OpenAI
OpenAI3d ago
Startups

How IT Admins can manage ChatGPT Work at scale | OpenAI

2 min video2 key momentsWatch original
TL;DR

OpenAI released admin APIs for ChatGPT Work that let IT admins monitor user activity, spending, and set usage caps across teams automatically.

Key Insights

1

Scoped admin keysAdmin APIs use scoped keys stored securely in Keychain, allowing IT agents to query usage data without exposing full workspace credentials.

2

Flags outliers automaticallyWork automatically flags spending outliers and recommends monthly per-user caps with projected impact before changes apply.

Want this for every new video OpenAI posts? Brevyd summarizes each upload automatically, the morning it drops.

Deep Dive

Admin APIs for ChatGPT Work

OpenAI introduced new admin APIs designed to help IT departments manage ChatGPT Work at scale. The APIs enable admins to monitor access, control spending, and measure adoption across their workspace. A demo shows how an admin creates a scoped API key, stores it securely, and then queries the system using natural language through a Work agent. The setup is straightforward: ask the agent which groups and users are most active daily, and it returns adoption data showing where teams are engaged and where support is needed.

Spending Oversight and Per-User Caps

The admin APIs surface average spend per user and identify teams spending above workspace average, with outliers automatically highlighted for investigation. When the highest-spending group is identified, the Work agent recommends a monthly per-user cap and shows the projected financial impact before implementation. The admin reviews the recommendation, confirms the change, and Work applies the cap while the admin UI verifies the adjustment. This workflow keeps admins in control — changes require explicit approval rather than automating blindly.

Takeaways

  • Start by querying your workspace activity patterns to find which teams need onboarding support or are overusing the tool.
  • Review spending outliers monthly and set per-user caps based on Work's recommendations — don't deploy caps without seeing the projected impact first.

Key moments

0:24Creating scoped admin keys

I create a scoped admin key and save it in Keychain. This allows our work agent to use the API securely.

1:05Recommending spending caps

For the highest spending group, I asked Work to recommend a monthly per-user cap. It showed the projected impact.

You just read one. Brevyd does this for every upload.

Follow OpenAI and every new video comes back as a summary like this, in your morning briefing. No watching required.