Share of anomaly/misuse alerts confirmed as genuine issues upon review, rather than false alarms.
Share of grounded prompt requests for which a complete, field-level provenance record can be produced on demand.
Share of automated cross-system entity matches confirmed correct in a sampled audit.
Count of any grounding event where data tagged with a residency or sovereignty constraint was routed to a processing endpoint outside its permitted zone.
Share of the domain's target recurring tasks (QBR-style summaries, headcount justifications, escalation responses) completed using the grounding platform…
Share of known confused-deputy / data-exfiltration attempt patterns, run against the pipeline in a red-team evaluation, that are actually caught before…
Estimated time no longer spent manually gathering and formatting data across systems before prompting an LLM, versus the pre-solution baseline (the docx's own…
End-to-end time from a resolved intent to a fully assembled context bundle, across all systems called.
Time from a template being submitted for review to a reviewer decision (approve or reject).
Share of grounded prompt requests served by a registered, approved template rather than resolved ad-hoc.
Share of SensitiveIntentDetected tags that a human reviewer subsequently confirmed were not actually risky.
Share of requests where one or more source systems failed but the response was NOT clearly flagged as partial.
Share of grounded responses that included at least one field older than its source-specific freshness threshold.