AI Disclosure
What this covers
Salvation's services are built around AI agents (an orchestrating consultant and specialist departments) working under human supervision. This page discloses, in plain language, how AI is used and what stays under human control.
Human oversight
High-risk actions always require explicit human pre-approval. Agent-claimed completion is never treated as proof - independent verification of actual system state is the intended standard before anything is reported as done. Today, this verification is a manual, human-performed process rather than an automated platform capability - see the Security & Governance page for full status labelling.
How client data is used with AI providers
We intend to route model calls through approved providers under standard data-processing terms, with client data not used to train third-party models. This commitment is required before activation and is not yet independently verified as implemented in this staging build.
Known limitations
AI-generated output can be wrong, incomplete, or reflect gaps in the data it was given. We do not claim "fully autonomous" operation, and we do not present AI-drafted work as final or approved until a human has reviewed it where required.