Oracle is embedding AI across every layer of its technology stack. In an exclusive interaction with AIM, Palanivel Saravanan, vice president of cloud engineering at Oracle India, said that the company is integrating AI into its infrastructure, platforms, databases, and SaaS applications.
“Oracle was the first to launch bare metal instances with GPUs to power AI model training, including large language models (LLMs),” Saravanan said. The company has built a supercluster with 130,000 GPUs housed in a large-scale data centre, making it one of the world’s largest AI training infrastructures.
AI in Databases
Beyond infrastructure, Oracle provides an AI-ready platform that allows developers and customers to use AI for diverse purposes, such as summarisation, visualisation, and identification. The company has built AI as a service, ensuring that it is embedded in the data layer and applications.
Oracle’s Select AI allows users to interact with databases using natural language without needing to know SQL or the underlying data structure.
“Users can query the database in 50 languages and extract insights instantly, making decision-making more efficient for businesses,” Saravanan explained. This feature benefits industries such as banking, where managers can access customer transactions and make real-time decisions without complex coding.
Moreover, Oracle has embedded LLMs directly into MySQL HeatWave, enabling AI to operate closer to enterprise data. This approach ensures faster response times and improves efficiency.
The service, known as Oracle HeatWave GenAI, integrates LLMs and vector processing within the database, allowing users to leverage generative AI without requiring AI expertise or data movement.
“We are moving beyond structured data by embedding vector search capabilities, enabling businesses to manage and query unstructured data such as customer feedback and location details more efficiently,” Saravanan said.
He further added that their latest Database 23ai has over 300 AI features.
“Every database customer is taking advantage of 23ai. Cognizant, the largest Oracle Applications user, is a case in point. They use our ERP, financial system, and various other applications,” said Saravanan, adding that they now want to build an agent to interact with the database and get the right queries responded to much faster than with traditional methods like report building.
He explained that 23ai with autonomous capabilities introduces machine learning functionality that allows the database to perform performance fine-tuning autonomously.
This means that tasks such as applying security patches and scaling the database—whether up or down in response to repairs—are handled automatically.
However, Oracle faces competition from Azure Microsoft, which offers Azure Cosmos DB, a fully managed NoSQL, relational, and vector database that integrates AI capabilities for tasks like retrieval-augmented generation (RAG). Azure also provides Azure Cognitive Search, which leverages AI for advanced search and data analysis.
According to Saravanan, Oracle, like other cloud services, lets companies bring their own LLMs to the cloud. The company handles the training and updates for these models. When they are added to MySQL, Oracle offers AI Quick Actions, giving customers a faster and more dependable way to use the models.
AI Quick Actions is a feature within Oracle Cloud Infrastructure (OCI) Data Science that simplifies the deployment, evaluation, and fine-tuning of LLMs.
RAG and AI Agents
Saravanan said Oracle is incorporating AI agents into its SaaS applications across ERP, HCM, SCM and CX. “These agents are built for specific use cases, such as customer support in financial services or safety compliance in manufacturing,” Saravanan said.
Notably, last year at CloudWorld 2024, Oracle announced the integration of over 50 specialised AI agents into its Oracle Fusion Cloud Applications Suite.
The company is also advancing RAG, allowing businesses to bring proprietary data closer to AI models. “RAG enables enterprises to generate AI responses that are contextualised with their own data, ensuring accuracy and relevance,” Saravanan said.
“AI works best when the data is best,” he concluded, highlighting Oracle’s commitment to delivering AI solutions that maximise business value.
