OpenAI Launches ChatGPT Work: Long-Running Agentic Task Automation

OpenAI launched ChatGPT Work on July 9, 2026, as an autonomous productivity agent that can execute multi-step professional tasks over hours by pulling context from Slack, Teams, Google Drive, SharePoint, email, and CRMs via plugins [1]. The tool includes Scheduled Tasks for recurring or event-triggered automation and a Sites feature for publishing interactive web apps directly from ChatGPT [1][2]. OpenAI reports over 5 million weekly Codex users—more than 1 million now using it for non-software tasks—and says nearly 100% of its internal teams use ChatGPT Work and Codex [1]. The launch directly addresses a time limit in the earlier Atlas Agent Mode, which stopped automated tasks after a few minutes [2].

ChatGPT Work moves OpenAI's product from reactive Q&A to sustained autonomous action across professional workflows. If the capabilities hold up in practice, this represents a meaningful shift in how enterprise teams can delegate repetitive knowledge work to AI—but early UX fragmentation and unresolved questions about oversight in complex multi-hour tasks suggest the product is still early-stage.

OpenAI launched ChatGPT Work on July 9, 2026, positioning it as an autonomous agent that can stay engaged with a professional project for hours, break goals into steps, pull context from connected apps, and deliver finished outputs for human review [1]. The tool integrates with Slack, Microsoft Teams, Google Drive, SharePoint, email, and CRM systems via plugins, and includes a Scheduled Tasks feature for recurring or event-triggered automation—such as refreshing meeting agendas or monitoring dashboards—without requiring user prompts each time [1][2]. A new Sites feature, in public beta, lets users publish interactive web apps and dashboards directly from ChatGPT [1]. OpenAI describes this launch as "the first step towards a broader vision for ChatGPT—where intelligence goes beyond answering questions" [1].

The launch directly addresses a limitation of Atlas Agent Mode, OpenAI's earlier browser-based agent, which Ars Technica notes stopped automated tasks after only a few minutes [2]. ChatGPT Work extends this to hours-long sustained operation, with the tool requiring user approval before executing important actions [2]. The underlying Codex platform had already introduced scheduled automations that can continue from existing conversation context rather than starting fresh [3], laying technical groundwork for the Work product. OpenAI reports more than 5 million weekly Codex users, with over 1 million using it for tasks outside software development, and says nearly 100% of its internal teams—including finance and sales—now rely on ChatGPT Work and Codex [1].

Independent commentary at launch has been limited but pointed. Simon Willison flagged a concrete UX inconsistency: Work conversations started on web or mobile run in the cloud and do not appear in the desktop Work interface, while the desktop app can additionally access local files and desktop applications [4]. Willison characterized OpenAI's own explanation of this split as failing to resolve the confusion—a sign of a rushed or under-coordinated release. Ars Technica's coverage was largely descriptive, contextualizing the launch against the Atlas precedent and conveying OpenAI's claims about multi-hour task execution and workflow automation [2]. No independent testing or third-party performance assessments have surfaced yet.