ZENPORT Enhances Contextual Data Matching in Its AI Agent “ZenRead” — Reconciling Shipping Documents Line by Line with Purchase Orders
Matching product details and transaction history to translate each partner’s differently coded item names into the customer’s own items
Tokyo, Japan — September [DD], 2026 — Zenport Inc. (Head Office: Chiyoda-ku, Tokyo; CEO: Fumiyuki Ota; “Zenport”) has strengthened the judgment capability of ZenRead—its AI agent that reads shipping documents and translates them into the definitions used in the customer’s ERP data—through enhanced contextual data matching. The enhancement has been available since August 10, 2026.
ZenRead matches the details written on a document—trading partner, quantities, unit prices, and the like—against the transaction history with each partner held in the customer’s synced ERP data: the data that shows the context of the transaction. On this basis, it translates item names written in each partner’s own scheme into the customer’s items, ties each one to the matching line on the corresponding purchase order (PO), and determines exactly which line of which PO the document is for.
■ Background — Beyond OCR: The Heavy Work of Judging What Has Been Read
In global supply chain operations, the product names on commercial invoices and other shipping documents arrive in a different format and wording from every trading partner. With AI-OCR now widespread, the step of reading the text has largely been automated. But judging which of the company’s own items a reading corresponds to, and which line of which PO the document is for, still rests on the experience and manual work of the people in charge. The further reading and digitization advance, the more the step that follows—reconciling to the company’s items and purchase orders—stands out as the heavy manual work left after reading.
■ Where ZenRead Stands
The standard Zenport sets for document AI is not only that it can read text. It is that what has been read is connected all the way through: to which of the customer’s items it is, and to which line of which PO it belongs.
Starting from the moment a document arrives, ZenRead carries the work through in a single flow, from identifying the item to reconciling it against the purchase order, line by line. It does not do away with judgment. It leaves human confirmation to the cases that need it—and that alone takes most of the burden out of the work.
ZenRead started as an AI-BPO service and has since automated more and more of this judgment. Today it runs as an AI agent, executing everything from reading to reconciliation in one flow.
It asks for no changes—not to trading partners’ document formats, and not to the customer’s ERP systems.
■ Key Features
1. Item identification through contextual data matching
ZenRead identifies which item in the customer’s item master a product written on a shipping document corresponds to. It does not use probabilistic scores to make that judgment. It matches the details on the document—trading partner, quantities, unit prices, and the like—against the transaction history with each partner in the synced ERP data, and identifies the item within the context of that transaction.
In this way, item names written in each partner’s own scheme are translated into the customer’s items.
2. Line-by-line reconciliation with purchase orders
ZenRead ties each document to the specific purchase order, and the specific line on it, that the document is for. From the moment a document arrives, it updates which PO quantities are in transit and which order balances remain open. Even with corrections, split shipments, or transactions that span several documents, it keeps each document tied to the right line of the purchase order.
[Figure 1: English version of the published three-column comparison]
Note: Line-by-line reconciliation with purchase orders is available when PO data is synced.
3. People step in only where confirmation is needed
Reading a document and making the business judgment are two different jobs. In ZenRead, AI carries the work through in one flow—reading, identifying the item, and reconciling it against the purchase order. People confirm only what needs confirming.
Rather than leaving AI-OCR output as raw text, ZenRead connects it to the company’s items and to the right lines of its purchase orders, and carries it forward to a business judgment. This is the Human in the Loop design ZenRead has held to since launch.
■ In Practice — Knowing What Is in Transit
Inventory that has been shipped but not yet received—goods in transit—cannot be physically counted, and the evidence for its quantities and breakdown is scattered across documents from each trading partner. With ZenRead’s reconciliation, customers know which items are in transit and in what quantities, line by line against each purchase order, from the moment the documents arrive. In-transit inventory is captured from the shipping documents already handled in daily import and export work—and the records needed to keep a ledger build up as part of the workflow itself.
■ Implementation
ZenRead can be implemented on top of each trading partner’s existing workflows and document formats and the customer’s existing ERP systems. Because it works from data that already exists, it adds no new manual entry as a rule.
Data that customers enter or sync into the service, and the results it generates, are managed so that no other customer can view or use them. The service uses generative AI services under contracts and settings that prevent this data from being used to train or improve shared AI models. Zenport does not use this data to train models either, unless the customer explicitly consents or instructs otherwise.
■ Related Information
One Grid Journal, “Reconciling In-Transit Inventory Line by Line Against Purchase Orders” (Japanese): https://note.com/fumiyuki_ota/n/n6398524ec87a
Taking an importer’s in-transit inventory as its example, the article explains from an accounting and audit perspective why reconciliation must happen at the level of each line on the purchase order, not the document.
■ About ZENPORT — A World Where Cognitive Diversity Creates Prosperity and Innovation
ZENPORT envisions a world where global data, people, and economies are seamlessly connected, enabling sustainable prosperity and innovation through cognitive diversity.
Under this vision, Zenport provides a data foundation that lets customers connect across organizations without asking their partners to change: ZenRead, which connects trade documents to operational data, and ZenMapping, which connects data to ERP systems and master data reproducibly and traceably.
■ Company Overview
・Company name: Zenport Inc.
・Address: GranTokyo South Tower 11F, 1-9-2 Marunouchi, Chiyoda-ku, Tokyo 100-6611, Japan
・Representative: Fumiyuki Ota, CEO
・Vision: “A world where differences become strengths.”
・Mission: “Connecting data across organizations and processes to accelerate business.”
・Business: Development and provision of an AI-native ontology platform for global supply chains. Proponent of “Fit to Ontology.”
・URL: https://zenport.io
■ Media Contact
Zenport Inc., Public Relations — info@zenport.io
