ChatGPT Voice Is Becoming a Way to Delegate Work, Not Just Talk
OpenAI is bringing Voice into Work on mobile, letting eligible users start documents, presentations, connected-app tasks, and browser workflows by speaking.
## BODY
ChatGPT Voice is becoming more than a way to ask questions and hear spoken answers. OpenAI is bringing Voice into Work on mobile, allowing eligible users to speak instructions for tasks such as creating documents, presentations, and spreadsheets, using connected apps, or working in a browser. OpenAI says an unfinished Work task can continue in text after a Voice call ends.
The change gives ChatGPT Voice a different role: it can now be an entry point for getting work started rather than only a conversational interface. The mobile app can capture the request while a user is away from a desk, while the task itself can continue through Work.
ChatGPT Voice Can Now Start Work Tasks on Mobile
OpenAI's September 23 release notes say Voice is available in Work on both web and mobile. Users can ask Work to create documents, presentations, and spreadsheets, use connected apps, or work in a browser. The exact tools and actions available still depend on the user's plan, connected apps, permissions, and usage limits.
TechCrunch reports that Plus and Pro users can use the Work tab on their phones for tasks such as creating a document, drafting an email, or summarizing Slack messages. OpenAI's own documentation confirms the broader Work capabilities, while noting that existing app permissions and action restrictions continue to apply.
This builds on the more conversational Voice experience introduced with OpenAI Launches GPT-Live - a Full-Duplex Voice Model That Talks While It Listens. The important difference here is where the conversation can lead: a spoken request can now become a Work task.
The Key Change Is What Happens After You Stop Talking
The most useful part of the rollout may be the handoff between Voice and Work. OpenAI says that if a task is still running when a user ends the Voice call, Work can continue the task in text. That means the voice session does not have to remain open until the work is finished.
Voice gives users a spoken way to start or coordinate Work tasks, but the underlying permissions, connected apps, usage limits, and approval requirements still apply.
OpenAI also says users can follow written responses and task results in the chat. For actions that require approval, the user must review the action on screen; spoken approval is not supported. This keeps the voice experience connected to the normal controls around app access and external actions.
The result is a workflow that can begin naturally on a phone and continue as ordinary text-based work. That is different from a voice assistant that simply answers a question and waits for the next one.
Mobile Becomes the Starting Point for Longer AI Workflows
Work already supports longer tasks such as research, analysis, and creating documents, spreadsheets, presentations, reports, or Sites. OpenAI says cloud Work chats sync across web, mobile, and desktop, so a Work chat started on mobile or the web can be continued on desktop.
That makes the phone less of a destination for the task and more of a starting point. Someone can describe what needs to be done while away from a computer, then return to the same Work context later. Saganote has previously covered the broader distinction between these workflows in ChatGPT Work vs Claude Cowork: Same Category, Different Customer.
The model is also expanding beyond documents. Connected apps can provide additional actions where the account supports them, while browser-based work can be handled inside Work. The available actions are still constrained by the user's permissions and the connected service's capabilities.
Voice Is Moving From Conversation Toward Delegation
The broader shift is straightforward: Voice can now be used to tell ChatGPT what work needs to happen, not just what information the user wants to hear. That makes natural speech an input method for agentic workflows.
There are already signs of this direction elsewhere in the market. Saganote previously covered Gemini 3.8 Live Lets Voice AI Think and Work in the Background, while Google's productivity tools have also moved toward voice-driven actions in Google Workspace Voice Features Land in Gmail, Docs, and Keep - Talk to Your Apps and They Listen.
ChatGPT's implementation keeps Voice tied to the Work environment rather than turning every spoken request into an unrestricted action. That distinction matters because the connected apps, permissions, and approval steps remain part of the workflow.
Availability Depends on the ChatGPT Plan and Access
OpenAI says Voice in Work requires both Voice and Work access. Free and Go users can use Voice in Chat with the plugins their plans support, while Voice in Work depends on having access to both experiences. Existing app connections, permissions, and usage limits continue to apply.
For eligible users, the practical change is that starting a Work task no longer requires typing the initial instruction. A user can speak the request, follow the written result, and let an unfinished task continue in text after the voice session ends.
That makes ChatGPT Voice less about replacing typing and more about removing typing as the first step into a longer AI workflow.