# Cult.fit teams build tools with Codex

> OpenAI describes an HR workflow and engineering support.

By BIGCHANGE.AI Editorial

Published: 2026-11-07T07:55:39.230Z
Updated: 2026-11-07T07:55:39.230Z
Canonical: https://bigchange.ai/blog/cultfit-codex-internal-tools-teams

![Conceptual illustration of colleagues developing an internal tool](https://bigchange.ai/api/media/file/cultfit-tool-concept.png)
AI-generated conceptual illustration; not an actual Cult.fit workplace or interface.

OpenAI said November 5 that Cult.fit's non-engineering teams built more than 20 internal applications with Codex in recent months.

## The big change

- Employees outside engineering use the assistant to create tools for their own operations.
- A reported HR workflow organizes performance-review scheduling and feedback.
- Engineering support and review matter when colleagues begin relying on an internal application.

## From an internal need to a tool

[OpenAI's account](https://openai.com/index/cultfit-builds-with-codex/) says Cult.fit's HR team used Codex to build a workflow that schedules reviews and collects and assembles feedback. The case does not publish the application or independently evaluate its reliability. It describes a tool for a business process, not a demonstrated replacement for the people managing it.

Cult.fit says it made company data easier for Codex to use, while engineering offers twice-weekly office hours and embeds engineers in selected functions. A request for an application is only the beginning: a team still needs to decide what information it should handle, whether its output is correct and how to maintain it.

## What the reported saving shows

OpenAI says an engineer rebuilt a dashboard with Codex over a weekend, avoiding over $100,000 in annual software costs. The figure is not independently verified.

Performance feedback can contain sensitive information. Before an internal tool is widely used, its owners need to decide who can see that information, how errors can be corrected and who maintains the application when the underlying process changes. These are questions for adopters, not safeguards demonstrated by this case.

## Keeping a review boundary

OpenAI's [separate code-review guidance](https://developers.openai.com/blog/custom-code-review-rules-for-codex) says automated review is an additional reviewer and that tests, branch protections and required approvals remain enforcement mechanisms. It does not verify Cult.fit's implementation. A company adopting generated internal tools still needs to determine who checks their behavior, data access and maintenance before colleagues rely on them.

## Sources

- [OpenAI: Cult.fit customer case](https://openai.com/index/cultfit-builds-with-codex/) — November 5 vendor and customer account; outcomes not independently audited.
- [OpenAI: Codex review rules](https://developers.openai.com/blog/custom-code-review-rules-for-codex) — Separate general guidance; not validation of Cult.fit.