Over the past several weeks, Michael Hichwa, SVP of Software Development at Oracle and one of the original creators of Oracle APEX, has shared an unusually detailed look at what is coming in Oracle APEX 26.1. Across a series of LinkedIn posts, Hichwa has outlined Oracle’s broader vision for generative application development, centered around a new open application specification language called APEXlang.
Many AI conversations in enterprise software feel focused on chatbots, copilots, and code assistants. APEXlang points to something more foundational. Oracle appears to be rethinking how APEX applications are represented, generated, managed, and evolved so they can participate more directly in AI-assisted development workflows.
What Is APEXlang?
APEXlang is a new optional application export and import format coming in Oracle APEX 26.1. Instead of exporting an application as a single monolithic SQL file, APEX applications can be exported as a structured package of readable .apx files and supporting assets.
That may sound like a file-format change, but Oracle’s messaging suggests something more strategic. APEXlang creates a structured representation of an APEX application that both human developers and AI systems can understand, generate, validate, and evolve over time.
In practical terms, APEXlang is designed to support human readability, AI-assisted generation, source control, validation, lifecycle management, and portability. For enterprise teams, the most important point is that APEXlang gives Oracle APEX a more modern foundation for AI-assisted development and collaborative delivery.
Why APEX Needed to Go Beyond SQL Exports
Traditional Oracle APEX exports were designed primarily for deployment and migration between environments. They served that purpose well, but they were not designed to operate as editable application source files.
SQL exports are difficult to diff cleanly because they contain environment-specific IDs and large blocks of generated metadata. They were also never intended to support direct editing or traditional Git-style merge workflows. In practice, that makes asynchronous collaboration difficult because developers cannot reliably merge changes the way they would with conventional application source files.
This matters even more in the context of generative AI. AI systems work best with structured, readable formats that clearly express relationships and intent. APEXlang appears to give Oracle APEX that kind of representation:
- Concise enough for AI generation.
- Readable enough for developers.
- Structured enough to support validation and lifecycle management.
One of Hichwa’s recurring themes is that APEXlang is intended to capture intent rather than raw implementation code. That idea may become one of the defining characteristics of Oracle APEX 26.1.
From Point-and-Click Development to Generative Development
Oracle APEX has always been declarative. Developers use a visual interface and configuration-driven components, along with handwritten code when needed, to define how an application should behave while the platform handles much of the underlying implementation.
APEXlang appears to extend that philosophy into the generative AI era.
At last week’s Oracle APEX Executive Roundtable and Bootcamp hosted by Traust in Minneapolis, attendees received an early preview of how this may work in practice. Oracle APEX product team members Jayson Hanes and Marc Sewtz demonstrated a workflow where a well-structured set of application requirements was provided to an AI model, which then generated APEXlang output capable of producing a functioning application.
The demo was impressive, but it also reinforced an important reality: generative AI development still depends heavily on architecture, requirements, and structure. The better the requirements, the better the output.
That is where much of the conversation around AI-driven development becomes oversimplified. Generative development does not eliminate the need for experienced developers and architects. It changes where those skills are applied. Developers may spend less time manually assembling pages, regions, and components inside the page builder and more time defining application intent, structuring requirements, validating generated output, and managing the broader application lifecycle.
Point-and-click development inside Oracle APEX is not disappearing. But with APEXlang, enterprise application development may become less dependent on manually constructing every screen and more focused on defining systems clearly enough that AI can help generate and maintain them.
Where Generative Development Meets Enterprise Architecture
For enterprise IT leadership, the significance of APEXlang extends beyond AI novelty. Many application platforms face questions about long-term maintainability, governance, portability, and modernization. Oracle appears to be addressing those concerns directly.
APEXlang positions Oracle APEX for:
- file-based development workflows,
- Git-based collaboration,
- AI-assisted generation,
- structured lifecycle management, and
- modern tooling such as VS Code.
Those capabilities matter because technical leaders are evaluating whether their existing platforms are ready for AI-assisted software delivery.
Oracle’s recent messaging suggests that APEX is not being left behind in that transition. In fact, APEX may be well suited for generative development because of its declarative architecture, structured metadata, and tight integration with the Oracle Database. Generative AI works best when it operates against clear patterns and well-defined relationships. Oracle APEX already emphasizes those qualities.
APEX Developers Are Becoming Architects
One of the more interesting themes emerging from Hichwa’s posts is the idea that developers will increasingly act as orchestrators of AI-assisted development workflows. He frequently references:
- “virtual staff,”
- reusable “skills,”
- specification-driven development, and
- lifecycle management for generated systems.
That does not mean technical expertise becomes less important. If implementation becomes more automated, the value of architecture, requirements engineering, UX design, governance, and business process understanding increases.
Organizations still need people who understand how enterprise systems actually work. AI can accelerate implementation, but it cannot independently resolve conflicting business requirements, define organizational workflows, establish governance models, or design thoughtful user experiences.
Generative tools can help produce applications faster. But they do not eliminate the need for disciplined application design, especially in enterprise environments where maintainability, security, scalability, and long-term support matter more than generating a prototype quickly.
What Organizations Should Be Doing Now
Even before Oracle APEX 26.1 is officially available (any day now 🤞), Oracle’s direction offers several clues about where enterprise development workflows are heading.
First, organizations should improve requirements discipline. Generative AI systems depend heavily on clear instructions and structured inputs. Teams with vague requirements, inconsistent business rules, or poorly defined workflows are unlikely to get reliable results from AI-assisted development.
Second, teams should modernize their development workflows. Oracle’s emphasis on APEXlang, Git, VS Code, and file-based development signals a continued shift toward more collaborative delivery models. Teams still relying on isolated development practices or manual deployment processes may find themselves increasingly constrained.
Third, organizations should continue investing in human expertise. AI-assisted development changes the role of developers, but it does not eliminate the need for people who can evaluate tradeoffs, define architecture, understand users, validate outputs, and align technology decisions with business goals.
At Traust, we are already utilizing generative AI to improve Oracle APEX development workflows. In our recent article, 7 Tips for Using AI for Oracle APEX Development, we discussed practical ways our team is integrating generative AI into real-world development processes while still maintaining strong architectural and UX discipline.
What We Still Don’t Know
While Oracle’s vision for APEXlang is becoming clearer, many implementation details are still emerging. Questions remain around:
- How mature generation workflows will be initially.
- How deterministic generated applications will become.
- What governance tooling Oracle will provide.
- How much traditional development work will still be required for highly customized applications.
Even with those questions, Oracle’s direction appears increasingly clear. APEXlang suggests that Oracle sees generative AI not as a side feature, but as a fundamental shift in how enterprise applications will be designed, generated, and maintained.
For organizations already invested in Oracle APEX, APEX 26.1 may represent one of the platform’s most important architectural evolutions in years. The future of AI-driven development will still depend on clear requirements, strong architecture, disciplined governance, and thoughtful user experience. APEXlang gives Oracle APEX a foundation for that future.
