In this episode of Cloud Wars Live, Bob Evans sits down with Tirthankar Lahiri, Senior Vice President for Mission-Critical Data and AI Engines. Lahiri explains how agentic AI is transforming enterprise applications from simple question-answer systems into action-driven platforms that can reason, remember, and securely execute tasks. He details Oracle’s strategy around unified agent memory, private agent factories, deep data security, and open development standards, all designed to help customers build scalable, secure, and flexible AI systems without added cost.

AI Built Securely

The Big Themes:

Agentic AI Becomes Action-Oriented: Tirthankar Lahiri explains that agentic AI represents the next major step beyond generative AI. While generative AI focused largely on answering questions and producing content, agentic AI is designed to take action. It allows businesses to build systems that can reason, decide, and execute tasks autonomously. Oracle sees this as the future of application development, where AI becomes embedded into workflows rather than functioning as a standalone tool.
Oracle Builds AI Directly Into the Database: Rather than forcing customers to move data across multiple isolated systems, Oracle’s approach is to bring AI directly to the data. Lahiri argues that data is the “ground truth” and moving it creates technical debt, silos, inefficiency, and security vulnerabilities. Oracle’s converged database architecture supports multiple data types, including relational, graph, spatial, and vector, inside one unified environment. This eliminates the need for separate repositories and allows AI agents to access all relevant context without fragmentation.
Deep Data Security Protects Against AI Risks: Lahiri strongly emphasizes that traditional application-layer security is no longer enough in the age of AI. Since AI can generate SQL and potentially bypass interface restrictions through prompt injection, businesses must secure data directly at the source. Oracle calls this “deep data security.” He uses the analogy of protecting valuables in a safe bolted to the floor rather than simply locking the front gate. Even if someone gets inside the house, the valuables remain protected. Similarly, Oracle enforces security policies at the database level, ensuring agents can only access data users are authorized to see.

The Big Quote: "You need to secure data. Need to lock your valuables into the safe deep inside the house."

More from Tirthankar Lahiri and Oracle:

Connect with Lahiri on LinkedIn or learn more about Oracle AI Database.

Visit Cloud Wars for more.

Tirthankar Lahiri on Why Agentic AI Must Live Inside the Database | Cloud Wars Live

Cloud Wars Live with Bob Evans