The first wedge into agentic logistics.
Turn confirmed, post-booking shipments into one agent-readable state graph — then use it to handle exceptions, documents, and the next action.
The data is everywhere. The state is nowhere.
ETA changed, B/L correction requested, discharge delayed — scattered across portals, threads and a stale spreadsheet, with no shipment state to reason over. That gap isn’t cosmetic; it’s expensive.
Before the agent — one question, a day of chasing.
With the state scattered, answering a single customer question means hopping portals, threads and phone calls — and mostly waiting. The very same question the agent will answer in seconds.
For Booking KRPU-240817, what should I tell the customer right now?
Ask the agent. Get a sourced answer.
Not a chatbot — the agent reads the normalized state graph and answers with sources, every line traceable to an event. The console is the output surface; the graph underneath is the product.
For Booking KRPU-240817, what should I tell the customer right now?
Why it matters The point isn’t a plausible-sounding answer — it’s a sourced judgment.
One confirmed shipment, one connected ontology.
Not a list — a graph. Scattered inputs become typed entities joined by typed relationships, and the agent reads that connected state to reason.
One event, several downstream obligations.
Watchthe event ripple along the connections — Journey & Document change, then Risk and Action nodes re-derive on the same graph.
Agent recommends. You approve. System records.
Exception → response → tasks → human approval → record. This is the concrete workflow the first PoC ships — deliberately human-in-the-loop.
- B/L teamVerify the consignee address correction
- Ops teamRe-confirm Container B gate-in status at 15:00
- CS teamApprove sending the ETA-change notice to the customer
Why it matters Stressing fully autonomous execution makes carriers uneasy. The structure is: agent recommends, human approves, system records.
The same graph, read by every party.
Future preview — built on the same shipment graph. Once one party can read the state, the next step is many parties reading it. This is the network effect, not the first PoC. Future · same graph, later
Why it matters Today a person reconciles state across portals and email. Tomorrow, carrier, forwarder, and port agents read the same shipment graph and coordinate their actions.
From one wedge to a network.
Phase 1 is the strategic entrance — the ontology slice + post-booking workflow you just saw. Every later phase adds parties and use cases to the same ontology; nothing is a separate module.
- 1Phase 1 · Now — the wedgeShipment Intelligence Agent
- post-booking ontology slice
- document readiness
- exception → next action
- human-approved CS response
- 2Phase 2Booking & Documentation Agent
- booking request validation
- SI/B/L automation
- eBL workflow support
- customer-specific SOP automation
- 3Phase 3Carrier-Forwarder Agent Network
- forwarder-facing white-label agent
- multi-party shipment collaboration
- proactive customer notification
- TMS/CRM integration
- 4Phase 4Port-Carrier Agentic Coordination
- port call event integration
- terminal event integration
- ETA/ETD impact propagation
- D&D / resource / priority handling
- 5Phase 5 · North StarAgentic Logistics Operating Network
- carrier agent
- forwarder agent
- port agent
- terminal agent
- shipper agent
Why it matters A single-scene demo extends into a single-line roadmap — from one wedge to a network where every party is expressed as an agent.
From shipment tracking to shipment reasoning.
An agent that doesn’t just show where your cargo is — it decides what to do next.
Post-booking shipments become one agent-readable state graph.
API · EDI · email · documents · portals → a DCSA-aligned shipment ontology.