Read-only start
Source coverage and remaining gaps are visible from the read-only start.
AUTONOMY WITH CONTROL
Runs continuously. Acts autonomously inside explicit authority.
Every run ends in a governed decision, a verified connector change or accountable human completion. Agents find, prepare, prioritize and move work on their own while consequential actions stay tied to the exact permission, decision and system change required.
Autonomy does the work. Authority controls the consequence. Proof closes the loop.
FindOpportunity or risk
BuildEvidence and work
ApproveExact authority
ExecuteInside permission
VerifyDestination state
ValueReceipt and confirmed result
CONTROL PLANE
Exact action. Named authority. Checked result.
CONTROL MODEL
Read-only first. Exact authority. Checked result.
Read-only first
Map sources and opportunity before granting write authority
Exact-action authority
Consequential changes bind to the approved operation
Approval-drift protection
A material change returns the action for renewed approval
Destination-verified
Connector changes are read back before verification
Allowed, held and blocked
The system classifies action state before execution.
Checked after action
SupraOS verifies the resulting state in the connected system.
TRUST CENTER
Inspect the control, data and proof model.
Security
Infrastructure, isolation, access, connectors, encryption and incident response.
Autonomy
What agents can do alone, what requires authority and what remains held.
Data handling
Source boundaries, stored state, hosting, retention, deletion and portability.
AI governance
Evidence grounding, model use, traceability, providers and VTWM.
Receipts and Value Ledger
Action records, connector verification, Recovery Records and confirmed value.
Security review
A buyer-ready diligence path covering data flow, controls, evidence and documentation.
Subprocessors
Managed infrastructure and deployment-specific provider disclosure.
Security contact
Security review requests, vulnerability disclosure and incident contact.