The race between Anthropic and OpenAI is accelerating around a new generation of AI coding tools that do more than write code when asked. They are increasingly designed to carry out recurring and continuous tasks with a high degree of autonomy.
The shift began with Anthropic’s launch of “Routines” in Claude Code, followed quickly by OpenAI’s major update to Codex, expanding its role from a coding assistant into a more persistent and independent agent.
Details
- Anthropic launched “Routines” in Claude Code.
- The feature allows developers to schedule automated coding tasks to run hourly, nightly, or weekly.
- These tasks can continue without the need to keep a laptop open.
- OpenAI responded with a new update to Codex.
- The update added background computer control.
- It also added persistent memory and more than 90 new plugins.
- The moves by both companies show that competition is no longer just about generating code, but about building a coding agent that can work continuously.
The core idea is to turn AI from a tool that is called when needed into a system that works inside the team on an ongoing basis.
What’s Next?
The next phase will not be only about improving code quality, but about which company can first turn the coding assistant into a reliable agent that can be left to work on its own for extended periods.
