Document Intelligence: How Titan Processes Contracts, Invoices, and POs

Titan is the document intelligence layer of the BLM OS. It classifies incoming documents by type, extracts structured data from them, parses contract terms and obligations, and matches invoices against purchase orders. Documents are processed on arrival rather than when someone has time to review them.

Document classification

The Document Classifier agent assigns a type to every incoming document (contract, invoice, purchase order, statement, application) based on content signals. Classification gates the downstream extraction workflow, ensuring that each document type is processed by the agent configured for it rather than a generic extractor.

Data extraction and contract parsing

The Data Extractor pulls structured fields from classified documents (counterparty names, amounts, dates, terms) and writes them to the Fusion Grid. The Contract Parser goes further: it identifies obligations, key dates, renewal clauses, and termination provisions, and surfaces items that require operator attention.

A document that sits in a queue is not being processed. It is being deferred. Titan removes the queue.

Invoice-to-PO matching

The PO Processor matches incoming invoices against open purchase orders, flags discrepancies in amount, quantity, or vendor, and routes matched invoices for payment approval. Items that cannot be matched surface in the Prime exception queue with the relevant context so the operator can investigate.