Start with the outcome.
The primary input is deliberately simple: “What do you want done?” The product works backward from that request instead of forcing users to assemble a workflow first.
Done is an xYz product concept built around execution rather than conversation: turn a request into a plan, move work through clear states and ask for approval only when the next action actually needs the user.
THE PRODUCT IDEA
Most assistants are optimized to answer. Done explores a different interaction model: the user states the outcome, the product turns it into executable work and keeps the boundary between what can proceed automatically, what needs approval and what is already complete.
The primary input is deliberately simple: “What do you want done?” The product works backward from that request instead of forcing users to assemble a workflow first.
Work moves through NOW, DOING and DONE so the user can understand what is waiting, what is active and what has actually finished.
Done separates tasks the system can handle from actions that require explicit approval, keeping execution fast without hiding meaningful decisions.
The MVP explores requests such as trip preparation, follow-up, messages and purchases to test whether one execution model can remain useful across very different personal tasks.
The shift is small in language and large in product behavior: stop asking AI only for answers and start asking what can be completed.