Company / Manifesto
Our direction is simple: understand the outcome, act with permission, and make the result something a person can examine.
Useful autonomy
has no shortcuts.
A useful assistant needs more than the latest message. It should connect the request to the context, commitments, and preferences that give the work meaning.
A question is part of a project. A project is part of someone’s responsibilities. Our aim is to keep that working context available without making people start from the beginning every time.
Preparation and coordination can be delegated. Consequential decisions need clear permission, and people should be able to see and direct what their agent is doing.
An agent should make the next decision easier to understand. It should keep the intended action, its scope, and the person who can authorize it close to the work.
We are building toward assistance that follows the outcome through changing conditions, handles failure honestly, and makes completion something you can check.
Progress matters, but so does knowing what is unfinished. We want the output, the supporting sources, and the next decision to be available for review.
