OpenAI launched ChatGPT Work on July 9, 2026 — an agent, not a chat mode, built to execute multi-step tasks across your connected apps and files rather than just answer questions about them.

The facts: it rolled out first to Pro, Enterprise, and Edu accounts on July 9, with Plus and Business following within days. It's powered by GPT-5.6. It's available to all plans on desktop today. Give it an outcome — 'analyze this month's budget variance,' 'turn these source materials into a marketing brief,' 'prep me for Thursday's sales meeting' — and it gathers information across connected tools, breaks the job into steps, and works independently for hours. Output lands as a finished document, spreadsheet, presentation, report, or website, with check-ins when it needs a decision only you can make.

Automation Squad's take: the bottleneck for most execs was never 'can AI write a paragraph.' It was assembling scattered inputs — a call transcript, a spreadsheet, three email threads — into one coherent deliverable a human could actually send. That's the specific gap ChatGPT Work targets. Whether it holds up on a genuinely messy internal project, versus OpenAI's tidy demo cases, is worth testing yourself before you trust it with anything client-facing.

Run this now: in ChatGPT, switch to the Work agent and paste something close to this: "Here's [a rough set of notes, a call transcript, or a messy email thread — paste it]. Turn this into a one-page decision brief: what's being decided, the options on the table, the tradeoffs of each, and my recommendation with reasoning. Flag anywhere you had to guess or fill a gap." Judge it on one thing: would you actually send that brief, or does it still need a rewrite?