American artificial intelligence research organisation,OpenAI, has launched an autonomous AI coding assistant designed to go beyond just auto-complete or code generation.
The new AI coding agent, Codex, can read and write files, use tools (like web browsers or terminals), plan and manage full coding jobs (like creating apps or correcting bugs), and execute sophisticated software development workflows using multi-step reasoning.
Although ChatGPT is quite good at developing and explaining code, especially when using GPT-4.5 or GPT-4-turbo, this new AI agent is more independent and can deconstruct human instructions into smaller goals, carry out those steps on its own, and debug and improve its own work.
In a statement, OpenAI said: ‘Major companies have already evaluated and adopted Codex. Cisco is testing the tool to speed up engineering workflows, while Superhuman uses Codex to improve test coverage and enable non-engineers to contribute code changes.”
‘‘Kodiak, an autonomous vehicle company, leverages Codex to enhance code reliability and gain insights into complex software stacks. Temporal uses it for background tasks like debugging and test writing.”
