Evidence · first-party tested

Rebuilds a scanned two-column page into a coherent reading flow, preserving the heading-to-paragraph sequence instead of outputting the columns in a disordered OCR dump.

✓ Worked🧾 artifact-verifiedTested Jun 20, 2026Llamaparse
Output — unretouched
Scanned two-column ecology text page with study area and stand prescriptions.
Scanned two-column ecology text page with study area and stand prescriptions.
Output — unretouched
Flattened hierarchical extraction of the study area section.
Flattened hierarchical extraction of the study area section.
The observation

Rebuilds a scanned two-column page into a coherent reading flow, preserving the heading-to-paragraph sequence instead of outputting the columns in a disordered OCR dump.

Criterion: Reading Order & Structure · Scenario: Scanned research paper (group: complex-pdf-to-markdown)
Query this
get_evidence({
  tool: "llamaparse",
  scenario: "complex-pdf-to-markdown"
})
MCP · mcp.aidemos.com/api/mcp
Free with attribution.
Same input, other tools
Real outputs, no retouching · every cell queryable via API & MCP · aidemos.com