Simplex Redesigns Software Development with Codex

Simplex is a technology partner spanning consulting, systems development, and operations, and it validates the effects of generative AI with numbers before scaling them across multiple projects. In 2023, it established an in-house CoE (center of excellence) to build the foundation for AI adoption, then rolled out ChatGPT Enterprise company-wide before adopting Codex as its primary coding agent.

Codex's scope of use goes beyond code generation. Based on design documents and reference implementations, it generates frontend and backend code, writes unit tests, and is also tasked with reviewing non-functional requirements and fixing defects found during internal integration testing. In Codex CLI, Simplex is also validating an automated workflow that runs Python scripts to go from server implementation to E2E test fixes.

The initial application target is CRUD-based web applications. As a result, screen design time was reduced by 40%, screen development time by 70%, and internal integration testing time by 17%.

Simplex's operating principles are as follows:

• Validate effects quantitatively first, then scale.
• View it as an operating model that includes governance, training, and support.
• Designate one primary agent to make knowledge accumulation and sharing easier.
• Separate experimentation from deployment and run them in parallel.
• Distinguish between areas AI will execute and areas where humans hold final responsibility.

The company views AI not as a supplementary tool for existing development stages, but as a starting point for redesigning the development process itself. It anticipates that as databases, API catalogs, and standardized design rules mature, Codex will take on more implementation and verification, and that simple systems could be generated from an RFP alone.