Oracle today announced the Deep Trust AI Architecture – a practical blueprint for accelerating enterprise AI innovation with trust and security built into the critical layers of the technology stack. This announcement has been positively received by leading industry analysts, as it provides proven methods to enforce trust deep within the underlying infrastructure, where AI models and agents cannot bypass or compromise it.
“Our work at theCUBE Research shows enterprise AI trust requires controls that remain effective when a model makes a mistake. Oracle’s Deep Trust AI Architecture addresses this need by coordinating controls across data, data access, application development and the platform itself. Its emphasis on enforcing access policies and business rules in the database, alongside security and operational services in a shared application runtime, gives enterprises a concrete engineering foundation for more controlled AI deployments. This approach has the potential to reduce the integration work customers must do to assemble and maintain the piece parts. Enterprises are still accountable to define policies, validate results and govern deployment. Oracle is taking a significant step toward closing the gap between what AI can generate and what enterprises can confidently put into production.”
Dave Vellante, Co-Founder and Chief Analyst, theCUBE Research
“Any enterprise deploying AI should have full confidence in what I call the triad of trust: Trust your data, Trust the system that maintains your data, and Trust the answers to data questions that AI provides. Similar to Oracle’s Maximum Availability Architecture, Oracle’s Deep Trust AI Architecture is a blueprint for enterprises that delivers on the triad of trust. It brings an integrated approach to trust instead of customers having to cobble together systems from multiple vendors. With a four-layer stack: Data, Data Access, Application Development and Platform, it’s architected from the onset to provide both Oracle and non-Oracle customers a consistent foundation for trusted outcomes.”
Carl Olofson, Principal, DBMSGuru
“The next frontier of enterprise AI isn’t simply generating answers faster. It’s generating answers that understand the business, improve with experience, and can be trusted to inform consequential decisions. Oracle’s Deep NL2SQL represents a significant leap forward, moving beyond traditional natural-language-to-SQL translation to deliver business-aware intelligence that continuously improves through feedback and verified knowledge. For enterprises, this has the potential to democratize access to critical business insights, accelerate decision-making, and deliver far greater value from their data investments. What’s particularly compelling is how Deep NL2SQL fits within Oracle’s broader Deep Trust AI Architecture, where intelligence, data security, and governance are designed to work together at the source. Deep Trust AI Architecture places Oracle at the forefront of a fundamental shift in enterprise AI, one in which trusted data becomes the foundation for realizing the full potential of agentic AI at enterprise scale.”
Brad Shimmin, Vice President & Practice Lead, Data Intelligence, Analytics, & Infrastructure, Futurum
“Agentic AI changes the trust equation. Agents don’t just answer questions; they query and act on an organization’s most sensitive data, often with little human review. Yet in most environments, security and governance are layered onto data spread across disconnected systems, where policy drifts and auditability breaks down. To be clear, the challenge is no longer building agents. It’s making them trustworthy at scale. Oracle’s Deep Trust AI Architecture addresses this at the data layer, building security, governance, and AI into Oracle AI Database, where much of the world’s critical enterprise data already lives.”
Matt Kimball, Vice President & Principal Analyst, Moor Insights & Strategy
“The enterprise AI race is entering a new phase, and Oracle is moving the competitive battleground from model intelligence to architectural trust. While much of the industry is still grappling with how to secure AI agents, govern their access to enterprise data, and improve the reliability of AI-generated answers, Oracle is addressing these challenges at their source, the database itself. With Deep Data Security enforcing access policies directly where data resides, and Deep NL2SQL bringing business context, continuous learning, and verification to natural-language interactions, Oracle is fundamentally changing the equation for enterprise AI. What makes Deep Trust AI Architecture particularly compelling is that these capabilities aren’t isolated innovations. They’re part of a unified architectural strategy designed to make trust intrinsic to how AI accesses, understands, and acts on enterprise data. In my view, Oracle is raising the bar for the entire industry, taking agentic AI to the next level by grounding intelligence more deeply in enterprise data, strengthening trust, and enabling AI to operate with greater precision, business context, and scale. While others are still grappling with how to connect AI intelligence with enterprise data and governance, Oracle is engineering that convergence directly into the foundation of the agentic enterprise.”
Steven Dickens, CEO, HyperFRAME Research
“As AI agents become increasingly integral to enterprise operations, CxOs are focused on scaling AI innovation while strengthening security, governance, and trust. Oracle is addressing this challenge head-on with its Deep Trust AI Architecture, applying the proven architectural principles of enterprise mission-critical workloads to the agentic AI era. By bringing together Deep Data Security, Deep Trust Data APIs, and an innovative approach to NL2SQL that enables AI to understand business context, learn from mistakes, and verify improvements, Oracle is advancing a more comprehensive foundation for trusted enterprise AI. Combined with Oracle’s fundamental advantage of bringing AI to enterprise data without requiring it to move, this gives customers a compelling path to accelerate AI adoption with greater confidence, control, and peace of mind.”
Holger Mueller, Vice President and Principal Analyst, Constellation Research
“Oracle is setting a new benchmark for trusted agentic AI, bringing enterprise data, business context, and continuous intelligence together in a unified architecture. Deep NL2SQL is a breakthrough, moving beyond basic SQL generation to deliver business-aware AI that learns from feedback and verifies improvements for greater accuracy. With its Deep Trust AI Architecture, Oracle is raising the competitive bar, transforming how enterprises turn governed data into actionable intelligence. This is a major leap forward in making agentic AI more intelligent, trustworthy, and scalable across the enterprise.”
Ron Westfall, Vice President, HyperFRAME Research
“Enterprise AI doesn’t need another layer of complexity. It needs an architecture that makes agents useful without compromising the data they depend on. Oracle’s Deep Trust AI Architecture takes on that problem at its source, building security, governance, and business context directly into the data foundation. That matters. An AI agent that can access data but can’t understand its meaning, respect its permissions, or act within business rules is a liability. Oracle’s approach has the potential to change the competitive equation, forcing vendors to rethink architectures that bolt those capabilities on after the fact. If Oracle delivers on that promise, it could set the enterprise standard for agentic AI for the next decade.”
Marc Staimer, Senior Analyst, theCUBE Research
“Most enterprise security still assumes that the application in front of the data enforces the rules. Agents that write their own queries and act across systems break that assumption. Oracle’s Deep Trust AI Architecture responds by moving enforcement into the data layer. Deep Data Security binds access to the identity of the user an agent acts for, SQL Assertions apply business rules no matter who or what initiates a change, and Deep Trust Data APIs expose governed business objects instead of raw SQL. Trust becomes a property of the architecture rather than of each application or agent, and that lets organizations put agentic AI to work on their most valuable data with policy enforced on every request.”
Alexei Balaganski, Lead Analyst & Chief Technology Officer, KuppingerCole Analysts