Highlights:
- The newly introduced OCI Generative AI service is a fully managed, cloud-hosted solution that empowers enterprises to harness robust large language models.
- Currently, in beta testing, the OCI Generative AI Agents service employs retrieval-augmented generation techniques to refine generative AI models.
Oracle Corp., the leading database provider, recently introduced its highly anticipated Oracle Cloud Infrastructure Generative AI service. The launch involves several innovations, empowering large enterprises to harness the latest generative artificial intelligence technology advancements.
The latest OCI Generative AI service is a fully managed, cloud-hosted solution that allows enterprises to utilize robust large language models like Meta Platform’s open-source Llama 2 and Cohere Inc.’s proprietary algorithms. These models can be seamlessly integrated with existing systems, enabling companies to automate numerous manual business processes currently carried out by humans.
Oracle states that the OCI Generative AI service supports over 100 languages, offers an enhanced cluster management experience for graphics processing units, and presents flexible options for fine-tuning.
Oracle has announced that its latest service allows models to be accessed through application programming interface (API) calls, catering to tasks such as text generation, summarization, and semantic similarity. Customers can securely integrate generative AI into their current technology stacks, ensuring robust data security and governance protocols.
In June, the company introduced the OCI Generative AI service, positioning it as a viable option comparable to Microsoft’s Azure OpenAI service. This service enables customers to deploy large language models capable of responding to human prompts conversationally.
Even though Llama 2 and Cohere’s AI models are well-known for their strengths, most businesses won’t find much use for them unless they refine and improve using internal data. To achieve this, users can enhance the capabilities of the LLM by using their own data with the help of the OCI Generative AI Agents service.
Currently in beta testing, the OCI Generative AI Agents service utilizes retrieval-augmented generation techniques to refine generative AI models, enabling them to offer more valuable and contextualized responses. This service includes an RAG agent that leverages enterprise search capabilities built on OCI Open Search, assisting LLMs in obtaining the latest information to enhance their responses and predictions.
While the initial version backs RAG via OCI OpenSearch, the company envisions incorporating a broader array of data search and aggregation tools. This expansion includes Oracle Database 23c featuring AI Vector Search and MySQL HeatWave with Vector Store. Additionally, the offering will extend support for prebuilt agent actions across the company’s suite of software-as-a-service applications, encompassing Oracle Fusion Cloud Applications Suite, Oracle NetSuite, and Oracle Cerner.
Oracle said the OCI Generative AI service is available in multiple Oracle Cloud regions and can also be accessed on-premises via the OCI Dedicated Region platform. The service will also be integrated with Oracle’s portfolio of cloud-based business applications, which includes Oracle Enterprise Resource Planning, Supply Chain Management, Human Capital Management, and Customer Experiences. In addition, Oracle announced that its generative AI service will be available through its Autonomous Database Select AI feature.
As part of a related development, Oracle has announced the expansion of its OCI Data Science platform. Next month, it will introduce a new AI Quick Actions feature in beta, offering a no-code pathway to access a range of open-source Large Language Models (LLMs). This feature aims to assist customers in building, training, deploying, and managing LLMs using open-source libraries such as Hugging Face Inc.’s Transformers or PyTorch.
Greg Pavlik, Senior Vice President of AI and Data Management at OCI, emphasized the company’s commitment to applying generative AI to address practical business challenges and facilitate broader technology adoption across enterprises. “To do this, we are embedding AI across all layers of the technology stack by integrating generative AI into our applications and converged database, and offering new LLMs and managed services. Instead of providing a tool kit that requires assembling, we are offering a powerful suite of prebuilt generative AI services and features that work together to help customers solve business problems smarter and faster,” he said.
The new service, with its API-based access, may prove appealing to certain enterprises, given its unusual availability in both cloud and on-premises environments, noted Andy Thurai, Vice President and Principal Analyst at Constellation Research Inc. However, Thurai highlighted significant limitations, including support only for LLMs from Cohere and Meta’s Llama 2. The service is specifically tailored to a narrow set of use cases, such as text generation and summarization.
Thurai said, “In terms of its overall generative AI offerings, Oracle remains far behind Microsoft, Google and Amazon, which provide greater functionality and more flexible deployment options. However, the option to integrate with their ERP, HCM, SCM and CX applications running on OCI could make this offering attractive to a number of enterprises, if it’s priced right. But if not, AWS might win this easily, as its generative AI services are currently far ahead of Oracle’s.”
The introduction of this new service is expected to position Oracle as a viable choice for enterprises engaged in generative AI development. The company is actively working towards establishing itself as a significant player in infrastructure, offering cloud-based access to Nvidia Corp.’s high-performance GPUs.