SDAIA’s guide to generative programming puts human review at the heart of the developer’s work

When an intelligent assistant proposes lines of code in seconds, the professional question becomes less about typing speed and more about who reviews what was written before it reaches a real user. This is the focus that the Saudi Data and AI Authority (SDAIA) placed in its new generative programming guide, released on 29 October. The guide does not treat the tool as a substitute for the developer, but as part of a development environment that requires clear direction, inspection and responsibility.

SDAIA describes the guide as a national reference that helps developers and specialists use emerging technologies efficiently, while preserving reliability, quality and cybersecurity. This comes at a time when the use of generative and agentic AI tools is expanding in designing digital solutions, writing code and testing it. Here, “agentic tools” refers to tools that can execute a sequence of steps within a specific task, not merely suggest a single answer in a chat window.

The developer does not hand over the decision to the tool.The guide positions the human to direct the tool, supervise its work, review its outputs and verify their quality and reliability before adopting or deploying them. This is not a decorative addition to the workflow. Code that appears acceptable in a small example may contain a functional error, a security risk, or a unsuitable assumption for the product environment. Therefore, understanding the problem, context and system boundaries remains part of the developer’s role, even when the tool handles part of the writing or analysis.

The guide tracks AI use across work stages, from idea generation and prototype building, to code writing, analysis and optimisation, then test case preparation, automation and result analysis. This map is professionally important because it shifts the discussion from a simple question, does the developer use AI or not?, to a finer one: at which stage is it used, and what needs independent verification before moving to the next stage?

Prompt formulation is a work skill, not a magic spell.The guide notes that an effective prompt specifies role, context, instructions, input data, constraints and output format. For a novice developer or a small team, this means the tool’s result improves when presented with testable requirements, not a generic phrase like “write an app”. For the CTO, it is a reminder that output quality also depends on clear product specifications, data and controls, not solely on the model’s capability.

The guide’s publication alone does not prove that a particular job has disappeared or that the Saudi job market is built around a single tool. It does, however, provide a practical signal of the type of readiness being regulated: the ability to work with code-generation tools, translate a vague task into specific instructions, and review results in light of quality and security. These are skills distinct from merely running a tool or copying its output.

What is worth the developer doing right now?Pick a small task from your work, such as writing a test or reviewing an existing function, and write a clear context, constraints and success criteria for it. Use the tool to generate a draft, then review the result as if a colleague had sent it to you: check the logic, edge cases, data handling, tests, and whether the code fits the project environment. This practice does not promise a functional outcome, but it builds a professional habit that the guide itself insists on: leveraging speed without abandoning human judgment.

In a market where development tools evolve rapidly, the most stable value for a developer may be the ability to explain why they accept or reject an output. The tool may suggest, but it does not bear product responsibility nor does it alone know what matters to the user or the organization. This dividing line makes human review both a technical and professional skill.

For developers in Saudi Arabia, the Gulf and Egypt, the practical message is clear: a national reference of this weight raises employer expectations in the region from “used the tool” to “reviewed the output and verified it”. So note in your résumé and interview how you audit generated code, and do not stop at mentioning the tool you used.