A two-direction system, not a one-way export
Ownership follows utilization
The plan does not assume that owning every asset is automatically efficient. Trucking is outsourced while shipment frequency is low. The trigger for direct ownership is a stable rhythm approaching a loaded vessel at the Puerto Rico port each day, at which point repeated carrier overhead can exceed the fixed cost of trucks and employed drivers.
The same test governs ships. Contracted carriage reduces initial capital exposure; owned small vessels become defensible only when Pluoton’s projected bulk flows make third-party rates structurally expensive. Inventory storage is minimized during the tenuous phase because idle stock and warehouse dwell consume capital without moving goods.
| Function | Hermes role in the research plan |
|---|---|
| External execution | Carry out trade and transport arranged with entities outside the inter-company system |
| Transaction design | Executed from plans negotiated by Phronesis; Hermes is not the negotiating desk |
| Ground and sea movement | Outsource first, own when measured frequency and cost justify integration |
| Inventory | Administer inventory and storage, while avoiding unnecessary early warehousing |
| Return cargo | Import merchant demand from Savannah to improve round-trip utilization |
The decisive measurements
The research asks for a lane-level model rather than a generic freight margin: tonnes and cubic metres by product, container compatibility, port dwell, loading time, sailing frequency, bunker and crew costs, trucking radius, demurrage, insurance, FTZ treatment, rail interchange, backhaul fill, and working-capital days. Owned transport is chosen only where the all-in cost per delivered unit beats contracted carriage at realistic utilization.
- Validate recurring outbound volume and handling classification.
- Secure indicative carrier, port, terminal, insurance, and inland quotes.
- Match inbound merchant purchase orders to return capacity.
- Compare charter, contract, and ownership on identical lane assumptions.
- Add vehicles, yards, or storage only after throughput makes them productive.
Puerto Rico pays for distance twice when freight leaves full and returns empty.
Hermes begins with a structural problem rather than a vehicle. An island industrial system cannot treat transport as a final invoice added after production. Vessel frequency determines inventory; port dwell determines working capital; container and bulk-handling choices reach backward into product packaging; unreliable return capacity raises the price of every imported input. For Pluoton, a profitable chemical or mineral process can become uneconomic when its output waits too long, moves in the wrong equipment, or pays a carrier whose network was designed for another cargo pattern.
The Savannah corridor is the initial answer because it joins several functions at one mainland node. Savannah offers a major port, rail access, a Foreign-Trade Zone, distribution infrastructure, and a large purchasing hinterland. Outbound material from Puerto Rico can terminate there, transfer to rail or buyer custody, or enter further processing. The vessel’s return leg can carry goods already ordered by Puerto Rican merchants. Hermes therefore treats import and export as one round-trip calculation. Return freight is not a secondary retail adventure; it spreads the voyage cost across two paying directions.
The asset is not the ship. The asset is a repeatable, loaded circuit with controlled dwell and known demand at both ends.
Why the plan begins small
Two small used cargo ships appear in the research because early volume is uncertain and operating frequency matters more than nominal fleet size. A smaller vessel that turns predictably can be more useful than a larger vessel that waits for cargo. Yet ownership is not sacred. Contracted shipping protects scarce initial capital while throughput is unproven. The decision changes when recurring Pluoton output gives an outside carrier pricing power over a lane the group cannot avoid.
Trucking follows the same logic. At low frequency, outsourcing converts idle trucks, maintenance, dispatch, and payroll into a variable cost. When loading and unloading produce a near-daily port rhythm, those margins repeat often enough that direct employment and owned vehicles may cost less. Integrate the part of the chain whose repeated external overhead exceeds the cost and management burden of ownership.
Inventory is time made visible
Material sitting in a yard has already absorbed excavation, processing, packaging, and inland movement but has not completed a sale. Imported goods sitting without merchant orders consume the same capital in reverse. Hermes must join production schedules, purchase orders, vessel stowage, customs documentation, trucking appointments, and customer acceptance into one flow plan. Storage is used where it buffers a real mismatch, not where it hides a planning failure.
A logistics company inside an inter-company system
The research separates negotiation from execution. Phronesis evaluates and negotiates outside transactions; Hermes carries them out, administers inventory, and owns or contracts the physical movement. That division keeps financial strategy from disappearing into day-to-day dispatch while keeping the negotiator accountable to real handling costs. It also makes lane economics auditable: agreed commercial terms can be compared with what the operation actually spent and how long the cargo actually took.
The mature Hermes imagined by the plan is the group’s memory of physical flow: which product tolerates delay, which port process creates damage, where containers accumulate, when ownership beats charter, how much backhaul demand is genuine, and which inventory belongs at sea rather than in a building. Its value comes from lowering the friction imposed on every other physical project.
A researched logistics company, not yet a carrier
Hermes currently exists as corporate and operational design work. Vessel acquisition, Merchant Marine participation, operating authority, customs processes, insurance, contracts, and service levels remain execution gates. The technical content above describes the actual research model without representing those gates as completed.