OpenAI says ChatGPT Work and Codex together have passed 10 million weekly active users, putting a fresh spotlight on its attempt to turn ChatGPT from a prompt-driven assistant into a workplace agent that can assemble documents, analyze data, and work across connected business apps.
The milestone is a combined figure, not 10 million users for each product. As reported by Bloomberg and echoed in OpenAI’s recent product messaging, the two tools have grown rapidly since ChatGPT Work launched on July 9, alongside GPT-5.6. For Windows users, the practical change is that OpenAI’s desktop app is increasingly becoming a place to hand off longer tasks—not simply generate text in a chat window.
ChatGPT Work is designed for the tasks surrounding the work people are actually accountable for: gathering project updates, reviewing meeting notes, cleaning spreadsheet data, producing briefings, and drafting presentations. It can pull context from apps and files that a user has explicitly connected, then create deliverables including documents, sheets, slides, reports, and simple sites.
OpenAI’s release notes describe Work as an agent for multi-step jobs that may take hours, with users able to monitor progress, redirect it, and approve important actions. Scheduled Tasks add another dimension: a workflow can run once, repeat on a schedule, trigger from an event, or watch for changes.
That matters more than another AI writing button. A user who needs a weekly executive project update may no longer have to manually reconcile Outlook mail, Teams discussions, task trackers, meeting notes, and a spreadsheet before writing the summary. The value proposition is the work around the work.
That makes the number a measure of early adoption and interest, rather than proof that millions of organizations have embedded autonomous agents into critical business processes. Still, it is a meaningful signal: the market for AI that completes multi-application workflows is moving quickly beyond developer tools and experimental pilots.
OpenAI has highlighted internal use across finance, sales, and other teams, while customer examples focus on reducing manual preparation. The company says NVIDIA used ChatGPT Work to organize GTC-related account and meeting information, and that Virgin Atlantic used it to speed competitor-analysis work tied to its five-year planning process. Such examples are vendor-provided case studies, not independently audited productivity benchmarks, but they illustrate the specific workflows OpenAI wants customers to delegate.
Microsoft 365 Copilot’s advantage is its native position inside Microsoft 365: Outlook, Teams, Word, Excel, PowerPoint, SharePoint, and the Microsoft Graph. For organizations standardized on Windows and Microsoft 365, that integration can make Copilot the more natural fit for governed collaboration, tenant controls, retention policies, and familiar Office workflows.
Google’s Gemini targets the equivalent Google Workspace territory through Gmail, Docs, Sheets, Meet, and Drive. Anthropic’s Claude Enterprise has focused on long-context reasoning, enterprise knowledge work, and coding-oriented workflows. OpenAI is attempting to compete across those boundaries by making ChatGPT Work a flexible orchestration layer for selected files, services, and apps.
For IT teams, the distinction is important. Microsoft 365 Copilot often begins with the question, “How can AI assist inside our existing Microsoft tenant?” ChatGPT Work begins closer to, “What outcome do you want, and what sources should the agent use?”
That makes permissions, connector scope, auditability, data retention, and human approval central deployment concerns—not secondary settings. Enterprises should be especially cautious about granting agents authority to send external messages, modify records, access financial data, or act on HR and legal materials.
The sensible early use cases remain read-heavy and approval-gated: creating a project brief, finding missing launch dependencies, preparing an account review, or flagging anomalies in a workbook. A useful instruction is not “handle this,” but “review this, show your sources, prepare a draft, and do not take action without approval.”
ChatGPT Work’s next test is whether the 10 million-user surge translates into repeatable, controlled workflows on real Windows desktops. The competition is no longer just about which assistant writes the best answer; it is about which platform can safely finish the most valuable preparation before a human has to make the decision.
