{"observation":{"id":"aa04c457-96be-40c1-87a5-5813f3f86c31","tool":"reducto","tool_name":"Reducto","criterion":"markdown-quality","criterion_name":"Markdown Quality","criterion_definition":"Produces clean, well-structured, usable markdown rather than a flat text dump.","criterion_evidence_type":"transformation","criterion_rank_role":"decisive","criterion_rank_role_reason":"The ranking is specifically about producing clean Markdown, so the quality and usability of the Markdown output directly measures success. (3 of 3 judges)","scenario":"financial-report-table-heavy","scenario_name":"Financial Report - Table Heavy","group_tag":"financial-report-table-heavy","scenario_description":"A table-heavy corporate financial report used to test extraction of dense, hierarchical financial statements with grouped columns, multi-row headers, segment-reporting tables, and narrative disclosures.","modality":"pdf","input_text":null,"input_artifact_refs":[{"alt":null,"url":"https://cdn.futuresmart.ai/public/aidemos/fba42ade5adb4814a3381dd4487a3910.pdf?v=1","role":"input","filename":"doc.pdf"}],"stresses":["Multi-page financial statement extraction","Hierarchical table reconstruction","Grouped columns and multi-row headers","Reading order in a report with mixed narrative and tables","Document structure retention","Markdown usability"],"verdict":"worked","score":null,"score_total":null,"note":"Keeps the syntax clean: no <signature>, <empty>, <b>, <i>, or <u> tags appear anywhere in the output, and the table-of-contents extract renders as a valid pipe table.","evidence_state":"verified","source":null,"artifacts":[{"url":"https://cdn.futuresmart.ai/public/aidemos/22ebace50155483ca7780413ab257455.png?v=1","role":"input","alt":null},{"url":"https://cdn.futuresmart.ai/public/aidemos/4c17b98e0cc94677861c239aec4808ca.png?v=1","role":"output","alt":null},{"url":"https://cdn.futuresmart.ai/public/aidemos/26a6176821d24574bf506ad8e85586e7.png?v=1","role":"input","alt":null},{"url":"https://cdn.futuresmart.ai/public/aidemos/e6e50071c0774c1fb1d9adef3e589dc0.png?v=1","role":"output","alt":null}],"run_id":"6e3160de-fe46-4b45-b071-72560b5c5d0e","study_title":"Convert a Complex PDF into Clean Markdown with an API","study_kind":"generation","research_task":"86b9h7t37","tested_at":null,"completeness":"input-and-output","input":{"state":"files","text":null,"files":[{"url":"https://cdn.futuresmart.ai/public/aidemos/fba42ade5adb4814a3381dd4487a3910.pdf?v=1","filename":"doc.pdf","alt":"Financial Report - Table Heavy","role":"input"}],"modality":"pdf","stresses":["Multi-page financial statement extraction","Hierarchical table reconstruction","Grouped columns and multi-row headers","Reading order in a report with mixed narrative and tables","Document structure retention","Markdown usability"]},"tool_page_slug":"reducto","tool_url":"https://aidemos.com/tools/reducto","permalink":"https://aidemos.com/evidence/aa04c457-96be-40c1-87a5-5813f3f86c31","api_url":"https://ai.aidemos.com/v1/observations/aa04c457-96be-40c1-87a5-5813f3f86c31"},"peers":[{"id":"91c495ea-e42f-469b-8dbf-70720d32cb86","tool":"llamaparse","tool_name":"LlamaParse","verdict":"mixed","score":null,"score_total":null,"note":"Renders the table of contents as sequential text with page numbers instead of a nested TOC structure, so the markdown is usable but flattened.","artifact_count":1,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/8bdf29da12ad4937a62553f452727c1b.png?v=1","evidence_url":"https://aidemos.com/evidence/91c495ea-e42f-469b-8dbf-70720d32cb86"},{"id":"9c7a0e04-472d-43b7-a08c-8cba8763be40","tool":"mistral-ai","tool_name":"Mistral AI","verdict":"worked","score":null,"score_total":null,"note":"The output is organized as a downloadable ZIP containing a consolidated markdown document plus individual page-level files for localized inspection.","artifact_count":3,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/b12d5cd772a647818ad1833789b090d6.pdf?v=1","evidence_url":"https://aidemos.com/evidence/9c7a0e04-472d-43b7-a08c-8cba8763be40"},{"id":"906dd315-1d24-43c4-a4d2-720f7a3c3aa0","tool":"pdf-ai","tool_name":"PDF.ai","verdict":"failed","score":null,"score_total":null,"note":"Markdown export failed on an 18-page table-heavy financial report; the report records \"Output MD: Output failed.\"","artifact_count":2,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/b12d5cd772a647818ad1833789b090d6.pdf?v=1","evidence_url":"https://aidemos.com/evidence/906dd315-1d24-43c4-a4d2-720f7a3c3aa0"},{"id":"517504c5-f1cc-4741-acfb-c00b4f29897f","tool":"tensorlake","tool_name":"Tensorlake","verdict":"worked","score":null,"score_total":null,"note":"Returns the extraction as structured markdown in the Document Markdown workflow, making the output usable rather than a raw text dump.","artifact_count":2,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/4f55f711b11b4124b7d3fdcd08810819.mp4?v=1","evidence_url":"https://aidemos.com/evidence/517504c5-f1cc-4741-acfb-c00b4f29897f"}],"other_criteria":[{"id":"2ca4b8f8-5317-4d78-bf58-1e1f88758a01","criterion":"advanced-features","criterion_name":"Advanced Features","rank_role":"context","verdict":"failed","score":null,"score_total":null,"note":"The confidence field is nearly useless here: the only low-confidence flag is a false alarm on the correct page-18 section header, while the genuinely broken Business Results table stays high-confidence throughout.","artifact_count":3,"evidence_url":"https://aidemos.com/evidence/2ca4b8f8-5317-4d78-bf58-1e1f88758a01"},{"id":"991345dd-5c2a-4f09-8dc3-76918b2843d9","criterion":"advanced-features","criterion_name":"Advanced Features","rank_role":"context","verdict":"worked","score":null,"score_total":null,"note":"Schema-driven extract.run fully recovers the broken Business Results table, separating the 2024 value and percent-change fields that parse.run had merged together.