AI Demos Research — the structured-intelligence platform. Every verdict on these pages opens to the execution behind it.
Graded 9 November 2026

Can Reducto extract text accurately from a clean scan?

Reducto read the clean scan without changing the values extracted from the digital original. Both versions returned all 20 requested values, preserving currency formatting and hire-period punctuation.

1 of 1 test case passed

Every test case in this scenario has a result.

Pass rate100%1 of 1 gradable
Coverage1 of 1test case with a result
1PassDid everything it was expected to do.
0FailNo test case failed.
0Not gradableEvery result here could be graded.
0UntestedEvery test case in this scenario has a result.
The pass rate is a summary. The evidence is the test case below: what we sent, what we checked, what the tool returned, and the proof.

The test case

Each test case is judged on its own: Pass, Fail, Not gradable, or Untested. The scenario result above is based on this test case.

Image-only twin, same schemaPassEvidence
What we sent
Input 1JSON · Schema pasted into Reducto Studio
{
  "instructions": {
    "schema": {
      "type": "object",
      "properties": {
        "invoice_number": { "type": "string", "description": "The invoice number." },
        "bill_to": { "type": "string", "description": "The customer the invoice is billed to." },
        "site": { "type": "string", "description": "The site the equipment was hired for." },
        "hire_period": { "type": "string", "description": "The hire period as printed." },
        "currency": { "type": "string", "description": "The currency of the invoice." },
        "subtotal": { "type": "string", "description": "The subtotal before tax, as printed." },
        "tax": { "type": "string", "description": "The tax / VAT amount, as printed." },
        "total_due": { "type": "string", "description": "The total amount due, as printed." },
        "line_items": {
          "type": "array",
          "description": "Every hired item line on the invoice, one entry per line.",
          "items": {
            "type": "object",
            "properties": {
              "description": { "type": "string", "description": "Description of the hired item." },
              "quantity": { "type": "string", "description": "Quantity hired." },
              "unit_price": { "type": "string", "description": "Unit price as printed." },
              "amount": { "type": "string", "description": "Line amount as printed." }
            },
            "required": ["description", "quantity", "unit_price", "amount"]
          }
        }
      },
      "required": ["invoice_number", "bill_to", "site", "hire_period", "currency", "subtotal", "tax", "total_due", "line_items"]
    }
  },
  "settings": {
    "citations": { "enabled": true }
  }
}
Input 2Document (PDF)
Input PDF: the original digital equipment-hire invoice
Open the PDF ↗
Input 3Document (PDF)
Input PDF: the image-only scan of the equipment-hire invoice
Open the PDF ↗
What the tool returned

Copied from Output file 1 (JSON), result › invoice_number › value

Text
TPH-3301

Copied from Output file 1 (JSON), result › bill_to › value

Text
Oakbridge Construction Ltd

Copied from Output file 1 (JSON), result › site › value

Text
Oakbridge Depot Extension

Copied from Output file 1 (JSON), result › hire_period › value

Text
03 Mar - 10 Mar 2026

Copied from Output file 1 (JSON), result › currency › value

Text
£

Copied from Output file 1 (JSON), result › subtotal › value

Text
£1,462.00

Copied from Output file 1 (JSON), result › tax › value

Text
£292.40

Copied from Output file 1 (JSON), result › total_due › value

Text
£1,754.40

Copied from Output file 1 (JSON), result › line_items · an excerpt

JSON
[
  {
    "description": "Telehandler, 4-tonne, weekly hire",
    "quantity": "1",
    "unit_price": "£850.00",
    "amount": "£850.00"
  },
  {
    "description": "Generator, 20kVA, weekly hire",
    "quantity": "2",
    "unit_price": "£210.00",
    "amount": "£420.00"
  },
  {
    "description": "Site fencing, per 10m panel, weekly",
    "quantity": "8",
    "unit_price": "£24.00",
    "amount": "£192.00"
  }
]

Excerpt ends here. Open the source for the complete output.

Copied from Output file 2 (JSON), result › invoice_number › value

Text
TPH-3301

Copied from Output file 2 (JSON), result › bill_to › value

Text
Oakbridge Construction Ltd

Copied from Output file 2 (JSON), result › site › value

Text
Oakbridge Depot Extension

Copied from Output file 2 (JSON), result › hire_period › value

Text
03 Mar - 10 Mar 2026

Copied from Output file 2 (JSON), result › currency › value

Text
£

Copied from Output file 2 (JSON), result › subtotal › value

Text
£1,462.00

Copied from Output file 2 (JSON), result › tax › value

Text
£292.40

Copied from Output file 2 (JSON), result › total_due › value

Text
£1,754.40

Copied from Output file 2 (JSON), result › line_items · an excerpt

JSON
[
  {
    "description": "Telehandler, 4-tonne, weekly hire",
    "quantity": "1",
    "unit_price": "£850.00",
    "amount": "£850.00"
  },
  {
    "description": "Generator, 20kVA, weekly hire",
    "quantity": "2",
    "unit_price": "£210.00",
    "amount": "£420.00"
  },
  {
    "description": "Site fencing, per 10m panel, weekly",
    "quantity": "8",
    "unit_price": "£24.00",
    "amount": "£192.00"
  }
]

Excerpt ends here. Open the source for the complete output.

