New Delhi: OpenAI has introduced a new AI tool, called ChatGPT Agent, which promises to complete multi-step tasks on behalf of the user with the use of a virtual computer. This is one of the significant tools in the development of AI, as it does not have the capabilities of ordinary chatbots but can accomplish much more challenging tasks such as event planning, presentation creation, or maintaining daily schedules. The feature is now out to ChatGPT Pro, Plus, and Team users, and it can be used through agent mode or by entering /agent in the tool.
ChatGPT Agent is an agent that is trained with a specialised model based on reinforcement learning to control tools, including a text browser, visual browser, and terminal. It uses the combined technologies of the previous tools of OpenAI, Operator and Deep Research. To aid in the creation of Agent, the company has combined the teams that produce these products, and Agent now has 20-35 researchers and product leads.
Performs real-world tasks like a virtual assistant
According to The Verge, the AI could verify calendars, cross-reference restaurant reservations, automate weekly office work and produce full research reports. The system is simple to use and flexible since the user can stop or edit tasks in progress. Before performing such a task as sending emails or making reservations, the AI will never act without the consent of the user.
The increased availability of the tool is one of its main advantages, as it does not only take control of a browser; it can take over the whole virtual computer. This implies that it is able to handle deeper workflows. But this is at a compromise of speed. OpenAI estimates that the tool can complete a task in 15-30 minutes; however, it can save time in comparison with manual work, which is enough to make it worthwhile.
Built-in safety and limited financial access
ChatGPT Agent is subject to high safety standards by OpenAI. These are set constraints in biological or chemical enquiries as well as monetary transactions, and a watch mode to detect any sensitive activities. The access to financial sites is already limited, and the AI is not able to make payments or trades.
Agent remains in its early rollout, and it will be offered to ChatGPT Enterprise and Education users later this summer. An implementation schedule of the European Economic Area and Switzerland has not been published yet.
Part of a larger race toward AI agents
The agent developed by OpenAI is one of the latest trends in an increasingly rapid industry built around AI tools capable of replicating the decisions and actions of people. Silicon Valley giants, such as Google, Meta, and Amazon, are in a rush to develop their own AI assistants, similar to the Klarna AI agent that removed an equivalent of 700 workers in only a month.
The goal of these tools is to become daily digital assistants – like J.A.R.V.I.S. of Iron Man. The release of ChatGPT Agent is an indicator that AI is quickly moving toward the category of useful to essential, though they are not quite there yet.