Documentation Index
Fetch the complete documentation index at: https://docs.commodityai.io/llms.txt
Use this file to discover all available pages before exploring further.
What is CommodityAI?
CommodityAI is a Commodity Operating System for physical commodity businesses. It combines document intelligence, standard commodity objects, workflow automation, source provenance, and API access in one operating layer. Instead of treating documents as the endpoint, CommodityAI uses documents, emails, and files as source material for operational records: contracts, counterparties, shipments, allocations, costs, payments, product and quality records, and trade finance workflows.Key Features
Standard Commodity Objects
CommodityAI provides a standard object model for commodity workflows:- Contracts for purchase and sale terms, pricing, delivery, payment terms, and status
- Legs and allocations for tranche-level delivery structure and buy-to-sell matching
- Shipments for physical movement, containers, ports, carriers, and milestone dates
- Counterparties for customers, vendors, brokers, carriers, banks, insurers, surveyors, and other roles
- Costs and payments for freight, insurance, inspection, handling, financing, invoices, and settlement tracking
- Letters of credit and document requirements for trade finance review and presentation workflows
Source Intelligence
Upload documents through email forwarding or direct file upload, and CommodityAI automatically:- Classifies document types using AI-powered classification
- Splits multi-document files into logical sources
- Extracts structured data based on your custom schemas
- Links source material back to the operational records it creates or supports
- Provides provenance for field-level review, audit, and downstream reconciliation
Configurable Object Management
Define customer-specific schemas alongside the standard object model:- Create source record definitions for document-based data (invoices, contracts, etc.)
- Build custom objects for company-specific workflows that do not fit a standard object
- Add extension fields to standard objects where customer-specific fields are needed
- Set up workflows that route documents to the right extraction and object-creation pipeline
- Configure triggers based on email properties, document classifications, or data conditions
Multi-Tenant Architecture
Enterprise-grade security and isolation:- Company-scoped data - All data is isolated by company ID
- Role-based access control - Admin, member, and custom role permissions
- Audit trails - Complete history of all data modifications
- API key authentication - Secure programmatic access to your data
Integration Capabilities
Access your data programmatically:- REST API with cursor-based pagination for efficient data retrieval
- Rate-limited access - 100 requests/min, 10k requests/day per key
- Filtering and querying - Query by definition, timestamps, counters
- Real-time access - Data available via API immediately after extraction
Who Uses CommodityAI?
Commodity Trading and Supply Chain Teams
CommodityAI is built for businesses that need a reliable operational record across trading, logistics, and finance:- Trading teams managing purchase and sale contracts, tranches, allocations, and P&L inputs
- Logistics teams tracking shipments, containers, bills of lading, ports, carriers, and milestones
- Finance teams managing cost items, invoices, payments, settlement status, and trade finance requirements
- Operations teams connecting source documents to the records and workflows they support
Compliance and Audit Teams
Streamline compliance workflows:- Automated data extraction reduces manual data entry errors
- Confidence scoring flags low-confidence extractions for human review
- Audit trails provide complete history of object and source changes
- Structured export enables compliance reporting and analysis
Data Analytics Teams
Analyze operating data with source context:- Normalized data - Standard and custom records follow defined schemas
- API access - Pull data into your analytics tools and dashboards
- Field-level provenance - Track where values came from in source documents
- Confidence metrics - Filter by extraction confidence for quality analysis
How It Works
1. Document Upload
Documents enter the system through multiple channels:- Email forwarding - Forward emails to your company-specific address
- Direct upload - Upload files through the web interface
- Supported formats - PDF, images, Excel, Word documents
2. AI Classification
Our AI engine automatically identifies document types:- Multi-class classification - Documents are categorized by type
- Confidence levels - Each classification includes a confidence score (high/medium/low)
- Document splitting - Multi-page PDFs are intelligently split into logical documents
- Trigger matching - Classified documents route to the appropriate workflows
3. Data Extraction
Structured data extraction using your custom schemas:- LLM-powered extraction - Uses advanced language models (GPT-4, Gemini, Claude)
- Schema validation - Extracted data conforms to your defined field types
- Confidence scoring - Each field includes confidence metrics and alternatives
- Human review workflows - Low-confidence extractions can route to review queues
4. Object Creation and Linking
Extracted source records can create, update, or support operational records:- Standard objects - Contracts, counterparties, shipments, allocations, costs, payments, and related records
- Custom objects - Customer-defined objects with primary keys and schema validation
- Source links - Documents, files, emails, and source records linked back to the operational record
- Activity history - Changes retained for audit and review
5. Access and Integration
Your data becomes immediately accessible:- Web interface - View and edit records in the dashboard
- REST API - Programmatic access for integrations
- Export options - Download data or sync to external systems
- Real-time availability - Data accessible within seconds of extraction