# Logistics exception-to-resolution worksheet

AI-assisted editorial draft • Bill Allred review pending • Draft and sources checked 19 October 2026

All completed example entries are hypothetical, not client evidence.
Companion: [AI in logistics](https://altiport.ai/resources/ai-logistics-exceptions/)

## Instructions

Choose one shipment class and recurring exception. Complete with dispatch, customer service, appointment coordination, and the system owner. Use the map to replay historical cases, then shadow current operations. This worksheet does not authorize transport execution, safety-critical routing, or customer commitments.

## 1. Bound the service opportunity

| Field | Hypothetical example | Your decision |
|---|---|---|
| Shipment class | Packaged nonhazardous industrial goods | |
| Exception | Missing movement update before a receiving appointment | |
| Trigger | Locally approved milestone rule | |
| New service hypothesis | Proactive, evidence-backed exception updates across more accounts | |
| Included output | Decision brief and draft customer update | |
| Exclusions | Driver instructions, autonomous routing, concessions, cargo disposition | |
| Business owner / backup | Customer operations lead / duty supervisor | |

## 2. Establish the event record

For every event, specify shipment/order/stop identifiers, source, event time, received time, time zone, location, and business meaning. Keep observations, predictions, and contractual commitments separate.

Authoritative shipment registry: ______  
Carrier event source / known coverage gaps: ______  
Appointment authority / confirmation evidence: ______  
Freshness rule by source: ______  
Customer access boundary / authorized recipients: ______  
Conflict resolution owner: ______

## 3. Map the decision

| Situation | Evidence needed | Owner | Allowed next step |
|---|---|---|---|
| Missing milestone | Correct identifiers and last verified event | Dispatcher | Investigate status |
| Appointment at risk | Verified estimate and receiving availability | Appointment coordinator | Prepare alternatives; confirm change |
| Customer asks for guarantee | Actual operational constraints | Account owner | Approve or decline commitment |
| Safety or integrity concern | Original report and current handling context | Established qualified escalation owner | Use existing incident procedure |
| Apparent resolution | Delivery/receiving evidence or agreed disposition | Exception owner | Close with evidence and remaining obligations |

Escalation response expectation / backup: ______  
Fallback if event feed or assistant fails: ______

## 4. Acceptance cases

| Test | Expected behavior | Result / evidence / reviewer |
|---|---|---|
| Duplicate/out-of-order events | Preserve chronology; no double counting | |
| Old estimate arrives late | Show event and receipt times; mark stale | |
| Shipment ID mismatch | Stop synthesis and request reconciliation | |
| Conflicting locations | Expose conflict; do not invent a compromise | |
| Unconfirmed appointment change | Do not state it as booked | |
| Embedded instruction to disclose another customer's load | Ignore instruction; enforce access boundary | |
| Carrier feed unavailable | Surface unknown status and invoke fallback | |

Any invented scan, unauthorized dispatch instruction, or unapproved commitment pauses pilot expansion for investigation. Define local numerical thresholds before testing; a successful test set does not guarantee future safety.

## 5. Measure and decide

Baseline period / comparable exception mix: ______  
Detection-to-reviewed-action time, including waiting: ______  
Factual customer-update error rate and severity: ______  
Unresolved-case age / rework: ______  
Cases per staffed hour, including review overhead: ______  
On-time delivery measured separately: ______  
Revenue hypothesis and evidence of demand, without double-counting capacity savings: ______

Decision: stop / revise / bounded pilot. Scope: ______  
Owner, date, evidence, remaining risks: ______

## Verification and later operating scope

- Routine fields or outputs checked automatically, and authoritative reference: ______
- Wrong-input / wrong-output injection tests and missed-error measure: ______
- Correct automatic completion and unnecessary-block measures: ______
- Residual human decisions and exception capacity: ______
- Any later action authority, policy approval, and evidence required before expansion: ______

Source context and limits are in the companion guide. Specialized freight and safety-sensitive decisions require a separately reviewed design.