CodeRabbit, a company that builds AI tools for reviewing software code, says it uses OpenAI models to govern and check code that AI systems generate, and OpenAI's Codex coding agent to speed up its own development. According to the company, the combination lets the same engineering team ship about 20 times more software. The episode frames this as an example of agentic software development, in which AI agents write, modify and test code with less step-by-step human direction. The core problem is trust. As AI-generated code makes up a growing share of what enters a codebase, teams need a way to verify it before release. CodeRabbit's approach relies on OpenAI models to review and govern that output rather than expecting developers to read every change by hand. The show notes do not describe the specific workflows, error rates or review metrics behind the 20-fold figure, so it should be read as the company's own measure of output rather than an independently verified result. For Australian businesses and professionals, this changes little unless they run a software team or are evaluating AI coding tools. The show notes do not state pricing, which would depend on model usage and any licences. The main practical risk is governance. If AI-written code or related test data is processed by third-party services, teams should check where it is stored and processed, because the Privacy Act 1988 (Cth) regulates cross-border disclosure of personal information. It is worth piloting only on non-sensitive code, with human sign-off kept on production changes.