OpenAI has released new data and commentary describing a shift in how ChatGPT is used worldwide — from a tool primarily used to ask questions to one increasingly used to complete tasks. The report, published by OpenAI, frames this as a move "from asking to doing," pointing to growth in usage patterns tied to document creation, coding, data analysis, and multi-step work rather than single-turn question answering.
According to OpenAI, the change reflects both product evolution — features like file uploads, code execution, browsing, and connected apps — and a shift in how users approach the tool once they understand its capabilities extend past simple Q&A. The company positions this as evidence that generative AI adoption is maturing past the novelty phase into embedded, repeated use inside real work processes.
Specific figures on task volume, industry breakdown, or the size of the shift relative to prior periods were not fully detailed in the source material available at time of writing; where OpenAI has cited proprietary usage statistics, those figures should be treated as company-reported and unconfirmed by independent third parties. What is confirmed is the directional claim: OpenAI's own framing of its usage trends emphasizes task completion over information retrieval as the growth vector for ChatGPT.
For B2B companies in the 10-200 employee range, this development matters less as a headline and more as a validation point. Most operators in this segment did not need OpenAI to tell them ChatGPT could do more than answer questions — many have already experimented with using it for email drafts, meeting summaries, or quick data pulls. What this report signals is that OpenAI itself is now optimizing the product, and its narrative, around task execution rather than conversation. That has downstream implications for how the tool gets built, priced, and integrated: expect continued investment in connectors, agentic workflows, and API-level task automation rather than chat-interface polish alone.
The practical takeaway for sales, support, and operations leaders is not to rush and "add AI" to a process, but to look at where their teams are currently doing repetitive, well-defined work — ticket categorization, lead scoring, invoice matching, follow-up scheduling — and evaluate whether a properly connected AI workflow (not just a chat window open in a browser tab) can absorb part of that load. The gap between casually asking ChatGPT for help and having it actually execute a task inside your systems is an integration and workflow-design problem, not a model-capability problem at this point. Companies that close that gap early get compounding time savings; those that treat this as another AI headline to skim past will be doing the same manual work in twelve months while their competitors aren't.
No pricing, availability, or feature-specific changes were announced alongside this report — it is a usage and positioning narrative from OpenAI, not a product launch.