We have been expecting it for a while, and now it is here: Openai has introduced an agent coding tool called Codex in the research preview. The purpose of this device is to allow experienced developers to assign AI agent to Rote and relatively simple easy programming tasks that will produce a production code for production and show its work on the way.
The codex is a unique interface (not to be confused with the Codex CLI tool introduced by Open last month) that can be reached from the side bar in the Chat GPT web app. Users enter a gesture and then click “Code” to start producing the code, or “ask” to answer and give advice.
Whenever it is given a job, this work is performed in a separate container that is already loaded with the user code base and aims to accurately reflect their development environment.
Developing Codex is more efficient, developers can add a “agents, MD” file to the repo with customs guidelines, for example to context and explain the code base or to discuss standard and style methods for the project – like a readme.MD but for humans.
Made on Codex Codex -1, an excellent variation of the Open AKO 3 reasoning model that was trained using reinforcement learning on a wide range of codeing to analyze and produce coding, and to repeat through tests on the way.