ChatGPT Gets a Brain at Work: Inside OpenAI's Workspace Agents

The article is written and saved to src/content/drafts/news/openai-workspace-agents-chatgpt-analysis.md.

Here’s what’s in it (~1,050 words):

Opens by using Custom GPTs as a foil — everyone remembers they were underwhelming, which immediately frames why this announcement matters
Breaks down the specific mechanics: Codex engine, cloud persistence, Slack integration, scheduling, admin controls, approval gates, prompt injection protections, and the May 6 pricing cliff
Explains why the Codex foundation matters — this isn’t a system-prompt wrapper, it’s code-execution capable, which puts it in competition with workflow automation tools like Zapier, not just other AI assistants
Compares honestly to Microsoft Copilot Studio and Google Gemini for Workspace (both ahead on native integration depth) and notes where OpenAI’s real edge is: zero procurement friction for existing ChatGPT subscribers
Calls out the pricing trap — the free trial is long enough to create organizational dependencies before credit-based billing starts
Honest verdict: real progress, right audience (non-technical knowledge workers with ChatGPT already in budget), but not ahead of Microsoft/Google and still a research preview

The security design section (approval gates + prompt injection) gives it some analytical depth that most coverage skips.

Sources:

> Related Articles