Assign a task to an AI the way you assign it to a person
No chat window, no prompt box, no separate tool. You put the teammate on the task, and the work comes back on that task.
The UnaTask team, Building UnaTask. Aug 2, 2026. 4 min read.
Most AI at work lives in a chat window that knows nothing about your work. So the day goes like this: you open the tab, explain the project from scratch, paste in the context, get something decent back, copy it out, and paste it into the place it was always meant to go. The AI did four minutes of work and you did eleven.
The copying is not the annoying part. The annoying part is that none of it is anywhere. The request is in a chat log, the result is in a document, and the task it belongs to has no idea any of this happened. A week later nobody can find which version was the real one.
The teammate goes on the task, not in a tab
An AI teammate in UnaTask is a member of your team. It has a name, a face, a job description, and it appears in the assignee picker next to everybody else. So you give it work the way you give anyone work: you put it on the task.
- Add the teammate to the assignees of a task. That is the whole action. There is no separate button and no prompt to write.
- The task itself is the brief: the title, the description, the checklist, the project it sits in.
- It starts within about a minute. While it works, the task shows how long it has been going and how long that teammate usually takes.
- The work comes back as a comment on the same task, with any files attached, and the task moves to Review.
If something goes wrong it says so on the task, in a comment, and tells the person who asked. No silent failures, no empty result you notice three days later.
Why the task is the right place
Because everything the work needs is already there, and everything the work produces belongs there. The brief does not have to be re explained, because the task is the brief. The output does not have to be filed, because it lands on the thing it was for. And the history stays in one readable line: this was asked, this came back, this is what we thought of it, this is the version we used.
It also means the rest of your team can see it. An AI teammate’s work is not hidden in somebody’s private chat history. It is a comment on a shared task, with a name on it, in the project where the work lives. Your colleague can read it, disagree with it, or pick it up on Monday.
The small shape of this matters more than it sounds. There is nothing new to learn, because you already know how to assign a task. The AI is not a mode you enter. It is a name in a list you were going to open anyway.
If you have to explain the project before the work starts, the AI is in the wrong place.