Trust
Controls are scoped to the workflow, data, providers, environment and current evidence rather than presented as universal certification.
Cannot conclude: That a method statement is an independent assurance report or that one engagement control applies universally.
Review the recordSources, permitted use, providers, retention, deletion and export expectations are mapped for the specific workflow.
Cannot conclude: A universal data-processing posture outside the website and a separately contracted engagement.
Review the recordAI is used inside a named workflow with explicit inputs, evaluation, review points, unsupported-answer behaviour and human release.
Cannot conclude: Universal accuracy, autonomous authority or fitness outside the tested workflow boundary.
Review the recordOwnership, documentation, exit and any support expectations are made proportionate to the engagement.
Cannot conclude: That CortexLabs remains an unbounded operator, permanent dependency or continuously available support provider.
Review the recordWebsite collection, analytics, contact and public claims are kept separate from the terms governing a contracted Assessment or Implementation.
Cannot conclude: That a website policy is legal advice or a complete contract for an engagement.
Review the recordProcurement
A precise control question should receive a scoped fact, the available evidence and any gap—not assurance language copied from a generic trust centre.
Controls and terms are agreed for the specific engagement.
Unsupported assurance language is refused rather than approximated.
Do not include credentials, personal records or confidential production material in an initial message.