Skip to content

The Control Tower home page

View as Markdown

“Control Tower”, the first entry of the sidebar, is the app’s home page. It answers one question: what needs you today. Analysis and history live in “Insights”; this page keeps to what to act on now.

From top to bottom it shows a header, “Today’s brief”, “Priorities”, today’s operations, a recovered-value line when there is one, the key metrics and, when any are open, “Alerts Requiring Attention”.

The header

Under the title, two chips say how fresh the page is:

  • The playbooks chip says when the findings and metrics were last computed. They are computed nightly, not live. When the last run is more than about a day old, the chip turns amber and says the playbooks are stale: treat what you see as old.
  • “Orders synced” means your orders loaded for the operations panels. “Orders sync failed” means they did not, and a banner above those panels offers “Try again”.

“Refresh” reloads the findings and metrics. It does not run the analysis again: it fetches the latest results of the nightly run.

Today’s brief

“Today’s brief” is a short written summary of your operation, built by AI from your other findings during the nightly run. It carries an “AI-generated” marker, the time it was generated and, when it has one, a “Suggested action:”.

“Ask the Analyst about this” opens the AI analyst with the brief already in hand, so you can ask follow-up questions. See Insights and the AI analyst.

With no summary from the last 24 hours, the card says so and explains when the next one is built.

Priorities

“Priorities” is your worklist of findings: the critical and warning findings from the latest run of each playbook, ranked by money at stake. A finding with a money figure comes before one without; among the rest, critical comes before warning. A finding you snoozed or resolved stays off the list.

Each row shows the finding’s title and the first lines of its suggested action, its severity, the playbook that raised it, how long it has been open, its current status and, when it can be priced, what is at stake. Each row offers:

  • “Investigate” opens the AI analyst with this finding in hand.
  • “View evidence”, or “Details” when the finding carries no structured evidence, opens the finding drawer, described below.
  • Snooze, for seven days, takes the finding off the list for a week, for the whole account. The row shrinks to a one-line receipt with “Undo”.
  • The cross in the corner (“Dismiss finding”) hides the finding.

A counter above the list, such as 2 of 5 handled, fills as you work through it: a row counts as handled once you snooze it, or once someone has marked it “Acknowledged” or “In progress”. When every row is handled, the list says “You’re clear for today.” The footer gives the week’s flow: how many findings opened, how many were resolved and how many are still open.

With nothing critical or warning to show, the block reads “All clear: no critical or warning findings.”

The finding drawer

The drawer shows the whole finding. Structured evidence comes first, as tables and charts; a finding without any shows the metrics recorded when it was computed. The raw evidence sits behind “Raw evidence JSON”, for checking.

Below the title, the issue panel tracks the problem across nightly runs: how long it has been open, how many times it has been reported and when it was first seen. Here you can:

  • set its “Status”: “Open”, “Acknowledged”, “In progress” or “Resolved”. A resolved issue is hidden from the attention surfaces until it recurs.
  • give it an “Assignee”, or leave it “Unassigned”.
  • “Snooze” it for one, seven or thirty days.

Status, assignee and snooze are shared by the whole account, so a teammate sees that someone has picked the issue up. The same drawer opens from “Insights”, so a finding reads and triages the same way in both places.

Today in operations

Two panels, built from your shipments and the orders imported from your stores.

“Needs intervention” lists the shipments that have been in transit the longest, oldest first, once they pass a set number of days since they were created. Each shows its order, origin and destination, and how many days it has been in transit; “Track” and “Open Tracking →” open “Tracking”. The age is counted from creation, so a shipment listed here is old, not confirmed stuck. With none, the panel says “No shipments need intervention right now.” See Track a shipment.

“Shipment flow” counts your shipments by status: “Label Created”, “In Transit”, “Delivered” and “Cancelled”.

Recovered value

Once findings have been resolved in the last 30 days, a green line says how many, and how much value they recovered when the findings could be priced, with the latest one named. It does not show until there is something to report.

Key metrics

The strip shows “Refund exposure (28d)”, “Margin per order”, “Carrier on-time rate”, “DPP compliance readiness” and “Working Capital”. Each tile is in one of four states:

  • a value, with its change and a small trend line where one exists;
  • good: refund exposure with nothing at risk shows a quiet positive tile rather than a zero;
  • a setup task: margin needs cost prices on your products (“Add costs” opens “Products”), and on-time rate needs carrier events (“Connect carriers” opens your carriers);
  • empty: “No orders in this period.”, or “Awaiting the first playbook run.” on a new account.

Alerts

“Alerts Requiring Attention” shows the five most recent of your open shipment alerts, with the parcel’s reference and how long ago each was raised, and the count still unresolved. “Acknowledge” clears an alert. This is the only place on the page where alerts are acknowledged.