This feature is currently in Beta and available to a limited group of merchants only.
Overview
Exception shipments often require teams to manually investigate between tracking events, carrier updates, internal notes, and customer communications. This takes longer to understand what went wrong and what action to take next.
AfterShip Tracking's built-in Exception Agent simplifies this process by automatically investigating shipment exceptions and preparing recommended actions in one place. Instead of manually reviewing every affected shipment, operations teams can review a ready-to-action exception task with supporting context, reasoning, and drafted communications.
Read the article to know how the Exception Agent works and understand the end-to-end exception handling workflow.
During Beta, Exception Agent operates entirely in Copilot mode, meaning every action still requires manual review and approval before execution.
What Exception Agent does
Exception Agent automatically detects shipment exceptions, including situations that carriers may not clearly report yet. It investigates the issue using shipment events, carrier patterns, delivery risks, and historical shipment behaviour to recommend operational actions and next steps.
Recommended actions may include:
Customer notifications
Carrier escalations
Internal operational alerts
Supported exception scenarios may include:
Multiple shipments are stalled in transit, with no movement for several days
Failed delivery attempts caused by address issues
Shipments at risk of returning to sender
Carrier or facility-level disruptions affecting multiple shipments
Step-by-step: Exception Agent workflow
The following workflow explains how to review and resolve shipment issues using Exception Agent.
Step 1: Review exception tasks
On the Tracking Home page, the Exception shipment tasks card displays shipment issues (pending tasks) detected by the agent.
Each task includes:
Exception type
Impacted shipment count
Priority level
Example tasks-
Stalled shipment cluster (8 Shipments affected)
Address verification failed (3 Shipments affected)
Customs document request (5 Shipments affected)
This helps teams quickly identify operational risks without manually reviewing individual shipments.
Step 2: Open the exception task queue
Click Check all from the Home page or navigate to the Exception task from the sidebar to open the task queue. The queue acts as the central workspace for reviewing and managing active shipment exceptions.
Operational metrics at the top of the page summarizes the current workload and performance. This may include:
Impacted shipments
Resolved shipments
Resolution rate
Average resolution time
Below the operational metrics, each task also displays:
Task name
Impacted shipment count
Recommended actions
Status
Priority
Timestamps
Step 3: Review the task summary
Go to the Home page or task queue > Open any task to review the task summary.
Each exception task starts with a summary explaining the shipment issue, its impact, the grouping logic, and urgency. This reduces the need to investigate tracking events across different systems manually.
Step 4: Review the agent’s reasoning
Select View reasoning to open the Agent reasoning panel. The Agent’s reasoning panel explains shipment signals and operational context behind the recommendations.
Depending on the issue type, reasoning may include shipment anomalies, carrier performance patterns, delivery risk indicators, or related historical cases.
The reasoning panel is typically organized into:
Task details
Agent analysis
Supporting signals and evidence
Ask follow-up questions directly within the task chat for additional clarification.
Step 5: Review grouped shipments
For tasks involving multiple shipments, the Grouped exception shipments section displays all impacted shipments together, including their tracking numbers, statuses, carrier information, and last known locations.
You can also:
Review individual shipments in detail
Remove shipments from the task
Split shipments into separate tasks
Mark shipments as resolved
Ignore shipments that no longer require action
This ensures that the task is aligned with the latest operational status before action is taken.
Step 6: Review recommended actions
Each task includes one or more recommended actions generated by the agent. Depending on the scenarios, recommendations may include customer notifications, carrier escalations, or internal operational alerts.
Moreover, each recommendation can be reviewed before execution, allowing you to take the following actions:
Edit the drafted content
Preview the final communication
Proceed with execution
You can also expand any recommendation to review the drafted message before taking action and create additional actions manually within the task if more follow-up is needed.
For example, you may send another carrier escalation or create a new external communication.
Step 7: Edit and preview communication
Select Edit to customize drafted communications before sending. Editable fields include subject line, message body, tone, and more. The changes apply only to the current task and do not affect future drafts.
Select Preview to review the final communication exactly as the recipient will receive it, including personalization and branded styling.
Separate templates can be configured for:
Customer notifications
Carrier escalations
Internal operational alerts
Step 8: Run actions
Once the recommended action has been reviewed, click Run to execute it.
Exception Agent displays the execution progress in real time and shows a Completed status once the action finishes successfully.
Step 9: Review the action log
Each task includes an Action log tab that records all executed activity, including Executed actions (what was sent), Execution timestamps (when it was executed), contacted recipients, and users who approved the action.
Teams can also manually log actions to maintain a complete operational record.
Step 10: Resolve or ignore tasks
Once the shipment issue has been handled, mark the task as Resolved.
Tasks that no longer require action can also be marked as Ignored to remove them from the active workflow.
