With ChatGPT Work, OpenAI is pursuing one of the most significant strategic shifts in the history of ChatGPT. Rather than simply answering questions or generating text, the new capability is designed to take ownership of entire workflows. ChatGPT Work is conceived as an agentic work assistant that can plan projects independently, combine information from multiple sources, interact with applications and ultimately deliver finished outcomes.
In doing so, ChatGPT is evolving from a conversational AI into a platform capable of working autonomously over extended periods. OpenAI no longer describes ChatGPT Work as a tool for individual prompts, but as a partner for complex knowledge work that accepts a goal and independently produces meaningful results.
From Conversation to Autonomous Work
The fundamental difference between ChatGPT Work and previous versions of ChatGPT lies in how it approaches work. Traditional chat interactions require continuous user instructions, whereas ChatGPT Work pursues a broader objective through multiple interconnected steps.
Once a task has been assigned, the agent develops a plan, gathers the necessary context from connected sources, breaks complex projects into manageable stages and works through them methodically. Users can monitor its progress, answer occasional follow-up questions or adjust its direction whenever necessary.
For larger projects, ChatGPT Work can remain active for hours, continuing to make progress without requiring every individual step to be triggered manually.
Working Across Apps, Files and the Web
One of the most significant innovations is that ChatGPT Work is no longer limited to information contained within a conversation. The agent can access connected applications, documents, files and online resources, combining them into a single workflow.
This allows it to consolidate information from multiple documents, analyse spreadsheet data, prepare presentations or generate reports, websites and other business materials.
OpenAI describes this as a shift from isolated answers to complete workflows. Rather than moving between several applications manually, users simply describe the desired outcome while ChatGPT Work organises and executes the individual steps.
Codex Becomes the Engine Behind the Scenes
Technically, ChatGPT Work is built on GPT-5.6 and incorporates the latest generation of Codex directly into its workflow.
Codex was originally introduced as a programming agent for software developers. It has since evolved considerably. Today, it no longer focuses solely on writing code but also manages structured workflows, automation and complex execution tasks.
As a result, ChatGPT Work can do far more than analyse or write. It can execute tasks, organise information and connect multiple tools into coherent processes. OpenAI increasingly positions Codex as a universal execution engine for knowledge work rather than simply a coding assistant.
Designed for Complex Projects
ChatGPT Work is intended for projects that evolve over time and require multiple sources, intermediate steps and sustained reasoning.
Typical use cases include extensive research, market analysis, project reports, presentations, technical documentation and even the creation of complete web applications. The agent gathers information, organises it, analyses it and produces polished deliverables.
Its key strength lies in maintaining context across multiple sources while preserving the overall structure of a project throughout a longer workflow.
For this reason, OpenAI increasingly describes the experience not as a conversation but as project-based work.
Sites Extends the Workspace
Alongside ChatGPT Work, OpenAI has also introduced a new capability called Sites.
This allows users to generate interactive websites, dashboards or lightweight web applications directly from their projects. Rather than exporting work as a PDF or presentation, results can be published immediately as fully functional websites.
For reports, internal dashboards or interactive project pages, this significantly expands what users can create without leaving ChatGPT.
Scheduled Tasks Keep Working Automatically
Another important addition is Scheduled Tasks.
These enable ChatGPT Work to perform recurring activities or respond automatically to predefined events rather than executing tasks only once. Regular research, recurring reports and ongoing monitoring activities can therefore be automated with minimal intervention.
This moves ChatGPT Work beyond being a reactive assistant towards becoming a continuously operating work agent.
A Unified Desktop Experience
ChatGPT Work also marks a major change in OpenAI’s product strategy.
The previously standalone Codex experience is being integrated into the new ChatGPT desktop application. Users will be able to move seamlessly between three primary workspaces: traditional chat, ChatGPT Work and Codex.
Depending on the task, they can switch between conversation, agentic project execution and software development without leaving the application.
OpenAI’s long-term ambition is to transform ChatGPT into a unified productivity platform rather than maintaining multiple disconnected products.
Built with Businesses in Mind
Although ChatGPT Work is suitable for a broad range of knowledge workers, the current rollout is focused primarily on organisations, educational institutions and professional users.
Enterprise and Education customers received access first, followed by a gradual rollout to additional paid ChatGPT plans.
For businesses, the key advantage is that ChatGPT Work can operate across existing files, connected applications and organisational workflows without requiring employees to supervise every individual action.
OpenAI therefore positions the product as a digital colleague for everyday office work rather than simply an intelligent search engine or writing assistant.
Powered by GPT-5.6
ChatGPT Work runs on the new GPT-5.6 model family.
Different subscription tiers provide access to different model variants. Higher-tier plans can choose between models such as Sol, Terra and Luna while also selecting the desired level of reasoning effort for particular tasks.
This combination of advanced language models and an agentic execution environment allows the system not merely to answer questions but to plan, reason, execute tasks and produce finished outcomes independently.
A Major Step Towards AI Agents
With ChatGPT Work, OpenAI is responding directly to the growing competition in the market for AI agents.
The industry’s focus is shifting away from increasingly capable chatbots towards systems that can complete meaningful work autonomously. The real competitive advantage is no longer simply producing better answers but reliably finishing entire projects.
This also changes the role of generative AI itself. Increasingly, these systems are becoming execution engines that gather information, organise knowledge, prepare decisions and deliver results rather than merely responding to user input.
What This Means for Businesses
For organisations, ChatGPT Work opens up new opportunities to automate knowledge-intensive processes.
Market research, competitive analysis, reports, presentations, project planning and documentation can increasingly be expressed as complete workflows rather than collections of individual prompts.
As a consequence, employees are likely to spend less time executing routine tasks and more time defining objectives, supervising progress and making strategic decisions where human judgement remains essential.
That may ultimately prove to be the platform’s most significant contribution. ChatGPT Work does not simply replace the traditional chatbot. It fundamentally changes how knowledge work is organised.
A New Chapter for Generative AI
With ChatGPT Work, OpenAI has embarked on its most ambitious vision since launching ChatGPT. The emphasis is no longer on generating impressive responses but on completing meaningful work.
The agent plans projects, operates across applications, files and the web, creates finished documents, websites and presentations, performs recurring tasks autonomously and integrates Codex directly into the execution process.
In doing so, it creates an entirely new category of software that sits somewhere between a conversational AI, office productivity suite and digital employee.
Whether this vision ultimately succeeds will depend on how reliably these agents can manage complex projects over extended periods. What is already clear, however, is that ChatGPT Work represents a decisive shift in the AI industry—from conversation towards the autonomous execution of work.
