ETA and milestone exceptions
Extract changed dates, reasons, affected bookings, and the next owner from carrier communication.
Carrier exception to reviewed system update
AI logistics agents that prepare the next action, then wait for your team
Turn a carrier email, attachment, or message into a structured exception. The agent proposes the TMS change, shows its source, and asks a human to confirm or edit it before anything is applied.
- Human approval
- Source trace
- Synthetic public demo
One observable workflow
From an unstructured exception to an auditable update
The example follows booking LW-2844 after a carrier moves the ETA and reports a customs inspection.
- 01
Read the operational signal
The agent identifies the booking, old and new ETA, exception type, and source message.
- 02
Prepare a proposed change
The change is structured for the target workflow instead of being left as a summary in another inbox.
- 03
Ask a human to confirm or edit
An operator sees what will change and can approve, correct, or stop the action.
- 04
Apply and retain the source
The approved update keeps a trace back to the operational signal and reviewer decision.
What the operator sees
Evidence before automation
The public walkthrough is intentionally narrow. It demonstrates the review boundary and the information path without pretending to be a connected customer system.
Control plane
Choose what the agent may read, propose, and apply
The implementation starts with one bounded workflow. We map the source, decision rule, target system, reviewer, failure state, and audit evidence before increasing autonomy.
Read the logistics intelligence case- Read scopeOnly connected mailboxes, queues, documents, and records.
- Action scopeDraft only, approval required, or automatic for an explicit low-risk rule.
- Failure stateUncertain or conflicting signals move to a named human queue.
- Audit trailSource, proposed values, reviewer decision, and applied state stay linked.
Where to start
Good first workflows for a logistics AI agent
Document intake
Classify incoming documents, extract agreed fields, and route missing or conflicting data for review.
Operations handoffs
Turn scattered updates into a structured task with the source, decision context, and responsible team.
Not a good first workflow: a broad autonomous agent with access to every mailbox and write permission across every system.
Guided product demo
Try the carrier exception workflow
Walk through extraction, human confirmation, and the resulting portal state. The demo uses synthetic booking data and does not connect to Lerine or any customer environment.
- Three interactive steps
- Visible confirm and edit boundary
- Clicks and active time tracked only after explicit demo-analytics consent
Access form
Open the guided demo
Questions
What the demo proves and what it does not
Does the AI agent update the TMS without approval?
Not by default. The workflow can require a named operator to confirm or edit the proposed change before it is applied.
What can the agent read?
A scoped implementation can process operational email, messages, documents, and system events that the team explicitly connects.
Is the guided demo connected to a live logistics system?
No. The public guided demo uses synthetic shipment data and does not connect to a customer or production logistics environment.
Have a workflow worth testing?
Bring one operational signal, one target update, and one human decision point.