Cult.fit, one of India's largest technology-enabled fitness companies, operates 708 fitness centres across 77 cities and serves nearly one million paid members. According to OpenAI's customer story, the company standardised its engineering teams on Codex and then extended the tool across the business. Its CTO, Santhosh Kumar, said the aim was to become more AI-native in how the company runs operations, without adding layers of process. The engineering shift came first. Developers had been using a mix of coding agents, which produced different conventions for repositories, agent instructions and workflows. Codex now serves as the common platform for testing, code review and feedback loops. A Site Reliability Engineering agent aggregates data from observability tools and logs, investigates incidents and often opens a pull request with a proposed fix for an engineer to review. Cult.fit is also moving toward spec-driven development, and Kumar estimated that a new social feed for its app would previously have taken more than a year but was built and shipped in two months. Beyond engineering, the company has made its data easier for Codex to query, so staff can run routine analysis without waiting on a central analyst team. Non-engineering teams have built more than 20 internal applications in recent months. HR, for example, built a workflow for performance reviews and scheduling. In one case, an engineer rebuilt a dashboard tool over a weekend after its licence cost rose sharply, avoiding more than $100,000 a year in software costs. The company supported the rollout with a hackathon, a one-day session for managers and twice-weekly office hours. Cult.fit is also using the OpenAI API for its AI Trainer, a guided workout product that about 20 per cent of its gym users currently use. The latest version is voice-first and built on GPT Realtime, aimed at the 45 per cent of members who are first-time gym-goers. The figures come from OpenAI's published case study and have not been independently verified. For Australian businesses and professionals, the direct impact is limited, since this is an Indian company's deployment and it changes no obligations under the Privacy Act 1988. The transferable lesson is practical: if your team is considering AI coding agents or internal app builders, start with a single approved tool, review AI-generated changes before they reach production, and avoid connecting customer personal information to any tool until you have assessed it against the Australian Privacy Principles. Subscription costs were not disclosed, so budget for a trial on non-sensitive internal work before committing.