OpenAI is Developing AI Agents

OpenAI reportedly is working on AI agents that take over a user's device to execute multi-step tasks autonomously

Deborah Yao, Editor

February 8, 2024

2 Min Read
Image of a profile of a head with letters AI written on it
Getty Images

At a Glance

  • OpenAI is reportedly developing AI agents that can execute multi-step tasks for users autonomously.
  • The Information is reporting that a type of agent executes tasks within an environment, like transfer data to a spreadsheet.
  • Another type of agent does web-based tasks such as booking airfares, even without using APIs.

OpenAI, the maker of ChatGPT, reportedly is working on AI agents that can execute tasks for the user autonomously.

The Information is reporting that one type of agent software OpenAI is developing would effectively take over a user’s device to automate complex tasks within an environment such as work. Normally, people will have to move the cursor, click and type to transfer between applications.

For example, ChatGPT could transfer the information on a document to a spreadsheet for analysis.

Another type of AI agent OpenAI is developing handles web-based tasks such as booking airfares or creating travel itineraries without access to APIs. ChatGPT currently can do agent-like tasks but it has to use the relevant third-party’s APIs.

Last November at its developers’ conference, OpenAI launched its Assistants API that lets users build agent-like experiences in their applications.

But Adept CEO David Luan, who used to lead engineering at OpenAI, told The Information that many enterprise apps do not have APIs. Here, agents can fill in the gap, Luan said.

AI agents differ from robotic process automation (RPA), which still needs developers to manually code steps needed to complete a task, Luan said.

In contrast, AI agents can do more complex, unstructured work with little guidance from users.

Related:Google's 'Gemini Era': Embedding Gen AI in Everything It Does

Ai agents have been under development for years by various companies. In 2018, Google demonstrated that a computer can do things like call a hair salon to make an appointment for a user or contact a restaurant to book a reservation - all by itself. The staff at these businesses did not know they were talking to a computer. Google, however, did not launch it in part due to fear of public backlash, the news outlet said.

Google is changing its tune now. CEO Sundar Pichai recently said that adding generative AI to search would make it “act more like an agent over time” to “go beyond answers and follow through for users even more” by autonomously executing on their search results.

Read more about:

ChatGPT / Generative AI

About the Author(s)

Deborah Yao


Deborah Yao runs the day-to-day operations of AI Business. She is a Stanford grad who has worked at Amazon, Wharton School and Associated Press.

Keep up with the ever-evolving AI landscape
Unlock exclusive AI content by subscribing to our newsletter!!

You May Also Like