Extraction result: the digital invoice's values with page citations

Open the whole JSON file ↗

Extraction result: the scanned invoice's values, identical to the digital extraction

Open the whole JSON file ↗
Expected vs. Found
✓Expected: The digital document and its image-only scan must return identical values, with both extracted for this comparison.Found: Both extractions returned the same 20 values, including a subtotal of £1,462.00 and a total due of £1,754.40 in each.
Supporting proof
Proof 1Screenshot
Studio screenshot: the digital invoice alongside extracted values and citation highlights
Open original ↗
Proof 2Screenshot
Studio screenshot: the scanned invoice alongside extracted values and citation highlights
Open original ↗
Cited values2 values from this file
job_id
199071d2-d155-4a24-a096-e1ccd4d9706d
duration
33.241049
Cited values2 values from this file
job_id
cafbcc81-2e63-4577-a61a-62e48f9b095e
duration
38.024216
Why this result

The scan had to yield the same values as the digital invoice. With no text layer to rely on, the identical outputs demonstrate reading from page images rather than embedded text.

Also observed in this test case

Extraction times. Processing the scan took about 38 seconds, versus about 33 seconds for the digital invoice. The roughly 5-second gap was too small for scanning to dominate processing time.

Tested by Shivohum Pandita · evidence dated 29 October 2026

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.

Software that produced the output
Reducto
Build or version
Not exposed by the hosted product
Plan or tier
Standard
Model
Extract v4 (the default extract model) and parse model r-1 (Preview)
Surface
Web app
Set up before the run
The Thistlewood equipment-hire invoice had to be processed digitally, then as an image-only scan, using the same extraction schema. The scanned version had to contain no text layer.
Schema and citations
The schema was pasted as JSON, with citations on. The same schema was used for both runs.
Schema
20 fields
Tested
29 October 2026 · Shivohum Pandita

How this scenario is graded

How we decide Pass, Fail and Not gradable. The same rules apply to every tool tested on this scenario.

How results are decided

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.

The rules
  • 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 test case's result says which one 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 test case's result 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.

LevelNameScope
BenchmarkStructured Document Extraction →19 scenarios · 10 tools
CapabilityOCR →
ScenarioA cleanly scanned document →
ToolReducto →

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.

from the publication record
18 November 2026First publishedStructured Document Extraction v1

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.

This matches what I see

You run the same kind of test against your own setup and get the same behaviour.

Agree →
This does not match

Yours behaves differently. Tell us what you got, with a screenshot if you have one.

Disagree →
Point out an issue

Something here is wrong — a reference value, a transcription, a grade.

Report an issue →
Request a retest

On a newer build, a larger dataset, or your own setup.

Request a retest →
We have fixed this

Tell us what changed and we schedule a rerun of the failing test case. The old result stays as history.

Vendor notice →
Filed against this evidenceNothing yet. Challenges, counter-evidence and fix notices appear here with their outcome, and stay on the page after they are resolved.
Cite this result
aidemos.com/benchmarks/structured-document-extraction/results/reducto/a-cleanly-scanned-document · 1 pass · 0 fail · coverage 1/1 · graded 2026-11-09

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.

Input 2 · Document (PDF) ↗application/pdf · 78 KB
Input 3 · Document (PDF) ↗application/pdf · 232 KB
Output file 1 (JSON) ↗application/json · 32 KB
Output file 2 (JSON) ↗application/json · 32 KB
Proof 1 · Screenshot ↗image/png · 189 KB
Proof 2 · Screenshot ↗image/png · 436 KB
File fingerprints (SHA-256)

A fingerprint identifies the exact file used for this result.

Input 2 · Document (PDF)fdcf03ac5ee53b083da58166f17fd41e7c72de79a9f3a3f6b67daa1360ddbca9
Input 3 · Document (PDF)6f1cc2a5f5e5e57494671341557d7a53d318b0f62bb8f8536165e1dac1b7a805
Output file 1 (JSON)71d04801031debc705e656ab15563695862377c02541e3524aec9f0737796036
Output file 2 (JSON)e5c5f495cedfba911b7a47b12bd8b013d5e4f277cf58b636990dd41634b8d543
Proof 1 · Screenshot14401cad731dd0889e3b1c5e235dafe6707b414292f16f559785fc6183332555
Proof 2 · Screenshot5701375be5d20bfb7b29fcae00905c5803b3cb3f6dc16a7ec1384fb203ee2852