As AI takes over coding, Oracle rethinks how applications are built

As artificial intelligence increasingly takes on the task of generating code, companies like Oracle are rethinking how software applications are built and maintained. While low-code platforms have aimed to accelerate application development, the rise of AI-driven coding presents new challenges in terms of maintenance, governance, and ensuring the software remains understandable to humans.

Mike Hichwa, senior vice president of software engineering at Oracle, revealed the company's solution to this evolving landscape at Oracle AI World. Hichwa predicted that by 2026, most code would be generated by machines rather than humans. To address this shift, Oracle has introduced its own application definition language called APEXlang. This highly abstracted language focuses on the intent behind the code rather than its implementation details.

Oracle unveiled APEXlang in conjunction with the APEX AI Application Generator and has made its grammar, skills, and documentation publicly available. This move allows coding agents such as Anthropic's Claude and OpenAI's Codex to generate APEXlang, facilitating the creation of machine-generated applications. Hichwa explained that the language is designed to be maintained and governed by people within enterprises, ensuring consistency, governance, and ease of understanding across an organization's portfolio of applications.

The runtime component of APEXlang, the APEX engine, handles the interpretation of the generated code. Oracle continuously updates and patches the engine to manage the evolving AI-generated applications, alleviating the need for developers to revisit their creations. Moreover, the compact and abstracted nature of APEXlang makes the generated code easier to read and less prone to AI-generated errors.

By leveraging APEXlang, Oracle aims to streamline the software development process and provide a more manageable solution for the growing volume of machine-generated code.