Codex is OpenAI's AI software engineer. You don't need to code to benefit from it — you (or someone on your team) describe what you want in plain English, and Codex writes, tests, and can even run the code.
The goal is for you to (a) recognize when a task has crossed from "a GPT can do this in chat" into "this needs a real automation/script," and (b) be able to describe that need clearly enough that Codex (or a developer using Codex) can build it fast.
When a recurring task is too complex for a Custom GPT alone — like pulling live data from an API — you now know the phrase to use: "Can we build this as an automation?" rather than doing it manually every week.
Video 1: OpenAI Codex Demo — A demonstration of Codex turning plain-English requests into working code.
The facilitator uses Codex live to build one small, real automation end-to-end from a plain-English request, e.g.:
"Write a script that reads a Zoho Expense CSV export and produces a summary table of total spend by category and by employee, and flags any single expense over 10,000 THB."
Watch Codex write the script, run it against a sanitized sample export, and produce the output — emphasizing this took minutes, not a dev sprint.
This module requires more setup than is available in-session.
Get a Codex/ChatGPT Business seat with Codex access and try rebuilding your Module 4 Automation Blueprint as a real script with facilitator support in a follow-up session.