OpenAI has launched ChatGPT Work, an agent designed to tackle ambitious, multi-step projects by integrating with user applications and workflows. This new capability allows ChatGPT to independently gather information, create finished materials like spreadsheets, presentations, and documents, and persist with complex tasks for extended periods. Powered by the new GPT‑5.6 model, ChatGPT Work demonstrates advanced reasoning for multi-step tasks and material generation that adheres to user-specified templates and references.

Users can now delegate tasks such as analyzing budget variances, converting source materials into marketing briefs, or preparing for sales meetings. The system offers transparency, allowing users to monitor progress, provide input, and approve critical actions. ChatGPT Work can manage entire workflows, for instance, transforming customer research into a campaign brief, generating marketing assets from that brief, and adapting them for different markets, maintaining context throughout the process. Scheduled Tasks enable projects to advance even when users are offline, with capabilities like automatically updating documents based on new messages from platforms like Slack and Microsoft Teams.

Early internal testing at OpenAI showed significant efficiency gains. Sales teams used ChatGPT Work to generate a tailored proof of concept within 24 hours, a process typically taking weeks. Finance teams reduced month-end close and forecasting from days to hours. The platform is rolling out across various ChatGPT plans, with a desktop app offering enhanced capabilities including a built-in browser for web-based research and a Computer Use feature that allows ChatGPT to execute tasks across local applications. The Codex app is now merged into the ChatGPT desktop app, retaining its coding agent functions with new performance improvements and multi-repository support.