Cross-border freight,
orchestrated.

FedEx, UPS, DHL, and USPS are wired at the client layer (src/integrations/carriers/carrier_client.py) with a unified tracking + rate + booking interface. They run in stub mode by default — returning realistic sample data — and flip to live API calls when the corresponding env vars are present (FEDEX_API_KEY, UPS_CLIENT_ID + UPS_CLIENT_SECRET, DHL_API_KEY, USPS_API_KEY). Ocean and air-freight forwarders (Maersk, Hapag-Lloyd, OOCL, Lufthansa Cargo) are on the dashboard as planned connectors — they are NOT live yet. We do not pretend otherwise.

Both, depending on availability. The primary path is Methodology #235 (Temporal Reasoning) at src/empires/logistics/logistics_engine.py:31-89 — it consumes shipment state, exceptions, and historical transit patterns to produce a predicted arrival with confidence and factor breakdown. When the methodology path is unavailable, the heuristic fallback at lines 66-89 quantifies delay as 24h per high/critical exception + 8h per medium, returns a transparent risk tier, and surfaces every exception that went into the calculation. No hidden model, no mystery number.

Methodology #237 (Spatial Reasoning) at logistics_engine.py:93-180 runs the primary optimization using your origin, destination, mode, available rates, and constraints. When it is offline, the heuristic fallback sorts your rate quotes: speed mode by transit days ascending, cost mode by rate ascending, balanced mode by a 50/50 blend of rate and transit time. The cheapest-vs-worst delta becomes your explicit savings number — not a marketing stat.

Gain-share is sales-led, not self-serve. The process: we sign an SOW, we measure your baseline freight spend over a 30–90 day pilot, we run the optimization and exception-prediction layers against your live shipments, and we conduct a quarterly savings audit against the measured baseline. You pay 10–15% of the verified savings — not projected, not estimated, verified. If the audit shows no savings, you pay zero. No trap. Target launch pricing and terms, finalized per deal.

Because we have not finished the logistics-tech competitor scan. Flexport, Project44, FourKites, e2open, and Descartes each price very differently — from per-shipment fees to seat licenses to percentage-of-spend — and publishing a confident $/mo before that scan would be vibes-based. We would rather tell you the self-serve tier is placeholder and lock the published number on the day it is set than quote you a figure we cannot defend on a sales call. Waitlist signups get the locked launch rate the day it is published. Gain-share is different — the 10–15% band is the stated business model, not a guess at what the market pays.

src/empires/logistics/logistics_knowledge.py targets 500+ structured entries across 10 domains (Incoterms, ports, carriers, HS codes, sanctioned entities, trade lanes, formulas, customs docs, carrier reliability, freight surcharges). The 11 Incoterms 2020 are fully wired today. The remaining domains are in active build — we do not claim completeness until each domain has its own test coverage.

Flexport is a licensed freight forwarder — they book your freight. Project44 is a visibility platform — they show you where your freight is. LiquidCortex Logistics is a cross-border 3PL orchestrator: it sits on top of your existing carriers and forwarders, predicts the delays, optimizes the routes, generates the customs docs, and runs the employee squad (Rowan + Maestro + Morgan + Scout + Finn) end to end. We do not replace your forwarder. We make the forwarder you already pay more legible, more predictive, and cheaper to operate against.