Elias unpacks the research behind the headlines in plain language.
Step-by-step guides on prompting, styles, and getting the most out of AI image generation.
Read the guidesPick a companion and get their take on this story
OpenAI is adding a dedicated Work tab to the ChatGPT mobile app, giving Pro and Plus subscribers the ability to trigger multi-step agentic tasks — sequences of actions the model carries out autonomously — using only their voice.
Agentic tasks, in plain terms, mean the model doesn't just answer a question — it takes a chain of actions to complete a goal: drafting a document, searching the web, summarizing results, and sending output somewhere, all in sequence. Until now, that kind of workflow on ChatGPT has been easier to manage on desktop. The Work tab consolidates those capabilities into a single mobile destination, and voice input is the primary interface.
The practical shift is meaningful. A creator who wants to research reference images, compile a mood board description, or draft a client brief while commuting can now speak the request rather than type it across a small screen. The model handles the sequencing.
The ChatGPT mobile app's new Work tab, which surfaces voice-driven agentic features for Pro and Plus subscribers.
Image: TechCrunch / TechCrunch AI
According to TechCrunch, the Work tab is specifically scoped to agentic use — it's not a general chat redesign, but a purpose-built space for task completion. That distinction matters: OpenAI is signaling that agentic workflows are a separate product surface from conversational chat, not just an advanced mode buried in settings.
Pro and Plus subscribers have access; free users do not. That's consistent with how OpenAI has rolled out its more compute-intensive features — agentic task chains are resource-heavy, and voice processing adds another layer on top. The gap does narrow the practical audience, at least initially.
For creators already on a paid ChatGPT tier, the immediate question is whether this replaces or supplements desktop agentic workflows. The answer, based on what's been announced, is that it supplements them: the Work tab brings parity to mobile rather than introducing capabilities that aren't available on desktop. The value is convenience and voice, not new underlying power.
OpenAI has been steadily expanding its model lineup — the GPT-6 Sol and Luna release earlier this year showed the company pushing on both capability and cost simultaneously — and the Work tab fits that pattern: shipping the same core capability across more surfaces and interaction modes.
The voice angle deserves attention on its own. Voice-based agentic control is a different interaction model from typing prompts, and it changes what kinds of tasks feel natural to delegate. Short, well-scoped requests — "summarize these three links and draft a one-paragraph description" — translate well to speech. Complex, iterative prompt engineering does not.
For AI-art creators specifically, that suggests the Work tab is most useful for the research and organization side of a project: gathering references, writing briefs, managing client communication. The image-generation step itself still happens in dedicated tools. Creators exploring what those tools can do will find the Charmloop model catalog a useful reference for what's currently available on the generation side.
What's still unclear is how well the voice interface handles correction and iteration — whether a creator can say "actually, make that shorter" and have the model revise mid-task, or whether each voice command starts a fresh sequence. OpenAI hasn't detailed the correction model publicly yet, and that will determine how fluid the mobile workflow actually feels in practice.
The Work tab rollout is happening now via standard app updates. Creators who want to get a sense of how agentic prompting differs from standard prompting before the feature reaches them can find a grounding in the Charmloop guides.
