AI / Vera
Conversation, reasoning, planning and assistance. Vera is personalised per Canonical Person only within permissioned boundaries.
System
One system. Clear authority.
VexlorGrid Home is the intended primary entry point. It presents capabilities without replacing the underlying authority that owns identity, consent, safety, data or external actions.
Apps
Focused workflows can appear as apps or experiences while reusing the same underlying authorities rather than creating isolated silos.
Black Box
Receipts and evidence expose what was proposed, authorised, executed, verified, held or recovered.
Device
Device capability remains subject to device identity, operating-system permissions, security and explicit action gates.
Authority model
Canonical Person remains the identity authority. Safety remains independent. Existing domain authorities remain authoritative for their own state. Models and interfaces do not acquire authority merely because they can see or describe information.
Failure model
Identical replay should produce no duplicate effect. Changed canonical content under the same identity is held as a conflict. Halfway failure preserves completed evidence and resumes only missing work. Restart should reconstruct deterministically. Work for one person must not contaminate another person's state.