ChatGPT Unveils General Purpose Agent

OpenAI has launched its ChatGPT agent, touted as its most advanced AI agent to date. This new agent is designed for iterative and collaborative workflows, offering greater interactivity and flexibility compared to previous models. Users can now interrupt ChatGPT at any point to clarify instructions, steer it towards desired outcomes, or modify tasks, with the agent adapting accordingly.

This launch signifies OpenAI's ambition to redefine generative AI's role in professional environments. The agent can operate across web sources and user-connected apps simultaneously. For example, users can direct ChatGPT to find, analyse, and synthesise content from sources like Google Drive to assemble comprehensive reports. By expanding beyond conversational assistance into agentic productivity, OpenAI aims to challenge Microsoft's dominance in enterprise productivity. The agent capabilities will allow users to generate and edit documents, spreadsheets and presentations within ChatGPT, potentially reducing reliance on traditional office software.

ChatGPT Agent can perform tasks on behalf of the user using a virtual computer inside the browser. It can locate hard-to-find information on the web by skimming content, avoiding distractions, and extracting relevant details. The agent also demonstrates improved reasoning and output across various benchmarks.