Step 11: Chat with the Exception Agent
You can interact directly with the Exception Agent within any task to:
Ask follow-up questions,
Request clarification,
Rewrite drafted communications, or
Explore alternative actions.
This ensures that the team can continue investigations without leaving the task workflow.
Create shipment dashboard views for exception handling
You can also customize the Shipment dashboard views, focused on exception handling workflow using exception-related filters and columns. This helps better monitor shipment issues.
For example,
Filter shipments by exception resolution status
View shipment data and exception information together
Monitor shipments linked to active or resolved exception tasks
Related exception tasks can also be opened directly from the shipment dashboard for additional investigation and action.
Custom exception types
In addition to carrier-reported exceptions, you can create custom exception types so the Exception Agent detects shipment issues based on your own operational rules, even when the carrier hasn't reported an issue.
Create a custom exception type
Go to Settings > Exception handling to manage custom exception types.
Each custom exception type includes:
A unique name
An optional description
Detection conditions
An on/off toggle
Detection conditions are created using shipment fields. You can combine multiple conditions in a single exception type. A shipment is flagged only when all configured conditions are met at the same time.
Commonly used fields include:
Shipment status (for example, In transit or Out for delivery)
No tracking updates for a specific number of days
Transit time exceeding a specific number of days
Carrier, carrier service, or shipping method
Many other shipment fields are also available when defining conditions.
For example, the preset Stuck in transit exception type combines these conditions:
Shipment status is In transit or Out for delivery, and
No tracking updates in the last 5 days
A shipment is flagged only when both conditions are true.
Example custom exception types include:
No movement – In-transit shipments with no tracking updates for 7 days
Slow standard shipping – Standard Shipping shipments that remain in transit for more than 10 days
Carrier stall – UPS Ground shipments with no tracking updates for 5 days
How detection works
Detection runs automatically every hour.
When a shipment matches a custom exception type, the Exception Agent automatically creates a new exception task or adds the shipment to an existing matching task before starting its analysis.
Flagging a shipment with a custom exception does not change its tracking status.
A default Shipment exception type is available for all carrier-reported exception sub-statuses. It is enabled by default and cannot be deleted, but you can modify the sub-statuses it includes.
Additional preset exception types, including Pre-transit delay, Stuck in transit, Delayed, and Potentially lost, are available in Settings and can be enabled or adjusted as needed.
How custom exceptions are resolved
A shipment is automatically resolved when it is delivered or no longer matches the exception type's detection conditions.
When all shipments within a task are resolved, the task closes automatically.
A shipment can belong to only one active task at a time. If the carrier later reports a native shipment exception for a shipment already assigned to a custom exception task, the carrier-reported exception takes priority and the shipment moves to the corresponding task.
Task grouping rules
Task grouping rules determine how the Exception Agent groups exception shipments into tasks so they match your team's operational workflow.
Configure the grouping rule
Go to Settings > Exception handling to configure task grouping.
Select up to 3 grouping dimensions. All exception shipments are grouped into tasks by the selected dimensions.
Available grouping dimensions include:
Exception type (default or custom exception criteria)
Carrier
Carrier service
Tracking event
Shipments with the same values across the selected dimensions are grouped into the same task.
For example, grouping by Exception type and Carrier creates one task for each carrier under every exception type.
At least one grouping dimension must remain selected. If you don't modify the configuration, the default grouping rule is used.
What happens after changing the grouping rule
Changes apply only to newly detected shipment exceptions.
Existing tasks keep their current grouping and are not regrouped.
Each shipment can belong to only one active task at a time, regardless of the grouping configuration.
If a task contains shipments from multiple carriers, the Exception Agent automatically adjusts its recommendations. For example, it does not recommend a carrier escalation email when multiple carriers are involved because there is no single carrier recipient.
Demo video
FAQs
1. Will the Exception Agent take actions automatically?
No. The Exception Agent currently operates entirely in Copilot mode. Every action requires manual review and approval before execution.
2. How is customer data handled?
Customer data is managed under standard AfterShip security and privacy policies. Personally identifiable information (PII) is not used to train AI models.
3. Can communication templates be customized?
Yes. Branded templates, sender email address, and default communication settings can be configured. You can also edit individual drafts before execution.
4. Can I undo actions after execution?
No. Executed emails, escalations, or alerts cannot be reversed. It is suggested to review recommendations carefully before running actions.
5. How is Exception Agent different from shipment alerts?
The existing shipment alerts notify your team when a shipment issue occurs. However, Exception Agents go beyond by investigating the issue, explaining the reasoning, recommending next steps, and preparing operational actions for review.
6. What if the recommendation is incorrect?
You can reject the recommendations, edit the draft content, or provide feedback directly within the task workflow. These feedback signals help improve recommendation quality over time.
7. Will changing the task grouping rule regroup existing tasks?
No. Changes to the grouping rule apply only to newly detected shipment exceptions. Existing tasks keep their current grouping until they are resolved or ignored.