# Over $100,000 a year saved! How Cult.fit rolled out Codex company-wide

Hi everyone, it's Shii-chan! Today's customer story from OpenAI is about Cult.fit.

## What was announced?

OpenAI News published a story about **Cult.fit**, one of India's leading fitness companies, with 708 centers across 77 cities and nearly one million paid members. It standardized engineering on Codex and then extended it across the company.

## The story so far

- Developers used a mix of coding agents and assistants, so repository conventions, agent instructions, and skills differed
- Even small data requests had to go through a central analyst team, which created a bottleneck

## What changes

- Repository structure, skills, testing, code review, and feedback loops follow common conventions
- Teams can analyze data themselves
- Non-engineering teams built more than 20 internal applications in recent months

## Dive Deep

### Engineering

- An SRE agent gathers information from observability platforms and logs, investigates incidents, and often creates a pull request with a proposed fix for an engineer to review
- Security agents help identify and address vulnerabilities
- The team is moving to spec-driven development, from specification to pull request
- A social feed for the app, which the CTO estimates would have taken more than a year, shipped in two months

### Company-wide

- HR built a Codex workflow for performance reviews, scheduling, and feedback collection
- When an internal dashboard tool's cost rose sharply, one engineer rebuilt it with Codex over a weekend, avoiding more than $100,000 in annual software costs
- The rollout used a large hackathon, a one-day session for leaders, twice-weekly office hours, and embedded engineers

### Member experience

- About 20% of gym users use the AI Trainer; the latest version is voice-first, using GPT Realtime for two-way coaching and workout logging
- Over 45% of members are first-time gym goers

## Wrap-up

- Standardized on Codex for a common development process
- Non-engineers built 20+ apps
- In-housing an internal tool avoided $100,000+ a year
- The AI Trainer gets a voice-first upgrade via the OpenAI API

This is for organizations that want AI beyond engineering!