Introduction to ODIN

ODIN (Oceanus Digital Intelligence Network) gives commodity traders nine AI agents that run their desk like a major trading house. The product has two layers: agents (the constant) and services (the variables).

Layer 1 — The Agents

Nine roles modeled on universal trading functions validated across Trafigura, Glencore, Cargill, Bunge, and Louis Dreyfus Company by four independent deep research sources.

Front Office

Market Intelligence - Supply-demand analysis, arbitrage identification
Origination - Supplier discovery, deal structuring
Merchant/Sales - Buyer discovery, contract formalization
Hedging - Derivatives management, basis risk monitoring

Middle Office (Independent)

Risk Management - Credit limits, VaR, concentration risk
Compliance - KYC/AML, sanctions, PEP screening

Back Office

Operations - Logistics coordination, document assembly
Trade Finance - LC management, payment instruments
Settlement - Trade capture, P&L, reconciliation

Layer 2 — The Services

Everything the agents access beyond the trader’s own data, gated through MCP (Model Context Protocol):

ODIN Trade Network - Counterparty discovery
Compliance Screening - Sanctions, PEP, adverse media
Market Pricing - Spot prices, forward curves, freight rates
Logistics Tracking - Vessel AIS, port data
Document Services - LC validation, discrepancy detection
Financial Services - Loan origination

Agent Tiers

Tier	Coordination	Execution	Context	Override
Sub-agents (Free)	Human	Sequential	Siloed	None
Team agents (Premium)	Orchestrator	Parallel	Shared	Risk/Compliance can block

Research Base

Two rounds of deep research (six sources: Perplexity x2, Claude x2, Gemini, ChatGPT) validated:

Nine universal trading functions
Seven-phase trade lifecycle
Twelve override triggers
Fourteen-document flow maps
Alignment with EY’s Autonomous Trade Operations framework