OpenAI keeps its standalone Codex app: AI coding agents are now an everyday tool

OpenAI has confirmed that its standalone Codex app — a tool where an AI agent writes, fixes and reviews code on a developer's assignment — is staying in the lineup for good. Not long ago such agents were considered a toy for enthusiasts; now it's a full product with its own roadmap rather than a feature inside a chat.

For the industry it's a point of no return: agentic development has become the norm. The programmer increasingly acts as task-setter and reviewer, while the routine part — boilerplate modules, tests, migrations — is done by the agent. The speed and cost of standard development change severalfold, and even those who never write code feel it.

For a business in Kazakhstan the consequences are practical. Custom development is getting cheaper — re-ask your contractors about their tooling and timelines: year-old estimates are stale. Internal improvements you "never got around to" — an integration, a report, a small service — deserve a fresh look: the entry threshold has dropped. And contracts with developers now warrant a clause on how AI-written code gets verified.