OpenAI introduced an AI agent called "Operator"

OpenAI has announced the launch of the AI agent "Operator" for automating online tasks. It can browse web pages, enter text, click buttons, and fill out forms. "Operator" simplifies routine actions such as booking, shopping, and submitting applications, saving time and making interactions with digital services more convenient.

How does "Operator" work?

The agent is powered by a new Computer-Using Agent (CUA) model, combining GPT-4o’s capabilities with advanced logical reasoning. The agent processes information through screenshots, analyzing visual data from the screen. It uses a mouse and keyboard similarly to how a human would, mimicking real user actions. Additionally, it is trained to request confirmation before executing critical actions, such as booking a hotel or sending an email, ensuring an extra layer of control and security.

Who gets access to testing?

OpenAI has introduced a research version of the AI agent for ChatGPT Pro subscribers in the U.S. for $200. It operates on a separate platform and will improve based on user feedback. While "Operator" does not always perform perfectly and sometimes requires manual intervention, this development opens up new automation possibilities and simplifies AI interaction.

Market competition

OpenAI is not the only company working on similar technologies. In October 2024, the startup Anthropic unveiled an updated version of its AI model Claude 3.5 Sonnet, which can also interact with computers, move the cursor, click buttons, and enter text.