ChatGPT Work agent launches as OpenAI targets enterprise automation
OpenAI introduced ChatGPT Work, an AI agent designed to automate complex workplace tasks, alongside the GPT-5.6 model family. The company reported that its Codex agent has surpassed 5 million weekly users, with significant adoption by finance teams for reporting and forecasting.

*this image is generated using AI for illustrative purposes only.
OpenAI is pushing ChatGPT beyond a question-and-answer tool and deeper into workplace automation with ChatGPT Work. The new AI agent can break down complex assignments, work autonomously over extended periods, and generate spreadsheets, presentations, documents, and web applications. The company also introduced GPT-5.6, a new model family that delivers stronger multi-step reasoning, better adherence to templates, and improved use of reference materials compared to previous versions.
The launch comes as AI companies race to win enterprise customers by embedding AI agents into everyday business workflows. Rival Anthropic has made a similar push with Claude Cowork, designed to help users plan and execute complex, multi-step tasks. OpenAI reported that adoption of its workplace tools has already expanded beyond software developers. Its Codex AI coding agent, now integrated into ChatGPT Work, has surpassed 5 million weekly users, with more than 1 million utilizing it for tasks outside traditional programming.
Finance teams have adopted the tools to accelerate month-end reporting and forecasting by pulling source data, reconciling figures, building presentations, and reviewing outputs. This automation allows employees to spend more time analyzing results and advising leadership rather than preparing reports. The deployment extends across the company's ChatGPT interface, Codex platform, and API services.
Availability and Security
The new models and workplace features are accessible through the ChatGPT, Codex, and API platforms. ChatGPT Work includes enterprise controls built on the existing ChatGPT Enterprise infrastructure. These controls feature permissions for tool access, allowed actions, and monitoring through a Compliance API. An "Auto-review" security feature blocked all tested attempts by external red teams to extract protected information during evaluations.
| Platform | Availability |
|---|---|
| ChatGPT | Yes |
| Codex | Yes |
| API | Yes |
Strategic Developments
OpenAI is folding the Codex app into the updated ChatGPT desktop experience while renaming the previous desktop client ChatGPT Classic. The company plans to transition users away from its standalone Atlas browser and expand ChatGPT features through a Chrome sidebar extension. The rollout comes as enterprises increasingly experiment with AI agents that can operate across software systems rather than simply generate text.
How will Anthropic and other competitors respond strategically to the integration of Codex into ChatGPT Work?
What specific metrics will enterprises use to evaluate the ROI of autonomous agents in non-technical departments like finance?
Will the consolidation of apps into the ChatGPT desktop experience trigger regulatory scrutiny regarding anti-competitive behavior?

































