OpenAI just dropped a research preview of Operator, an AI agent that can navigate a web browser and complete tasks on its own.

It's rolling out first to U.S. users on ChatGPT’s $200 Pro plan, with plans to expand to Plus, Team, and Enterprise users later.

If Operator lives up to the hype, we’re one step closer to an AI assistant that actually does the boring stuff—so maybe one day, booking a flight won’t feel like solving a CAPTCHA from hell.

CEO Sam Altman confirmed Europe will have to wait.

For now, you can test Operator at operator.chatgpt.com, but OpenAI wants it baked into all ChatGPT clients down the line.

The idea is to automate everyday tasks—booking hotels, making restaurant reservations, online shopping—without you lifting a finger.

It’s not fully hands-off, though. For sensitive actions like payments, you’ll need to step in.

Operator runs in a separate browser window, showing you every step it takes.

It’s powered by OpenAI’s Computer-Using Agent (CUA), a model that blends vision and reasoning to interact with websites—no special APIs required.

It also asks for confirmation before completing major actions, so you can double-check before anything becomes final.

That said, Operator isn’t perfect yet. OpenAI admits it struggles with complex tasks like building slideshows or managing intricate calendars.

But if it works as promised, it could be a serious upgrade for digital multitasking.