GWS
Governed Work System
An architecture for letting AI do real work by separating intelligence, authority, execution and verification, so autonomy is bounded and governance stays independent.
How GWS worksMontgomery PBC · A Delaware public benefit corporation
Powerful intelligence. Governed action. Human agency.
We build technology that lets people put capable AI to work without surrendering control, and tools that help people choose how they respond to their own lives.
What we believe
Capability and control should grow together.
AI is becoming able to do more of the world's work. The question that matters now is not only what it can do, but who decides what it is allowed to do, and how anyone can tell it did the right thing.
Powerful intelligence
We want the most capable models working on real problems. Capability is not the risk to be avoided; ungoverned capability is.
Governed action
Every consequential action should trace back to explicit, bounded, revocable authority, and be checked by someone other than the actor.
Human agency
The point of all of it is to widen what people can do and choose, never to quietly make the choice for them.
Our work
Two projects, one commitment.
One project governs what AI systems may do on our behalf. The other helps people govern how they respond to what happens to them.
Oath
From reaction to deliberate choice
Technology that aims to reduce unnecessary suffering by helping people recognize their reactions and choose how they want to respond.
About OathStatus: GWS and Oath are under active development. They are not yet generally available products, and what you read here describes our direction and principles rather than finished features.
Intelligence may propose. Only accountable authority may permit. Independent verification confirms what was actually done.
Building something that needs governed AI?
We welcome conversations with researchers, partners and organizations who care about accountable autonomy.