OpenAI Is Targeting the Preparation Layer
ChatGPT Work is designed for the tasks surrounding the work people are actually accountable for: gathering project updates, reviewing meeting notes, cleaning spreadsheet data, producing briefings, and drafting presentations. It can pull context from apps and files that a user has explicitly connected, then create deliverables including documents, sheets, slides, reports, and simple sites.OpenAI’s release notes describe Work as an agent for multi-step jobs that may take hours, with users able to monitor progress, redirect it, and approve important actions. Scheduled Tasks add another dimension: a workflow can run once, repeat on a schedule, trigger from an event, or watch for changes.
That matters more than another AI writing button. A user who needs a weekly executive project update may no longer have to manually reconcile Outlook mail, Teams discussions, task trackers, meeting notes, and a spreadsheet before writing the summary. The value proposition is the work around the work.
The 10 Million Figure Needs a Little Context
The reported 10 million total combines ChatGPT Work with Codex, OpenAI’s coding agent. OpenAI has not publicly broken the total down by product, company size, paid status, or the precise activity threshold required to count as an active user.That makes the number a measure of early adoption and interest, rather than proof that millions of organizations have embedded autonomous agents into critical business processes. Still, it is a meaningful signal: the market for AI that completes multi-application workflows is moving quickly beyond developer tools and experimental pilots.
OpenAI has highlighted internal use across finance, sales, and other teams, while customer examples focus on reducing manual preparation. The company says NVIDIA used ChatGPT Work to organize GTC-related account and meeting information, and that Virgin Atlantic used it to speed competitor-analysis work tied to its five-year planning process. Such examples are vendor-provided case studies, not independently audited productivity benchmarks, but they illustrate the specific workflows OpenAI wants customers to delegate.
Microsoft 365 Copilot Still Has the Native Windows Advantage
ChatGPT Work enters a workplace-AI market where Microsoft, Google, and Anthropic already have distinct strengths.Microsoft 365 Copilot’s advantage is its native position inside Microsoft 365: Outlook, Teams, Word, Excel, PowerPoint, SharePoint, and the Microsoft Graph. For organizations standardized on Windows and Microsoft 365, that integration can make Copilot the more natural fit for governed collaboration, tenant controls, retention policies, and familiar Office workflows.
Google’s Gemini targets the equivalent Google Workspace territory through Gmail, Docs, Sheets, Meet, and Drive. Anthropic’s Claude Enterprise has focused on long-context reasoning, enterprise knowledge work, and coding-oriented workflows. OpenAI is attempting to compete across those boundaries by making ChatGPT Work a flexible orchestration layer for selected files, services, and apps.
For IT teams, the distinction is important. Microsoft 365 Copilot often begins with the question, “How can AI assist inside our existing Microsoft tenant?” ChatGPT Work begins closer to, “What outcome do you want, and what sources should the agent use?”
The Governance Problem Arrives Before the Productivity Gain
The more useful an agent becomes, the more sensitive the data it needs. An effective meeting-prep or project-status workflow may require access to mailboxes, calendars, shared drives, CRM records, tickets, customer files, and internal chats.That makes permissions, connector scope, auditability, data retention, and human approval central deployment concerns—not secondary settings. Enterprises should be especially cautious about granting agents authority to send external messages, modify records, access financial data, or act on HR and legal materials.
The sensible early use cases remain read-heavy and approval-gated: creating a project brief, finding missing launch dependencies, preparing an account review, or flagging anomalies in a workbook. A useful instruction is not “handle this,” but “review this, show your sources, prepare a draft, and do not take action without approval.”
ChatGPT Work’s next test is whether the 10 million-user surge translates into repeatable, controlled workflows on real Windows desktops. The competition is no longer just about which assistant writes the best answer; it is about which platform can safely finish the most valuable preparation before a human has to make the decision.
References
- Primary source: InfotechLead
Published: 2026-07-31T12:45:57+00:00
Loading…
infotechlead.com - Related coverage: community.openai.com
Loading…
community.openai.com - Related coverage: openai.com
Loading…
openai.com - Related coverage: techradar.com
Loading…
www.techradar.com - Related coverage: help.openai.com
Loading…
help.openai.com - Related coverage: chatgptaihub.com
Loading…
chatgptaihub.com