Disclose the AI.
A named digital persona should make responsibility easier to understand, not trick someone into believing they are talking to a biological employee.
07 / Trust
Autonomous
does not mean uncontrolled.
Afluma’s governance model is designed around transparent AI identity, least-authority execution, evidence, human approvals and operational recovery.
Trust model
The boundary is
part of the product.
A system is not trustworthy because it sounds confident. Trust comes from knowing who owns the task, what context was used, what the system was allowed to do and how the outcome was checked.
Bound the action.
Tool access, data access and consequential decisions are permissioned by task. A role title is never blanket authority.
Show what supports the claim.
Knowledge and public claims should preserve lineage, uncertainty and review status rather than laundering assumptions into facts.
Verify and recover.
Changes need an accountable owner, observable result and recovery path before greater autonomy is justified.
Governance loop
Request. Scope.
Act. Verify.
Every increase in authority should preserve a clear chain from human intent to bounded execution and observable result.
Human intent→Permissioned work→Verified outcome
Afluma / Pilot