What happens when your chatbot and your AI coworker become one app?

The line between "asking" and "handing it over" is about to disappear, and nobody has said how the limits will work.

Right now, in one big AI app, there are two very different things living side by side. Chat is where you ask a question and get an answer. Work is an agent that takes a long job and actually does it. Next step: they get merged into one experience.

Wait, what can the agent already do?

Quite a lot. On the desktop it can use your computer on your behalf: click, type, move files between apps. You decide what it is allowed to touch and when it must ask for approval first. That part already exists today.

So after I give permission, is the chat just a remote control for my whole machine?

More or less, yes, within the permissions you grant. The real change in the merger is who decides when a conversation turns into action. Today you pick the mode. Merged, the model gets to notice that your question is really a task and start doing it.

Another lab is heading the same way: its agent mode began running in the cloud for paid tiers in October and is being described as one single experience with chat.

What nobody has said yet is when this lands, or how usage limits will be counted when a quick question and an hours-long job live in the same box. Expect that to be the thing people argue about first.