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/991345dd-5c2a-4f09-8dc3-76918b2843d9"},{"id":"bfda059c-a8f9-41ce-8796-bcc01ea2026d","criterion":"complex-document-handling","criterion_name":"Complex Document Handling","rank_role":"decisive","verdict":"mixed","score":null,"score_total":null,"note":"Processes the 18-page report in 9.7 seconds with no truncation; the 9-column segment table at page 17 is flawless even though the first table on page 2 is broken.","artifact_count":3,"evidence_url":"https://aidemos.com/evidence/bfda059c-a8f9-41ce-8796-bcc01ea2026d"},{"id":"cb41b326-14c7-4854-ae36-24b1588cda74","criterion":"reading-order-structure","criterion_name":"Reading Order & Structure","rank_role":"decisive","verdict":"failed","score":null,"score_total":null,"note":"Provides zero markdown heading markup across all 18 pages; the Roman-numeral sections and numbered subsections are flattened into plain text despite the source’s visible section hierarchy.","artifact_count":4,"evidence_url":"https://aidemos.com/evidence/cb41b326-14c7-4854-ae36-24b1588cda74"},{"id":"8c8cffd6-9942-4cc3-95c3-b837dfd045e3","criterion":"table-preservation","criterion_name":"Table Preservation","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"Preserves the balance-sheet hierarchy and values, including Assets > Current assets > Cash and deposits, and also reconstructs the later 9-column segment table with all values correct.","artifact_count":4,"evidence_url":"https://aidemos.com/evidence/8c8cffd6-9942-4cc3-95c3-b837dfd045e3"},{"id":"f597cbbc-cbe9-4694-8b5e-078b1489986b","criterion":"table-preservation","criterion_name":"Table Preservation","rank_role":"decisive","verdict":"failed","score":null,"score_total":null,"note":"Breaks a 5-column business-results table by duplicating the 2025 header and merging the 2024 value and percent change into one cell; the Net sales row ends up with 254,811 and 2.6 smashed together.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/f597cbbc-cbe9-4694-8b5e-078b1489986b"},{"id":"2d1ae485-b6cc-4e88-813c-fce6d08fdd87","criterion":"text-ocr-completeness","criterion_name":"Text & OCR Completeness","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"Converts all 18 pages with no omitted text; a dense financial-condition paragraph preserves multiple JPY-billion values exactly, including JPY1,252.7 billion, JPY623.6 billion, JPY27.2 billion, JPY10.8 billion, JPY5.0 billion, and JPY9.7 billion.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/2d1ae485-b6cc-4e88-813c-fce6d08fdd87"},{"id":"cd90d364-bb6e-4ca0-9248-fe516d1a8765","criterion":"visual-content-retention","criterion_name":"Visual Content Retention","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"Retains the recurring page-header logo consistently across all 17 pages that carry it; the returned crops on pages 2 and 18 are the same mark and wordmark, with only a 1px width difference (453x36 vs 452x36) consistent with rounding.","artifact_count":4,"evidence_url":"https://aidemos.com/evidence/cd90d364-bb6e-4ca0-9248-fe516d1a8765"}],"appears_in":[{"page_type":"ranking","slug":"pdf-to-markdown-apis","title":"Best AI Tools to Convert Complex PDFs into Clean Markdown with an API","url":"https://aidemos.com/best/pdf-to-markdown-apis","binding":"run"}],"same_scenario":[{"id":"91c495ea-e42f-469b-8dbf-70720d32cb86","tool":"llamaparse","tool_name":"LlamaParse","verdict":"mixed","score":null,"score_total":null,"note":"Renders the table of contents as sequential text with page numbers instead of a nested TOC structure, so the markdown is usable but flattened."},{"id":"9c7a0e04-472d-43b7-a08c-8cba8763be40","tool":"mistral-ai","tool_name":"Mistral AI","verdict":"worked","score":null,"score_total":null,"note":"The output is organized as a downloadable ZIP containing a consolidated markdown document plus individual page-level files for localized inspection."},{"id":"906dd315-1d24-43c4-a4d2-720f7a3c3aa0","tool":"pdf-ai","tool_name":"PDF.ai","verdict":"failed","score":null,"score_total":null,"note":"Markdown export failed on an 18-page table-heavy financial report; the report records \"Output MD: Output failed.\""},{"id":"517504c5-f1cc-4741-acfb-c00b4f29897f","tool":"tensorlake","tool_name":"Tensorlake","verdict":"worked","score":null,"score_total":null,"note":"Returns the extraction as structured markdown in the Document Markdown workflow, making the output usable rather than a raw text dump."}]}