ChatGPT Users Are Twice as Likely to Finish Work Tasks as Personal Ones — But What Gets Finished Differs by Product
People are more than twice as likely to use ChatGPT to complete a work task — write, code, analyze — than to do the same thing outside work, according to new country-by-country usage data OpenAI published August 6, 2026 as part of its ongoing “Signals” series. The report frames this as a shift from asking to doing: less Q&A, more task completion. Multimedia use — image and video generation or upload — is the fastest-growing category globally, now 7.8% of all messages and more than one in ten messages in Brazil and Colombia. Adoption is also broadening past early-adopter markets: Peru, Uruguay, and Costa Rica posted the fastest usage growth among countries in the Q2 rankings, and usage among people over 35 rose in nearly every country OpenAI studied.
The “more task completion at work” framing is worth holding next to Anthropic’s own usage telemetry for Claude Cowork, published July 7, 2026 from a sample of roughly 1.2 million sessions across more than 600,000 organizations. There, software development was just 8.7% of sessions — the largest category was business process and operations at 33.4% (consolidating updates into a report, building onboarding checklists, reconciling spreadsheets), with content creation and copywriting second at 16.4%. OpenAI’s data measures who is doing more finishing versus asking; Anthropic’s measures what that finishing actually consists of. Together they suggest the “task completion” story isn’t mostly about code — it’s operations and content work that used to be manual.
For companies deciding where to point a first agent deployment, neither data set alone answers the question, but together they narrow it: usage is shifting from asking questions to finishing tasks, and the tasks getting finished skew toward operations and content, not just engineering. A rollout plan built only around developer tooling is optimizing for the smaller share of what people actually do with these tools once they’ve moved past novelty use.