Does FutureSmart Document Intelligence trace a value to where it came from?
FutureSmart Document Intelligence failed to trace the subtotal to its document location. The subtotal was on page 2, but the saved result had no location field, the chat only named a chunk, and the magnifier highlighted text in the extracted-text panel rather than a returned pointer.
1 of 1 test case failed
Every test case in this scenario has a result.
The test case
Each test case is judged on its own: Pass, Fail, Not gradable, or Untested. The scenario result above counts this row.
Pointer to where the value sitsFailEvidence
“What is the subtotal on main_invoice_multipage_blackwood.pdf? Tell me which page and where in the document that value comes from.”
The subtotal is **£6,103.00**. It appears on **page 2 of 2**, in the **invoice totals table at the bottom**, on the row labeled **“Subtotal.”**
Chat response JSON: subtotal answer and source chunks
Copied from Output file 1 (JSON), line 3 · an excerpt
Open the whole JSON file ↗Source excerpt · partial file · response 1
"message": "The subtotal is **£6,103.00**. It appears on **page 2 of 2**, in the **invoice totals table at the bottom**, on the row labeled **“Subtotal.”**",
Copied from Output file 1 (JSON), line 3 · an excerpt
The tool’s response is printed above, under What the tool returned. Show this file
{ "object_id": "a24ef21c-747c-4a35-b8e5-442ec07b32ef", "file": "main_invoice_multipage_blackwood.pdf", "addedvia": "File", "status": "Completed", "created": "2026-09-15T09:38:31.232539", "updated": "2026-09-15T09:38:31.233182", "document_outline": ["Invoice Header", "Billing Information", "Line Items", "Totals", "Payment Instructions"], "document_type": "invoice", "entities": ["Desmond Ayotte"], "keywords": ["retail shelving", "checkout counter", "barcode scanner", "invoice"], "organizations": ["Meridian Retail Group", "Blackwood Distribution LLC", "Continental Trust Bank"], "summary": "This invoice from Meridian Retail Group to Blackwood Distribution LLC details the supply of various retail fixtures and store equipment. The total amount due for the items provided is £6,591.24, including VAT.", "temporal_context": "June 2020", "title": "Invoice MRG-INV-20481 for Blackwood Distribution LLC", "topics": ["retail equipment", "store fixtures", "procurement"], "branch_phone": null, "currency": "GBP", "damage_exceptions_0_date": null, "damage_exceptions_0_details": null, "damage_exceptions_0_location": null, "damage_exceptions_0_description": null, "damage_exceptions_1_date": null, "damage_exceptions_1_details": null, "damage_exceptions_1_location": null, "damage_exceptions_1_description": null, "damage_exceptions_remaining_date": null, "damage_exceptions_remaining_details": null, "damage_exceptions_remaining_location": null, "damage_exceptions_remaining_description": null, "delivery_date": null, "document_reference_number": "MRG-INV-20481", "document_type_label": "invoice", "fuel_surcharge": null, "incident_entries_0_severity": null, "incident_entries_0_description": null, "incident_entries_0_incident_date": null, "incident_entries_0_incident_type": null, "incident_entries_1_severity": null, "incident_entries_1_description": null, "incident_entries_1_incident_date": null, "incident_entries_1_incident_type": null, "incident_entries_remaining_severity": null, "incident_entries_remaining_description": null, "incident_entries_remaining_incident_date": null, "incident_entries_remaining_incident_type": null, "line_items_0_code": null, "line_items_0_amount": null, "line_items_0_quantity": null, "line_items_0_unit_price": null, "line_items_0_description": "Retail Shelving Unit - Steel, 5-Tier", "line_items_1_code": null, "line_items_1_amount": null, "line_items_1_quantity": null, "line_items_1_unit_price": null, "line_items_1_description": "Checkout Counter - Laminate Top", "line_items_remaining_code": null, "line_items_remaining_amount": null, "line_items_remaining_quantity": null, "line_items_remaining_unit_price": null, "line_items_remaining_description": [ "Security Tag Deactivator - Handheld", "Price Label Printer - Thermal", "Shopping Basket - 20L, Set of 10", "Storeroom Pallet Racking - 4-Bay", "Barcode Scanner - Wireless", "Receipt Paper Roll - Case of 50", "Mannequin - Adjustable, Full Body", "Display Signage Kit - Window Set" ], "loyalty_membership_number": null, "party_names": ["Meridian Retail Group", "Blackwood Distribution LLC"], "receipt_number": null, "statement_period": null, "total_amount_owed": 