OpenAI is reportedly gearing up to launch ‘Operator’, a new AI agent tool through which it looks to capitalise on an ever-growing market for AI agents across sectors. This comes in via a Bloomberg report.
The development though is hardly unexpected, given that OpenAI CEO Sam Altman had hinted at this earlier. In an “Ask Me Anything” forum on Reddit, Altman noted that the company will “have better and better models,” but “I think the thing that will feel like the next giant breakthrough will be agents.” Later, at an OpenAI press event, the company’s chief product officer Kevin Weil said, “I think 2025 is going to be the year that agentic systems finally hit the mainstream.”
The company aims to launch Operator as a research preview in January, making the tool accessible to developers through its API. According to reports, Operator can perform tasks on behalf of the user, performing multi-step, independent actions. This marks a shift to agentic AI (wherein a series of tasks can be performed autonomously, such as coding. The concept of agentic AI has spread in recent months, and the likes of Google and Anthropic have been working on integrating it into their AI agents. Google’s offering in the area is said to be Project Jarvis (yes, it seems to be named after the AI in the Iron Man movies).
Anthropic, in particular, announced that it will roll out Computer Use, an AI agent that comes with the capability to process a user’s real-time computer environment, executing actions as directed, while Microsoft has developed AI tools that assist with tasks like email management and record keeping for professionals.
Coming back to OpenAI, the firm aims to leverage advanced natural language processing and reinforcement learning for its Operator. More specifically, models like GPT-4 will provide the necessary linguistic understanding, while reinforcement learning will enable the agent to improve its responses based on previous interactions.
This will allow Operator to grow beyond a basic digital assistant and allow it to effectively “learn” and adapt over time, accumulating knowledge through interactions and handling more personalized and complex requests. If Operator rolls out, it will be able to enhance productivity by reducing time spent on repetitive or administrative tasks. For individual users, Operator could simplify tasks such as booking appointments or managing digital records. For enterprises, Operator can perform tasks like automating customer service responses or handling data entry tasks.
