Two line-item columns remain suspect: Reconciliation is empty across all 8 rows, and Reference Number is fixed at 1 for all 8 rows.
What was measured
Extraction Accuracy
Are field values correct, complete, and free of OCR or parsing errors, including numerical precision on financial fields?
decisive for this rankingtransformation
Correct field values are the core of the job; wrong or incomplete extraction means the tool failed to retrieve the structured data from the document. (3 of 3 judges)
What was given, what came back
Test input: Invoice PDF · pdf · group: business-document-extraction
Input — what we sent
A two-page broadcast advertising invoice PDF with nested header metadata, billing and remit addresses, and eight line items spanning a page break. It was used to stress hierarchical line-item extraction, amount precision, time/day parsing, code extraction, and summary validation.
Why this input is hard
- · Nested line-item hierarchy extraction
- · Multi-page continuity across a page break
- · Precision on large dollar amounts and totals
- · Parsing complex time slots and day patterns
- · Extraction of Ad IDs and reconciliation codes
- · Mapping structured metadata sections correctly
- · Financial summary validation
- · Handling political advertising compliance text
Output — unretouched
Also checked on this input — same tool, 2 other criteria
Semantic Field Enrichment✓ WorkedThe tool correctly splits compound scheduling data for all 8 line items, with Day of Week and Day Pattern both populated correctly instead of being garbled together.Table & Record Completeness✓ WorkedThe line-item table keeps all 8 rows, including the row that spans the page break, without introducing a duplicate line item.
Provenance
- Observation
- 9790387d-d986-4d2c-97c7-fb2ed6765292
- Evidence run
- a061b9e7-a9c5-443d-a171-b296aaf51b8c
- Study
- Extract and query structured data from documents using natural language
- Research task
- 86b9y25e5
- Tested at
- not recorded
- Source
- first-party
- Evidence state
- verified
- Proof shown
- input + output shown
- Cost / latency
- not captured
- Repeat run
- not captured
- Tester
- not captured
The last three rows are honest blanks, not placeholders — our capture has no field for them yet.
Query this
get_evidence({
tool: "docsumo",
scenario: "business-document-extraction"
})MCP · mcp.aidemos.com/api/mcp
Free with attribution.
Same input, same check — 8 other tools
measured on Extraction Accuracy
Datalab✓ WorkedRecovers invoice metadata and financial totals accurately: invoice number 4064621-1, invoice date 10/28/12, gross total 29750, agency commission 4462.5, net amount due 25287.5, and 8 aired spots match the source.Extend AI✗ FailedIt over-reads the payment-term field, outputting `Payment Terms 30 Days` instead of the requested `30 Days`.Landing AI✗ FailedMisreads at least one alphanumeric identifier: line 1's source Ad-ID NRCCWI071005 is extracted as NRCCW1071005.LlamaParse✓ WorkedThe invoice totals and headline metadata are extracted consistently: aired_spots is 8, gross_total 29750, agency_commission 4462.5, net_amount_due 25287.5, and payment_terms 30 Days match the source invoice.Nanonets✓ WorkedExtracts invoice and finance values accurately, including invoice number 4064621-1 and summary totals of 8 aired spots, 29750 gross, 4462.5 commission, and 25287.5 net amount due.Reducto✓ WorkedCorrectly extracts the line-item frequency value as '1x' and preserves citation plus confidence metadata for that field.Retab◐ MixedRetains the source label in payment_terms, returning 'Payment Terms 30 Days' instead of only the requested value, so the field is not fully normalized.Unstract✓ WorkedCaptures the invoice header fields and financial totals exactly, including invoice_number 4064621-1 and the printed gross_total 29,750.00, agency_commission 4,462.50, and net_amount_due 25,287.50.
This evidence is published in
Real inputs and real outputs, no retouching · every cell queryable via API & MCP · aidemos.com