OpenAI's Coding Assistant Codex is Getting a 24/7 Operating Mode

OpenAI is developing a new continuously active operating mode for its coding assistant, Codex. Discovered in code within the company's GitHub repository, this feature aims to allow the AI to execute tasks 24/7 without user intervention.

Innovation Discovered in GitHub Code

New features prepared by OpenAI for its coding assistant, Codex, were detected in the command-line version within the company's GitHub repository. This discovered code indicates a significant step toward continuously operating AI systems.

Always-On Mode and Operating Principle

When this persistent mode is activated, it will continue to run uninterrupted until put into sleep mode by the user. This development aims to eliminate complaints regarding assistants shutting down on their own and leaving tasks unfinished.

Processing Power and Smart Reasoning Option

According to details in the code, the system will offer users a reasoning effort option. This way, the processing power spent by the AI model can be limited to prevent excessive token consumption overnight.

Proactive Task Management Feature

Additionally, traces of an extra feature named proactivity were found among the code. Thanks to this feature, the AI agent will be able to generate follow-up tasks on its own without requiring a new command.

Industry Competition and Sleeping Agents

The trend of continuously operating AI assistants has recently gained momentum in the industry. Microsoft previously announced its own similar assistant, and Meta is known to be working on an option called Hatch.