we_are_coded.by CODE · The world, decoded
БГ

OpenAI · Research

AI is reshuffling who does what: almost half of profession-specific tasks come from another profession

On 27 July OpenAI released the first report in its new Work at the Frontier series: an analysis of over 800,000 work messages in ChatGPT shows that 43.5% of profession-specific tasks come from another profession. The designer handles his own budget. The salesperson analyzes data. The marketer fixes his own website. The division of labor is being rearranged before job descriptions have caught up.

Source
OpenAI
Published
The event
Checked on
Responsible editor
Cvetelin Ivanov
How we work
MethodCorrections

Most research on AI and work asks the same thing: can the model do the tasks of a given profession. OpenAI's new report flips the question. Who is doing the tasks now that AI is at hand? The answer is that the boundaries between professions are already shifting.

Fact

The facts: On 27 July 2026 OpenAI published Work at the Frontier - the first report from its economics research team's new series. The analysis covers over 800,000 work messages from US users of ChatGPT. 16.8% of all work messages concern tasks from another profession; for profession-specific tasks - after excluding general activities like writing, summarizing and planning - the share reaches 43.5%. The biggest boundary-crossers are customer service (77% of their profession-specific messages are foreign tasks), designers (75%) and human resources (69%). Engineering and marketing tasks 'travel' the farthest - almost every other profession does them. In small companies (2-5 employees) foreign work is 18.9% of messages, versus 16.3% at large ones (over 100). All figures come from OpenAI's own platform data. Source: OpenAI, 27.07.2026.

The report calls the phenomenon 'task crossover.' Put simply: work that by the book belongs to one profession shows up in the hands of another. The small business owner reviews his own contract and does his own math. The marketer fixes his own website without waiting for a developer. The task no longer travels to the specialist - it gets solved by whoever ran into it first.

There is one detail that tells me more than the big number. For designers, 35% of the work done with AI is foreign - finance, law, marketing. In the other direction, almost nothing: design work shows up in other professions in only 1.7% of messages. The designer takes from everyone. Almost no one takes from him. Some professions absorb tasks, others give them away. Engineering gives away - everyone 'borrows' something technical, usually to fix on their own a problem they used to wait on a person for.

The specialist doesn't disappear. The waiting for the specialist does.

One caveat before we draw big conclusions. This is one company's window onto its own product. OpenAI sees what passes through ChatGPT, not all the work in the world - and it has an interest in the picture looking exactly like this. But the direction is confirmed by the daily life of every small business: there are no departments there, just people close to the problem. Job descriptions get rewritten slowly. People are already working the new way. Want to see where your profession is headed - look at what you actually do every day, not what's on your business card.

The visual is generated code-art, no third-party images.