Docs
Sign in

Daily ops standup

Updated May 21, 20262 min read

Pull the operational summary, layer in failures and exceptions, narrate the result as a one-paragraph briefing the ops lead can read in 30 seconds.

This is the canonical "how is the business doing?" prompt. The agent answers it from three angles — overall, failures, exceptions — and stitches the result into a tight summary.

When to use

  • Slack-bot morning post.
  • Copilot opening response when the ops lead types "good morning".
  • Internal status page that an executive reads on the way to a meeting.

Tool sequence

report_dashboard_summary           (overall snapshot — 30-day fixed window)
       │
       ▼
report_failed_delivery_analysis    (recent failure trends — last 7 days, daily buckets)
       │
       ▼
report_exception_history           (broader errors — last 7 days, daily buckets)
       │
       ▼
narrative summary                  (a single paragraph)

Example agent prompt

"Daily ops standup — what's the state of shipments today?"

"Give me a 30-second briefing on shipment ops."

Walkthrough

Step 1 — overall snapshot

report_dashboard_summary(tenantId="…")

The dashboard summary covers a fixed 30-day window — status counts, shipment trends, transit-time summary. Reads as the "big picture" in the briefing.

Step 2 — recent failures

report_failed_delivery_analysis(
  tenantId="…",
  from="<7 days ago>",
  to="<today>",
  bucket_size="day",
  failed_delivery_group_by="reason"
)

7-day window, daily buckets. The agent picks out the top reason and the trend direction (rising / falling / flat).

Step 3 — broader errors

report_exception_history(
  tenantId="…",
  from="<7 days ago>",
  to="<today>",
  bucket_size="day",
  group_by="error_code"
)

Same window. The agent surfaces the top error code and how it compares to the prior week if the data permits. Per the input schema, valid group_by values are carrier, error_code, delivery_type, and shipment_type.

Step 4 — narrate

The agent collapses the three responses into a paragraph:

"Volumes look steady — 2,450 shipments confirmed in the last 30 days, transit-time mean unchanged at 2.3 days. In the last 7 days, failed deliveries are up 8% (mostly customer unavailable, KSA), and we've had a small rise in label-generation errors against Aramex. Worth a quick look at the Aramex integration before volumes pick up later this week."

The shape: lead with overall, follow with failures, end with a single recommended next step.

Variations

  • Per-merchant briefing — pass merchant=<id> on each tool to scope the report to one brand. Useful for multi-brand tenants where each brand owner wants their own daily.
  • Per-region briefing — same shape, with country_access set.
  • Weekly instead of daily — extend the windows on steps 2 and 3 to 30 days, change bucket_size to week. Same narrative shape.
  • Copilot greeting — same three calls, but the agent only surfaces the briefing if there's a change worth surfacing. If everything is flat, the agent says "nothing unusual in ops this morning" instead of restating numbers.

Pitfalls

  • Don't restate numbers without interpretation. A briefing that reads "volume is 2,450, transit time is 2.3 days, failures are 3.1%" is just a dashboard with extra steps. The value is in the trend interpretation.
  • Watch context size. All three reports return rich payloads. Use select aggressively — for the briefing, the agent only needs totals and trend direction, not full breakdowns.
  • Time zones. Use tzoffset on each call so the "last 7 days" window matches the operator's day, not UTC.