Three-way match exception tasks from manufacturer AP, 11 firms

Summary

Accounts payable tasks built from the ERP and AP inbox records of 11 US manufacturers. Each task holds a purchase order line, its receipt and the supplier invoice that failed to match, plus the supplier's note, and asks the model to name the cause and the fix: price variance, short shipment, freight billed separately, unit of measure mismatch or duplicate invoice. Answers come from how the AP team actually cleared the exception, and the grader checks both the cause and the adjustment. Supplier and staff names, bank details, addresses and tax ids are removed or replaced with placeholders.

Collected to order. We collect it from US companies for your order. Scope, volume and timing are confirmed in writing before you sign.

Pricing on request

Request scope and pricing

At a glance

Planned scope. Volume, sources and timing are confirmed in writing before you sign.

Records: 31Kgraded tasks
Size: 50 MB
Tokens, estimated: 10M
Years: 2016 to 2025
Firms: 11

Owner profile: 11 US discrete manufacturers, 60 to 400 staff each, making machined parts, molded plastics and fabricated metal products
Industry: Manufacturing
Data type: Evaluation and task packs
Headcount: 60 to 400 staff per firm
Region: United States (Midwest, Southeast)
Languages: English
Formats: JSONL, Parquet
Updated: 2026-09-23

Sources

System	Kind	Records
NetSuite	Accounting and ERP	2.4M PO, receipt and invoice lines
Epicor	Accounting and ERP	1.7M PO, receipt and invoice lines
Microsoft 365 mail	Email	380K AP inbox emails

Good for

training finance agents to clear three-way match exceptions
evaluating cause classification on real AP exceptions
testing reasoning across PO, receipt and invoice tables plus free-text notes
measuring whether models propose the adjustment the team actually posted

Fields

Field	Type	Description
task_id	string	Stable task id
purchase_order	object	PO number, item, quantity, unit price and unit of measure
receipt	object	Quantity and date received
invoice	object	Invoice number, quantity, unit price and freight
supplier_note	string	The supplier's email or invoice note
exception_cause	enum	price_variance, short_shipment, freight_separate, uom_mismatch or duplicate
resolution	string	How the AP team cleared the exception
grader	enum	cause_and_adjustment

Example record

{
  "task_id": "T-57439-011902",
  "purchase_order": {
    "po": "[PO_88123]",
    "item": "Housing, injection molded, PC/ABS, black",
    "qty": 5000,
    "unit_price": 0.94,
    "uom": "EA"
  },
  "receipt": {
    "qty_received": 5000,
    "date": "2025-02-11"
  },
  "invoice": {
    "invoice": "[INV_40917]",
    "qty": 5000,
    "unit_price": 1.01,
    "freight": 0
  },
  "supplier_note": "Resin surcharge of 0.07 per piece applies to all Q1 shipments per our notice of 12/18.",
  "exception_cause": "price_variance",
  "resolution": "Buyer approved the surcharge through Q1, PO line amended to 1.01, invoice released for payment.",
  "grader": "cause_and_adjustment"
}

One record, de-identified. Placeholders such as [PERSON_12] and [COMPANY_4] stay the same across the whole dataset, so threads still read correctly.

Licenses available

Exclusive
One buyer. Nobody else can license it during the exclusive term, usually 2 years.
Non-exclusive
Licensed to more than one buyer.

What ships with it

Datasheet: what it is, what it is good for, verified size, sources, years, format and limits.
Provenance file: the owner shown by code, staff notice and consent status, exclusions, mapped to the EU AI Act Article 53 training-data summary and California AB 2013.
Manifest with a hash for every file.
QC report and de-identification report.
Your own copy, with unique canary records.

How it is de-identified

Default exclusions removed, files flattened to text, identifiers found in three layers and replaced with consistent placeholders, and the result measured against a 99% bar for each identifier type in each source.

The full method