OpenAI’s ChatGPT has been a global phenomenon, enough to spark an AI race amongst the leading tech companies. Everyone is out there trying to build a ChatGPT competitor. But ChatGPT appears intent on capitalizing its lead with new AI products in development that could further automate several tasks and, unfortunately, put some more people out of jobs as a side-effect.

According to a report from The Information, OpenAI is developing two types of AI agent software that would help do even more tasks traditionally done by humans.

The first type of AI agent could be used to automate complex tasks by taking over a customer’s device. For instance, the report mentions that the ChatGPT agent could transfer data from a document to a spreadsheet or fill out expense reports and enter them into accounting software. These complex requests typically involve mouse clicks, cursor movements, text input, and other tasks humans perform as they work on data across multiple apps.

This agent will require the user’s permission to work, as it will essentially take over the device. Files related to the agents working may also need to be stored on the user’s device, and OpenAI could also need permission from users to train the AI bot on personal data. The report also mentions that OpenAI has trained its computer-using agent models on examples of humans using computers, including how we humans work on different document types like charts and PDFs.

The second AI agent class is said to handle web-based tasks, such as gathering public data about a set of companies, creating itineraries, or booking flight tickets.

It isn’t clear when OpenAI plans to release these agent products, but they are said to have been in development for over a year. With advancements like these, AI will replace humans in roles that conventionally require human input and supervision due to the complexity and variability of the task.