Legal readiness · Draft
Responsible AI standard
Aspinall People Systems is in development. This standard sets a deliberately strict boundary: AI supports controlled work; people remain responsible for consequential decisions.
Permitted support
AI may help retrieve authorised information, summarise supplied records, identify missing evidence, calculate transparent dates and draft source-linked administrative material. Outputs must show limitations and be reviewable.
Human decision required
A named, accountable person must verify evidence, interpret obligations, approve policy changes and decide any recruitment, grievance, disciplinary, capability, absence, redundancy, pay or safety outcome.
Prohibited automation
The system must not autonomously hire, reject, rank candidates, discipline, dismiss, set pay, diagnose health, infer protected traits, score employee value or make another significant employment decision.
Data boundary
No real employee, applicant, case, payroll, health or safety data is accepted by the current public preview. Production use requires identity controls, tenant isolation, privacy assessment, retention, incident response, contracts, security testing and a clear human-review design.
Traceability
Future consequential outputs must preserve the source set, effective date, prompt or task, model/template version, confidence/uncertainty, reviewer, approval and later override. Model confidence is never treated as evidence.
Current preview
Aspinall Intelligence in the preview uses deterministic, pre-authored responses to synthetic records. It does not call a live AI model and does not send text externally.