Trust is infrastructure
ElseHour becomes more useful as it understands more of your work and life. Its authority should not grow with that context. Every source, inference, recommendation, and effect needs a visible boundary.
Account separation
Each item is designed to stay attached to its source account, identity, portfolio or workspace, time zone, permission, and evidence. Work and personal accounts can inform one view without being silently merged.
Permissioned context
You choose which sources may contribute context and for what purpose. Derived facts are designed to retain their provenance and permission scope; an inference never becomes permission to inspect unrelated information.
Visible uncertainty
Confirmed facts, observations, inferences, estimates, and unknowns remain distinguishable wherever the difference could change a decision.
AI boundary
AI may help form an insight, option, or draft. It does not decide canonical workflow state, grant itself permissions, or quietly turn a suggestion into a fact.
Bounded authority
ElseHour is designed to keep external effects outside AI authority. A suggested reply remains a draft unless an explicitly supported workflow and your confirmation make it an action.
Outcome history
The estimate, choice, correction, and actual outcome stay connected. Potential time return remains labeled as estimated until evidence supports a verified result.
The trust contract
Vision and present state
The principles above describe the trust model ElseHour is being built to enforce. For the precise state of connector access, AI processing, external actions, and website collection, read the current limitations before relying on a capability.