Oracle announced the general availability of Oracle APEX AI Application Generator on Tuesday as part of Oracle APEX 26.1. The tool runs on OCI Autonomous AI Database Serverless and introduces a new approach to generating enterprise applications with AI.
The generator produces complete data-centric applications from natural language descriptions. Rather than generating thousands of lines of source code for developers to review and maintain, the system outputs APEXlang - a structured Application Definition Language that Oracle APEX compiles and executes within its managed runtime.
"The question is no longer whether AI can help developers build applications; it clearly can," said Michael Hichwa, senior vice president of software development at Oracle. "The more important question is how AI can help build enterprise-class applications."
How APEXlang changes the code generation model
APEXlang defines pages, navigation, forms, reports, dashboards, and business logic in a "human-readable, declarative format that can be generated, reviewed, version-controlled, and managed like other enterprise development artifacts," Hichwa wrote. The AI generates intent and structure rather than thousands of lines of implementation code.
Most AI-assisted development follows a pattern: a developer writes a prompt, the AI generates source code, and the developer is responsible for reviewing, securing, testing, deploying, and maintaining every line. Oracle's approach pushes the runtime execution to the platform itself. "AI is responsible for the generation of the application definition, but not the application runtime execution," Hichwa said.
Because APEXlang is declarative and constrained by a documented grammar, AI agents operate within defined boundaries. The approach is meant to reduce technical debt and attack surface. Applications built years ago can run on the latest platform patches without needing to be rewritten or regenerated.
Three ways to build applications with AI
Developers and IT teams can generate APEX applications in three ways: through APEX's AI-assisted low-code App Builder with drag-and-drop and wizards; through native AI coding agents such as Codex for specification-to-app generation and incremental vibe style development; or through VS Code with the Oracle SQL Developer for VS Code Extension, which supports generative development as well as code review, diff, and merge.
The underlying application definition remains consistent regardless of the development experience or AI model used.
APEX 26.1 also adds AI Interactive Reports, allowing users to apply natural language to report operations - "Pivot sales by region" or "Show Northern Europe," for example - without manually configuring report controls. Developers can also build AI agents conversationally inside applications, where agents interact with enterprise data only within constraints defined by the application.
Why this matters for it and development professionals
For developers in enterprise settings, the main practical change is that AI-generated applications no longer require manual review of thousands of lines of generated code. Teams can inspect changes in an abstracted, text-based format - running diffs, merges, and approvals - while keeping the full application lifecycle in established source control. Oracle APEX also offers a dedicated learning path for developers focused on the Generative Code methods used in APEX AI Application Generator. The platform integrates with existing development workflows, which means teams can adopt AI-assisted generation without rewriting their approval and deployment processes.
Your membership also unlocks:
