OpenAI is bringing voice-based agentic features to the ChatGPT mobile app, letting users trigger real work — drafting documents, summarizing emails, building presentations — simply by talking, the company announced Wednesday, as reported by TechCrunch.

The rollout extends to smartphones capabilities that OpenAI previously offered on desktop, and it lands as voice becomes one of the fastest-growing ways users interact with AI assistants for complex tasks.

What subscribers can do by voice

Plus and Pro subscribers will be able to use the Work tab in the ChatGPT mobile app to create documents, draft emails, or summarize Slack messages by voice. The same voice workflow covers heavier jobs the Work tab already supports, such as building sites, creating presentations, using the cloud browser, and digging into the finances area of ChatGPT.

Free and Go users get a narrower slice of the update, gaining the ability to work with plugins and connected apps.

OpenAI said voice conversations will now produce richer text output, and users can move fluidly between voice and text modes. Perhaps the most practical addition for people juggling commutes and desktops: a conversation started on the go can be resumed later on the desktop app.

The road from GPT-Live to mobile agents

The update is the mobile chapter of a story OpenAI began in July, when it launched GPT-Live, its new conversational model, and later wired it into the desktop app so users could complete Work tab tasks and build apps in the Codex tab by voice. Wednesday's announcement closes the loop by bringing the same voice-driven task execution to phones.

Why voice agents on mobile matter

The shift reflects how usage patterns are changing. More users are turning to voice to issue commands to AI assistants for complex tasks, and the phone is where those commands most naturally happen — between meetings, in the car, or away from a keyboard.

Until now, the most powerful agentic features in ChatGPT lived on desktop. Users who wanted ChatGPT to build a presentation or run a multi-step research task needed to be at their computer, typing instructions into the Work tab. The mobile app, for all its popularity, was largely a conversational surface. Wednesday's update collapses that distinction: the phone becomes a legitimate place to kick off substantive work, with the voice channel as the interface.

The subscription split

The rollout also sharpens the lines between ChatGPT's subscription tiers. Plus and Pro subscribers — the plans that already carry the highest usage limits and access to OpenAI's most capable models — get the full voice-driven Work tab experience, including document creation, email drafting and Slack summarization. They can also build sites, create presentations, use the cloud browser and access the finances area of ChatGPT, all by voice.

Free and Go users receive a more limited set of capabilities centered on plugins and connected apps. That tiering follows a familiar pattern for OpenAI: prove a capability on desktop, then bring a mobile version whose most valuable features push users toward paid plans. For OpenAI, the move deepens the moat around its paid tiers at a moment when rivals are aggressively courting the same users.

How it compares with Anthropic's approach

The competitive backdrop matters here. TechCrunch notes that Anthropic recently made handoffs between mobile and desktop easier for its own users and merged its Cowork and Chat interfaces into one experience. OpenAI, by contrast, is still keeping chat and workspaces separate — a deliberate product split that keeps casual conversations distinct from the workspace where files, projects and tools live.

The two companies are effectively running opposite experiments in AI-assistant workflow design. Anthropic's merged interface bets that users want a single unified surface that follows them across devices. OpenAI's bet is that chat and structured workspaces serve different needs and should stay distinct, with voice acting as the connective tissue — start a task by talking on mobile, refine it with a keyboard on desktop.

What to watch next

The obvious next question is how far voice agents travel from the Work tab. OpenAI's desktop integration already reaches into Codex, its app-building tool, and mobile parity there would turn phones into full development surfaces. OpenAI has not announced timing for that.

Also unresolved is reliability. Agentic voice tasks — drafting, summarizing, browsing — involve multi-step execution where errors compound, and mobile environments add interruption and context-switching. Early user experience will determine whether voice-driven work becomes a daily habit or remains a demo-friendly feature.

Either way, the direction of travel is clear: the assistant that answered questions by voice two years ago is now expected to finish the task before you reach your desk. With Wednesday's update, OpenAI's mobile users can start that task with a sentence — and pick it up on a bigger screen when they arrive.

---

Stay Ahead of AI

Get the latest AI news, analysis, and breakthroughs — all in one place.

Read more AI news →