1. Proactive delivery-exception updates
Customers find out about a delay by chasing you — a bad first signal.
Trigger: Webhook (carrier tracking event)
How the workflow runs
- Trigger
Carrier posts an event. A Webhook trigger receives each tracking update.
- Condition
Detect exceptions. A Condition node flags delays, failed attempts, or address issues.
- Agent
Explain + offer options. An Agent node writes a clear, calm update and the choices (redeliver, reschedule).
- Output
Notify proactively. An Output node reaches the customer on WhatsApp before they notice the problem.
Channels & connectors
- Webhook
Outcome
Customers hear about problems from you first, with a fix already offered.
Why it helps
Proactive, contextual updates turn a complaint moment into a trust moment.
Build spec
1 agent — 1 updates agent; exception detection is a workflow Condition node.
System prompt (paste-ready)
You are a delivery-updates agent for {{company}}. Given a carrier exception event {tracking_id, status, reason, eta}, write a calm, clear WhatsApp update in the customer's language that explains what happened and offers the available options (redeliver, reschedule, pickup). State only the facts in the event — never invent a new delivery date or a cause the carrier didn't report. Keep it short and reassuring, with one clear next action.MCP connectors
- WhatsApp — send channel
- Carrier tracking — inbound via Webhook trigger
Built-in tools
- http_request (fetch full tracking detail)
Guardrails
- Only the carrier's reported status and ETA are communicated — no invented dates or causes
- The Condition node fires the message only on genuine exceptions (delay/failed attempt/address issue), not routine scans
- Options offered are limited to what the carrier actually supports for that shipment
Output & delivery
When the Condition node flags a real exception, the agent's calm, factual update is sent on WhatsApp — before the customer notices — with the available options as one-tap replies.
