Authority
How can a person express exactly what an agent may decide, where it must ask, and how permission expires?
Independent research & product company
for useful, human-aware AI.
Research agenda
We study what has to become true before an AI system can participate meaningfully in human work.
How can a person express exactly what an agent may decide, where it must ask, and how permission expires?
Can systems reason about roles, relationships, timing, tone, and consent without flattening people into metadata?
What should persist across devices and conversations—and what should be deliberately forgotten?
How do we make an agent’s evidence, actions, uncertainty, and responsibility legible at the moment they matter?