Skip to content

Virtual Agent call audit

Virtual Agent Call Audit is the operational history for automated voice interactions and the decisions made during them. It is available within your assigned site scope; platform administrators can query all sites.

Open General Administration and select Virtual Agent Call Audit. If a site was already selected, the dialog starts there; otherwise it can show All Sites within the available scope.

Find a call

The default date range is the last 30 days. Filter by site, AllBox, date range, call status, outcome, or search.

Filter Behavior
Site Limits the provider request to one site; users with only one relevant site may not see this selector.
AllBox Limits the request to one AllBox available in the selected scope.
Date range Last 24 hours, 7 days, 30 days, or any time, evaluated from Started.
Call status Requested, Denied, Canceled, Connected, Access Granted, or Closed.
Outcome Completed, Access Granted, Escalated, Denied, Canceled, or Failed.
Search Matches site, AllBox, status, outcome, reason, and summary.

Search updates after a short pause; use Apply for popover filters and Refresh for newly recorded calls.

Each underlying call-search request returns at most 200 calls before the date-range, outcome, and text-search filters are applied in the portal. When an expected older call is absent, select one site and a specific call status, then retry before concluding there is no record.

Interpret status and outcome

Status describes the call's lifecycle state. Outcome describes the operational result. They answer different questions: a call can be closed after escalation, denied before connection, or complete without granting access.

The grid shows Started, Site, AllBox, Status, Outcome, Duration, Cost, and Summary. Dates are converted to the site's time zone when available. Cost is displayed with the recorded currency and should be treated as operational usage data rather than a finalized customer invoice.

Open the eye action or double-click a row for full details.

Call overview

The Overview tab shows:

  • site, AllBox, and post;
  • status, outcome, and reason;
  • started and ended time;
  • total duration and billable time;
  • voice provider;
  • total cost, resale amount, and currency;
  • the recorded call summary.

Not recorded means the audit service did not return that value. A missing end time can indicate an active or incomplete call, but confirm its status and timeline before treating it as stuck. A missing summary does not mean no events occurred.

Event timeline

The Event timeline contains up to 300 events for the selected call. It shows Time, Event, Category, Status, Outcome, and Reason. Sort Time to reconstruct the progression through request, policy, model, approval, provider, access-control, escalation, media, and closure stages.

Use the timeline to identify the first stage where actual behavior diverged from the expected workflow. For example:

  • policy or approval denial before a provider connection;
  • a completed conversation followed by failed access control;
  • escalation after a timeout or unsuccessful automated outcome;
  • a provider event without a corresponding device or access event.

Event details

Open an event's eye action or double-click it. Event Details shows the time, event, category, performed-by type, status, outcome, reason, and a flattened view of the recorded payload.

Technical correlation identifiers such as site, post, AllBox, policy, call, provider-call, and correlation IDs are intentionally hidden from this payload view. Other payload details are shown up to 100 rows and eight nested levels. Unsupported payload formats produce a generic notice rather than raw content.

The detail grid is an interpretation of the event payload, not a complete raw-log export. Preserve the parent call, visible event fields, site, AllBox, and timestamps when support needs deeper correlation.

Warning

An Access Granted outcome does not prove that a gate, reader, or relay physically operated. Correlate it with access-control or AllBox activity, controller health, device configuration, and onsite confirmation.