Not long ago, I wrote that the AI agent was the future of AI: such equipment that could work for you, such as grocery or booking meetings. This reality looks a bit closer with the latest launch of Openai.
ATSO: AWS VP says
On Thursday, during A Live streamOpenai launched a Chatgpt agent, which the company claims that it can handle complex tasks for you from beginning to end. Some examples OpenIAIs were looking at your calendar and writing a briefing based on your upcoming events, or even plans and buying and buying materials for food you were thinking about cooking. Let’s dive.
how it works
The most state -of -the -art facilities of Openai, including operator and deep research, tasted the company’s agent capabilities and now gave power to this new agent mode. The operator launched in January was designed to interact directly with a web browser to take action for you, while Deep Research is an agent feature that can find the web for you and create a detailed report in minutes that will take humans for hours.
After seeing that many questions being fed to the operator were a better fit for intensive research, Openi decided to combine both in this new experience – and add some new equipment.
Also: Microsoft is saving millions with AI and is away from thousands – where do we go from here?
For the beginning, the Chatgpt agent uses a visual browser that interacts with the web via a graphical user interface (GUI), a text-based browser, a terminal and direct API access. It also uses Chatgpt connectors, a feature that allows users to connect apps such as Gmail and Github with Chatgpt to draw relevant information to fulfill their requests.
With all those various sources of information, Chatgpt argues that it is best for work in hand and draw information accordingly. This processing is done using its own virtual computer and distinguishes between logic and action based on human instruction, which allows it to maintain reference when pulling it from several devices.
The Chatgpt agent is flexible and stereble. This allows you to interrupt a request to interrupt the mid-process and gives clear instructions to cooperate with it that is better to suit your desired result. Even though it will use new information, it will not lose the track of old instructions, allowing users to take advantage of the coupled reference. It is also necessary to work for further details and classifications from you.
What can you do with the agent of Chatgpt?
the possibilities are endless. You can automate simple tasks as scheduled an appointment for yourself in your favorite salon, or complicated as updating the spreadsheet with new financial data keeping in mind your intended nature.
Also: Openai, anthropropic, meta, and google released researchers joint AI Safety Warning – Why is here
During Live Demo, the Chatgpt agent was asked to look for a pair of black men’s dress shoes in size 9.5, start the process of making and ordering from the image of a pet, handle some aspects of the wedding plan, and even draw from Google Drive to make slides.
If all go according to the plan, the AI book is a trip for you or re -organizing your meeting schedule can be possible through Openai’s Chatgpt agent (and contestants like it). Ultimately, only time and testing will explain how to execute those tasks, but in principle, it should be as simple as you are asking in conversation what you want to do, and AI is to handle the rest.
Privacy and security
Of course, an AI that can reach your personal information and naturally enhances security and privacy concerns for you. Openai addresses these head-ons, in addition to the general model card, offers an entire page within the Vlog post dedicated to these concerns. Openai says it has added security measures for the challenges exposed in the operator Research Preview, such as handling sensitive information on live web and limited terminal network access.
Also: Does your generic AI protect your privacy? This study is best to make them best
During the live stream, Openai said that it enables the model so capable that it can browse the Internet – but that the Internet is a “very scary place.” In particular, the company was most concerned about early injections; For example, if the agent using your credit card information on a website to keep the order fell victim to a malicious scam. While the CHATGPT agent was trained to help find out the fishing efforts, the company still emphasized risks to users.
Openai says that it has also considered specific risks that agents are exposed by adding additional safety measures. The company warns that even if the agent can make a series of complex functions well, it can also make mistakes – a current range is that it cannot make slideshows.
For a complete understanding of boundaries and security risks, it is worth taking a look at the blog post and model card.
Who can use a chat agent, and how?
Unlike Openai’s most state -of -the -art features, which are usually limited to the most paid users when launching, Openai Pro, Plus, and are providing a chatgpt agent for team users. Pro users will get access to the end of the day, while the users of the plus and team will have enterprise and education users within the next few days, and within the coming weeks.
Also: 7AI iOS 26 features that I can’t wait to use (and how you can try them)
Pro users have the most bandwidth on 400 messages per month, while other paid users get 40 messages monthly with an option to expand through flexible credit-based options.
To activate the feature, users only select “Agent mode” from the dropdown of the tool during conversation with the chatbot.
Want more stories about AI? Sign up for innovationOur weekly newspapers.
