enables clients to harness recent innovations within their established business processes. Furthermore, Oracle is integrating generative AI capabilities into its database offerings, empowering customers to develop AI-driven applications. The Autonomous Database Select AI facilitates rapid application development and innovative business solutions by blending private enterprise data with the power of generative AI.
To assist customers in building, training, deploying, and managing LLMs using open-source libraries like Hugging Face’s Transformers or PyTorch, Oracle is expanding OCI Data Science capabilities. The upcoming OCI Data Science AI Quick Actions feature, entering beta soon, provides no-code access to a variety of open-source LLMs, including prominent providers like Meta or Mistral AI.
