Skip to main content

Insights Dashboard

Overview

The Dashboard is where you triage findings produced by Aurora Insights' nightly analysis. Findings are grouped by priority — Critical, Important, Info — so you can clear the most urgent items first, filter by owner or rule, and mark each finding as reviewed, dismissed, or actioned as your team works through it.

How to Access

  • Click Insights in the left sidebar.
  • The Dashboard tab is selected by default.
  • Available to all team members. Non-admin users see only findings on conversations they own or that belong to teams they manage.

Interface Overview

The Dashboard has three areas, top to bottom:

  1. Quick-pill strip — coloured pills showing how many New findings exist at each priority level (Critical, Important, Info), plus a total count. Click any pill to filter the list down to that priority. Click the same pill again to clear the filter.
  2. Filter bar — five filters across the top:
    • Owner — narrow to findings on conversations belonging to a specific team member (or "All").
    • Priority — multi-select dropdown (Critical / Important / Info).
    • Status — multi-select dropdown (New / Reviewed / Dismissed / Action Taken / Auto-Resolved).
    • Rule — multi-select dropdown of every built-in and custom rule.
    • Date range — calendar picker; defaults to the last 29 days.
  3. Priority groups — three collapsible sections (Critical, Important, Info), each listing the findings that match your filters. Empty groups are hidden.

Features & Actions

Filter Findings

What it does: Narrows the list to a specific subset of findings.

Steps:

  1. Pick any combination of Owner, Priority, Status, Rule, or Date range from the filter bar.
  2. The lists refresh automatically.
  3. Click a quick-pill to filter by a single priority; click it again to remove the filter.

Important notes:

  • Filters combine with AND — for example, selecting Priority "Critical" and Status "New" shows only New Critical findings.
  • The Owner filter only appears when your account has team users to choose from.
  • Non-admins always see results restricted to their own conversations and team-managed conversations; the Owner filter cannot widen this.

Open the Underlying Conversation

What it does: Lets you read the original chat thread that triggered the finding.

Steps:

  1. On a finding row, click the eye icon to preview the conversation in a side panel without leaving the page, or click the open-in-new icon to open the conversation in a new tab.

Triage a Finding

What it does: Updates the finding's lifecycle status so your team knows it has been handled.

Steps:

  1. On a New finding row, click the three-dot menu on the right.
  2. Choose one of:
    • Mark Reviewed — someone read it and acknowledged it.
    • Dismiss — not actionable, or a false positive.
    • Action Taken — the team has resolved or addressed the issue.

Important notes:

  • Once a finding's status changes from New, the three-dot menu disappears. Status changes are intentionally one-way — you cannot move a finding back to New.
  • Findings in Auto-Resolved state never had the menu — the underlying condition was fixed before anyone needed to act.

Read Why a Finding Was Flagged

What it does: Shows the rule, summary, and recommended next step for each finding.

  • The row header shows the conversation name, the rule name that triggered it, and the current status (if not New).
  • The summary line below explains what the AI noticed. Hover for the full text in a tooltip.
  • For findings produced by the AI, suggested actions are written in the same language as the conversation.

Paginate Within a Group

What it does: Lets you walk through groups that have more than 5 findings.

Steps:

  1. Each priority group shows up to 5 findings per page.
  2. Use the chevron buttons at the bottom of the group to move between pages.
  3. The group header shows the total finding count for the current filters.

Collapse a Priority Group

What it does: Hides a group's findings to focus on a different priority.

Steps:

  1. Click the priority header (the row with the colored dot, label, and count).
  2. The chevron rotates and the findings collapse. Click again to reopen.

Fields Reference

ColumnDescription
ConversationThe room name where the finding originated
RuleThe detection rule that produced the finding (built-in or custom)
StatusCurrent lifecycle state (New / Reviewed / Dismissed / Action Taken / Auto-Resolved)
SummaryShort description of what was detected (full text on hover)
DateThe date this finding was produced
ActionsQuick preview, open-in-new, and the triage menu (only for New findings)

Business Rules

  • Non-admin users only see findings on conversations they participate in or on conversations that belong to teams they manage. Admins see everything.
  • A single conversation can produce multiple findings if it triggers more than one rule (for example, a stale conversation that also contains a broken promise).
  • Stale conversations and missed handoffs auto-resolve when the underlying problem is fixed (for example, an agent replies, or someone is assigned). Auto-resolution happens at the start of the next nightly run — not in real time.
  • Auto-resolved findings appear with that status badge but cannot be triaged manually — there is nothing left to do.
  • Findings older than 90 days are automatically removed once they have reached a resolved state (Reviewed, Dismissed, Action Taken, or Auto-Resolved). New findings are never purged.
  • The same conversation can be re-flagged on a later night if a new problem appears, or if a previously resolved condition returns.

FAQ

Q: Why doesn't a finding I dismissed reappear the next night? A: Dismissing a finding records that the underlying issue isn't actionable or is a false positive. The next night the system sees that the same rule already produced a finding for that conversation and won't duplicate it.

Q: Why is the same conversation flagged for two different rules? A: Each rule operates independently. A conversation that's been silent for 24+ hours and contains a broken promise can fairly produce both a "Stale Conversation" and a "Broken Promises" finding — they describe different things.

Q: Why didn't I get a notification when a Critical finding appeared? A: Insights does not push real-time notifications today. Findings are produced as a batch overnight and visible the next morning. Treat the page as a once-a-day triage tool.

Q: I can't see findings for conversations I think I should have access to. Why? A: Access is determined at the moment the finding is created. If you were added to a team after the analysis ran, you will see findings produced from the next run onward. Access changes are not back-applied to historical findings.

Q: A finding's summary is in Spanish but my UI is in English. Why? A: AI-generated findings are written in the same language as the underlying conversation, so the agent reading the dashboard sees a summary in the language the customer uses. Built-in rule names and standard statuses are translated to your UI language.