6591.24, "totals_tax": null, "totals_subtotal": 6103, "totals_total_due": 6591.24, "totals_closing_balance": null, "totals_opening_balance": null, "transaction_entries_0_amount": null, "transaction_entries_0_description": null, "transaction_entries_0_running_balance": null, "transaction_entries_0_transaction_date": null, "transaction_entries_1_amount": null, "transaction_entries_1_description": null, "transaction_entries_1_running_balance": null, "transaction_entries_1_transaction_date": null, "transaction_entries_remaining_amount": null, "transaction_entries_remaining_description": null, "transaction_entries_remaining_running_balance": null, "transaction_entries_remaining_transaction_date": null, "filled": 19 }
| ✗ | Expected: A pointer to the place the returned value actually appears, at the precision the product can provide.Found: The subtotal sits in page 2's totals block, but the saved result has no location field and the magnifier only highlights text in the extracted-text panel. |
The tool’s response is printed above, under What the tool returned. Show this screenshot
Open original ↗The tool’s response is printed above, under What the tool returned. Show this screenshot
Open original ↗The tool’s response is printed above, under What the tool returned. Show this screenshot
Open original ↗It had to point to the place the subtotal appears, but it only gave a chunk reference, a prose page mention, and a text highlight in the extracted-text panel. That is not a returned location.
Chunk reference only. The source reference names a chunk, but it gives no page, region, or span for the subtotal.
Repeated value check. On the repeated invoice number, the magnifier lands on the page-2 header copy rather than the labelled line.
Prose not location. The page-2 wording in the chat comes from the chunk text, not from a structured location field.
The extraction ran on 15 September 2026; the 6 October session reread the stored result and asked the chat.
The highlight appears only in the extracted-text panel; no page number, box, or offset is shown.
Configuration and setup
How this tool was set up for the run and what the test needed in place. Each row is a fact from the run's records; a fact the records do not hold is left out, not guessed.
How this scenario is graded
How we decide Pass, Fail and Not gradable. The same rules apply to every tool tested on this scenario.
Each test case gets one result per tool: Pass, Fail or Not gradable. A test case we haven't run yet shows Untested. There are no partial results.
- Pass: the tool did everything the test expected, and nothing it said contradicts the correct answer.
- Fail: at least one expected behaviour clearly didn't happen; the row says which and quotes the tool.
- Not gradable: our evidence couldn't settle the outcome (for example a record we needed is missing). It is never counted as a fail, and the row says what's missing.
Where this sits in the benchmark
This page is one cell of a larger study: one tool, one scenario. Only this benchmark's frame appears here.
| Level | Name | Scope |
|---|---|---|
| Benchmark | Structured Document Extraction → | 19 scenarios · 10 tools |
| Capability | Source Grounding → | |
| Scenario | Trace a value to its location → | |
| Tool | FutureSmart Document Intelligence → |
Global scenario definition → · Global capability definition →
History of this result
What has happened to this result since it was first published. Runs and grades are never overwritten: a retest or a re-grade publishes a new result and keeps the earlier one readable.
Act on this result
Nothing filed here edits the run or the grade. A challenge opens a review, and a review can produce a new run or a re-grade — which becomes the current result and leaves this one in the history.
You run the same kind of test against your own setup and get the same behaviour.
Agree →Yours behaves differently. Tell us what you got, with a screenshot if you have one.
Disagree →Something here is wrong — a reference value, a transcription, a grade.
Report an issue →Tell us what changed and we schedule a rerun of the failing test case. The old result stays as history.
Vendor notice →The same record is available as structured data through the AI Demos MCP server, with the counts, the coverage and every per-test-case reason carried as fields.
Verify the proof files
These files support this result. Open a file to inspect the original evidence.
File fingerprints (SHA-256)
A fingerprint identifies the exact file used for this result.