OpenAI is preparing to launch an autonomous artificial intelligence (AI) agent, code-named “Operator,” as soon as January.
The software, which would take direct actions on a person’s computer, would initially premiere as a research preview through OpenAI’s developer API, according to Bloomberg, which first reported the news Thursday. Operator could be a general-purpose tool that executes tasks in a web browser, the news organization reported.
OpenAI’s bid came shortly after rival Anthropic launched its “Computer Use” feature; Alphabet Inc.’s Google is reportedly readying its own consumer-focused version for release in December.
The ChatGPT maker’s foray into agentic AI has been expected as the industry races to embrace an enterprise-wide use of generative AI that could quickly blossom into a multibillion-dollar business. For weeks, OpenAI executives hinted strongly at an imminent release. “I think 2025 is going to be the year that agentic systems finally hit the mainstream,” company chief product officer Kevin Weil said OpenAI’s Dev Day last month.
The use of AI agents is “the next giant breakthrough” in the field, OpenAI CEO Sam Altman said during a recent Reddit “Ask Me Anything” session.
But analysts caution OpenAI may have moved too late. Some of the industry’s biggest names — Microsoft Corp., Box Inc., Salesforce Inc., ServiceNow Inc., Cisco Systems Inc., LinkedIn, Adobe Inc., and others — have plunged into the market in recent weeks.
“OpenAI’s Operator strategy is mirroring what many other AI companies and enterprise software vendors are already doing, which is leveraging generative AI and automation to enable agents to make decisions and take actions based on vetted company data, and learn from past interactions,” Keith Kirkpatrick, research director at The Futurum Group, said in an email. “OpenAI is a little late to the party, and will need to demonstrate how its agents will be able to provide value quickly, with minimal integration issues.”
“Ultimately, I see a mixed environment at enterprise organizations, where agents from several vendors — which could include Salesforce, ServiceNow, Google, Anthropic, and others — will need to interact with one another, based on where they’re deployed, the data upon which they act, and their level of complexity,” Kirkpatrick added.
News of Operator landed the same day that OpenAI released a policy paper on the nation’s AI strategy that reportedly proposes the federal government create AI “economic zones” to build AI infrastructure, and the establishment of a bloc with U.S. allies to compete with China.
The proliferation of autonomous agents, and their ability to perform repetitive work tasks within enterprises, is widely considered the next phase in the growth of GenAI. Executives have stamped AI agent use as a validation of their enormous investments in AI.
The latest entrant in the agentic AI sweepstakes, ThoughtSpot, launched Spotter, an autonomous analytics agent designed to give anyone from front-line workers to executives enterprise-level insights.
