Research2026-09-01

OpenAI's finance team used the company's Codex coding agent to automate part of its monthly-close reporting, cutting one compute-cost analysis from about five days to roughly five hours. Director of product finance Kyle Kober said the work let finance staff build their own tools rather than wait for engineering. Advisers caution that such code needs auditability guardrails and that token-based costs are rising.

What changed

Reconciling product-usage data with accounting records for the monthly close was done manually and took about five days.

What it unlocks

Finance staff building their own small automation tools with an AI coding agent instead of queuing for engineering help.

  • compute analysis: 5 days to ~5 hours
  • 90% of enterprise engineers using AI code assistants by 2028
  • under 14% in early 2024

What you need to act on it

  • access to an AI coding agent such as Codex
  • internal guardrails for auditability of AI-written code

Send this to someone who needs it

Shares the story and its sources. Nothing about you.

What does this mean for your job?

This is the story as everyone gets it. Once a week we send you the version written for your role — what changed, why it matters for the work you actually do, and one thing to try. Free while we tune it.