Intent
A human, AI agent, service, or workflow proposes an exact action.
How it works
From intent to evidence, without collapsing the steps.
AtlaSent inserts an organizational authority decision before a consequential effect and keeps authorization, permit verification, execution, and proof distinct.
Authority
Standing and delegated organizational authority is resolved for that actor and scope.
Policy + evidence
Current approvals, duties, resource state, and other required evidence are evaluated.
Decision
The canonical outcome is allow, deny, hold, or escalate.
Permit + verify
Where required, a bounded permit is issued and verified at the execution boundary.
Execution + evidence
The native operation may proceed; execution/effect are recorded only when actually observed or verified.
Why this matters
A plausible request, a valid identity, a tool permission, or an approval can each be necessary inputs. None alone establishes that the organization authorized the exact operation being attempted now.