OpenAI unveils ChatGPT Work, an agent for hours-long projects
OpenAI has unveiled ChatGPT Work, a new agent that can take on long-running projects. It works with your apps and files, can stay on a project for hours, and turns a stated goal into finished work. The launch moves AI assistants toward handling more complex, longer tasks.
AI-processed from OpenAI Blog; edited by Hamidun News
OpenAI has unveiled ChatGPT Work, an agent that can take actions in apps and files, stay on a project for hours, and carry a goal through to finished work. It is a new class of assistant: one that goes beyond text chat and takes an active part in getting tasks done.
What sets ChatGPT Work apart
The traditional ChatGPT works as an adviser: you describe a task, get a recommendation, and then act on it yourself. ChatGPT Work takes a different approach. It is an active agent that can:
- Act inside your apps and files
- Work on a project for hours on end without breaks
- Turn a high-level goal into a finished result
The agent doesn't hand you instructions; it carries them out. It clicks through interfaces, fills in data, copies information between apps, and keeps track of context over long hours of work.
The evolution of AI agents at OpenAI
Over the past few years, OpenAI has been experimenting with a new type of AI agent. Its first major project was Operator, a browser agent that carries out actions on the web on its own, such as buying tickets, ordering food, and filling out forms. ChatGPT Work follows the same philosophy but focuses on more complex, long-term projects that require coordination across multiple apps and files.
This marks a strategic shift from "AI as a text generator" to "AI as a doer of actions." Systems like this could significantly cut the time spent on routine multistep work.
Practical uses
ChatGPT Work may be useful for projects that demand hours of careful work across many tools: analyzing large datasets and moving the results into a report, syncing information between a CRM and spreadsheets, processing documents, and compiling reports from disparate sources.
The key difference is its ability to stay within a project's context for hours. That lets the agent handle long-term tasks that don't fit into a single chat session.
What this means for users
ChatGPT Work marks a new stage in the development of AI assistants. Instead of getting advice and applying it themselves, users will be able to delegate hours of work on complex projects to AI. It brings OpenAI closer to its vision of systems that can act as true digital partners on ambitious tasks.
Key Questions Answered
What sets ChatGPT Work apart?
The traditional ChatGPT works as an adviser: you describe a task, get a recommendation, and then act on it yourself. ChatGPT Work takes a different approach. It is an active agent that can: Act inside your apps and files Work on a project for hours on end without breaks Turn a high-level goal into a finished result The agent doesn't hand you instructions; it carries them out. It clicks through interfaces, fills in data, copies information between apps, and keeps track of context over long hours of work.
What are the key details on The evolution of AI agents at OpenAI?
Over the past few years, OpenAI has been experimenting with a new type of AI agent. Its first major project was Operator, a browser agent that carries out actions on the web on its own, such as buying tickets, ordering food, and filling out forms. ChatGPT Work follows the same philosophy but focuses on more complex, long-term projects that require coordination across multiple apps and files. This marks a strategic shift from "AI as a text generator" to "AI as a doer of actions." Systems like this could significantly cut the time spent on routine multistep work.
What are the key details on Practical uses?
ChatGPT Work may be useful for projects that demand hours of careful work across many tools: analyzing large datasets and moving the results into a report, syncing information between a CRM and spreadsheets, processing documents, and compiling reports from disparate sources. The key difference is its ability to stay within a project's context for hours. That lets the agent handle long-term tasks that don't fit into a single chat session.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.