- A new focus in GenAI development is readying data for effective use
- Oracle’s new GenAI Agents simplify deployment with direct data integration and built-in accuracy checks
- Despite fierce competition from AWS, Google and Microsoft, Oracle is establishing itself as a "strong competitor" among CSPs in the AI space
Early excitement over generative artificial intelligence (GenAI) has been tempered by the realization that without strong data infrastructure, even the most advanced AI models can struggle to deliver real business value.
“As enterprises accelerate AI/generative AI adoption, they face ongoing challenges in preparing data for AI,” said Gaurav Dewan, Research Director at Avasant. Those challenges include breaking down data silos, consolidating diverse data sources and ensuring data accuracy and recency, Dewan told Fierce Network.
With GenAI challenges stacking up, a 2024 report from Lucidworks found only 63% of companies plan to increase their AI investments this year, a significant drop from 93% in 2023. Sensing the threat, cloud service providers (CSPs) are responding with solutions designed to address these pain points head-on.
Oracle is one such company with its newly launched GenAI Agents, AI-powered tools within Oracle Cloud Infrastructure that can help enterprises automate tasks and improve decision-making by integrating AI with their data. These agents use retrieval-augmented generation (RAG) capabilities, which combine GenAI with data retrieval processes, allowing businesses to create AI models that are both creative and contextually accurate.
Oracle's database services are used by hundreds of thousands of enterprises worldwide. And the company’s approach to data “for years” has been to create a converged database that has native support for all modern data types built into one product, according to Sherry Tiao, senior manager of product marketing for AI and Analytics at Oracle.
“Great AI starts with great data,” Tiao said.
That means enterprises can make complex queries to the GenAI Agents, Tiao explained, and the agents will take multiple data types and sources into consideration, making answers even more contextually relevant.
In a sea of competition, Oracle’s AI advantage is in its “existing access to the data,” said Avasant Senior Research Director Dave Wagner.
“It will be much harder for a third party to create an AI to pull insights out of Oracle's core products than it is for Oracle to keep upgrading its own offerings with AI to do the same thing.”
Another “unique strength” for the CSP, Dewan noted, lies in enabling customers to use RAG and advanced search functions without needing to transfer data to a separate vector database, leveraging its large customer base.
Upping the ante
"Despite being a late entrant, Oracle has positioned itself to compete with other major GenAI players by offering agents specifically tailored for its existing database customers," Dewan noted. The provider's integration of data with AI, particularly through its Oracle Database 23ai launched in May 2024, has set it up as a “serious competitor" to leading cloud service providers.
Like AWS, Google and Microsoft, Oracle uses vector search to search documents, images and relational data based on their conceptual content, rather than relying solely on specific words, pixels or data values. “It employs natural language processing (NLP) on private business data to deliver more accurate and relevant search results,” Dewan added. “The vector search feature is paired with indexing capabilities for fast data retrieval.”
However, competition continues to advance. Google, for example, introduced a new vector search UI in November 2023, allowing developers to create and deploy indexes directly from the UI without coding. It also reduced indexing latency from hours to minutes for smaller datasets.
For its part, Microsoft offers multiple built-in options for its Azure AI Search, such as using a vector index, retrieval augmentation or even employing a search index as a vector store in a prompt flow. AWS, meanwhile, made vector search for Amazon MemoryDB publicly available in July 2024.
“These developments reflect a rapidly evolving competitive landscape in AI-powered data integration,” Dewan concluded.