Santhosh Kumar, CTO at Cult.fit, told OpenAI News the company standardized its software development lifecycle on Codex to end a sprawl of coding assistants.
Cult.fit, which operates 708 fitness centers across 77 cities for nearly one million paid members, said the move let it unify repository structures, testing and review conventions and then push the same platform beyond engineering, per OpenAI News. The company reports more than 20 internal applications built by non-engineering teams in recent months, more than $100,000 in annual savings after rebuilding an internal dashboard over a weekend, and a 50 percent plus reduction in customer support costs after embedding AI agents in its Sprinklr workflows.
That standardization matters because Cult.fit's earlier setup made reuse hard.
Kumar said developers had used a mix of agents and assistants with different conventions for instructions, skills and workflows. Once Cult.fit picked Codex as the foundation, Kumar said, the team could build a repeatable path from spec to pull request, with specialized agents for site reliability and security that aggregate logs, investigate incidents and draft fixes for an engineer to review. He pointed to a social feed for the Cult.fit app that the team shipped in two months, a project he estimated would previously have taken more than a year. The draft still lands as a pull request, it does not merge on its own.
The second shift was away from engineering. Kumar said Cult.fit made company data easier for Codex to work with so teams could self-serve analysis that previously queued through a central analyst team. He called that the biggest step toward data democratization. HR, he said, built its own workflow for performance reviews, scheduling and feedback aggregation in Codex without engineering owning the build.
HR's example is not an outlier. Cult.fit said non-engineering teams built more than 20 internal apps in recent months, and that the weekend dashboard rebuild came after a vendor price increase, per OpenAI News. The claim shows savings but it also shows the trade off, the same democratization required a company-wide hackathon, a one-day build session for managers, twice-weekly office hours and engineers embedded in functions to shepherd ideas to production.
On the member side, Kumar tied the same tooling to a voice-first AI Trainer. About 20 percent of Cult.fit gym users use the trainer, the company said, and more than 45 percent of members are first time gym goers who need guidance. Cult.fit said the latest version uses GPT Realtime through the OpenAI API for two-way voice coaching and workout logging, building on an earlier text-based trainer the company had run for several years.
Context puts the Codex choice in sharper relief. Codex also ships as a lightweight coding agent that runs in your terminal, per its GitHub releases, while Encore notes Codex Cloud is proprietary and built around OpenAI models. Encore's 2026 ranking contrasts that with Cursor, which pairs an AI code editor with an agent mode and Cloud Agents that run tasks remotely and gate commands through a terminal allowlist. Cult.fit chose the centralized platform route over the editor-centric route, betting that shared conventions and centrally governed skills would scale faster than letting each team pick its own assistant.
What Kumar did not detail is how Cult.fit governs those 20 plus team-built apps once they touch production data, how it measures the accuracy of support decisions like weather-based penalty waivers, or what Codex costs at enterprise scale versus the $100,000 plus vendor saving it cites. The support agents now check signals such as local weather and attendance patterns before waiving a missed-class penalty, per OpenAI News, but the company did not say what the reversal or escalation rate is or whether members can appeal an AI decision.
The question left hanging is who owns reliability when anyone can ship. If software is democratized, who signs off when a non-engineering app breaks a workflow that frontline staff depend on?
