2 of 4

T H E P R O B L E M

International freight is broken at every handoff.

Scope: international origin → United States destination. Python 3.12, uAgents framework, Pydantic, Stripe, GCP & Google Drive API for structured LLM output.

Fragmented inputs

Weight, volume, declared value, and timeframe live in different places. One missing field breaks the entire quote pipeline.

Opaque routing

Air vs ocean vs inland trucking is decided by gut feel. No cost model, no Haversine distance, no transit-time tradeoff.

No total landed cost

Entry tax, freight, tolls, and inland trucking are never summed in one place. The shipper has no true cost picture.

Disconnected invoicing

Route, cost breakdown, and service fee are assembled manually after the fact — no unique transaction ID,

no audit trail.

S C O P E & C O N S T R A I N T S

Route:

International origin → United States destination only

Transport:

AIR or OCEAN — decided by weight, declared value, and timeframe preference (SPEED / COST)

Weight tiers:

≤500 kg or luxury or SPEED → AIR only | 500–2,000 kg → AIR or OCEAN compared | >2,000 kg + COST → OCEAN only

High-value:

Declared value >$2,500 USD triggers high-value classification and additional MPF + tariff calculation

3 of 4

CURRENT EVENTS WE CONSIDERED

Travel and shipping logistics have changed rapidly in the past year.

Rising Tariffs

Tariff pressures, especially unpredictable changes in rate, are putting immense pressure on businesses
In a market working with 20.4 trillion dollars of product, this is immensely impactful to day-to-day life.

Shift in B2B Models

Historically, businesses planned two major seasonal shipments each year. But that model is quickly eroding.
Zara, for example, drastically reduced the time between design and delivery—often turning around new collections in as little as five weeks.

Consolidation

The top ten players adds up to only 45 percent of the market, larger freight forwarders are increasing pressure on the rest.
Small and midsize freight forwarders have limited capacity to fund investments required to remain competitive— reducing competition.

Investor Preferences

In this environment, investors are likely to favor deals grounded in a clear vertical thesis and a well-tested synergy case rather than large balance sheet bets

ARTICLES

McKinsey:

https://www.mckinsey.com/capabilities/m-and-a/our-insights/travel-logistics-and-infrastructure-a-fragile-reset-paves-way-for-m-and-a-momentum

DCL:

https://dclcorp.com/blog/supply-chain/tariffs-are-reshaping-the-b2b-supply-chain/

4 of 4

T H E W O R K F L O W

One shared state. No forms.

Aniket — Orchestrator

Conversational intake

LLM extracts origin, destination, weight, volume, declared value, and timeframe from plain-language chat

into a ShipmentRequest.

Validation loop: halts and re-prompts until all

fields are confirmed complete.

→ ShipmentRequest

Ashwin — Economics

Constraints & entry tax

Classifies cargo by value (high-value if >$2,500 USD) and weight tier.

Applies transport logic to return AIR or SHIP.

Calculates US Merchandise processing Fee + standard category tariffs.

→ EconData

Riya — Routing

Pathfinding & cost

comparison

Haversine routing across origin city →port/airport → US port → destination city.

AIR and SHIP sub-agents each return freight cost, inland trucking, tolls, and transit time.

Picks optimal

mode per timeframe preference.

→ RouteData

Neel — Treasury

Downloadable invoice

Generates a PDF invoice with unique transaction ID, full shipment details, optimal route nodes, itemised cost

breakdown (freight, tolls, entry tax, service fee), and total landed cost.

User pays a small fee to view/download invoice.

→ SettlementStatus

1 of 4

2 of 4

3 of 4

4 of 4