Forensics for documents and text. We examine file structure, typography and scan traces for signs of editing, and treat statistical text analysis as supporting evidence, never as proof of who wrote something.
[ 01 / Paradigm shift ]
Target vs Reality
Current
Edited Without a Trace
A changed figure or retyped line looks right to the eye, and the editing history is buried inside the file.
Target
Every Change Surfaced
Structure, typography and scan sensors surface edits by page and region, each with the measurement and its limits.
[ 02 / Method ]
Built for scrutiny
01Preserve
The source file and rendered pages are hashed and sealed.
02Parse
Internal objects and revisions are read from the file.
03Compare
Typography and scan sensors compare regions on each page.
04Report
Findings are listed by page and region with their limits.
Document Analysis Chain
Control planeAnalysis plane
01Structure Parse
PDF OBJECTS · REVISIONS
Reads internal structure, revision history and embedded objects of the file.
02Typography Sensors
FONTS · GLYPHS · SPACING
Compares fonts, glyph shapes and spacing across the page for regions that were retyped.
03Scan & Print Traces
NOISE · HALFTONE
For scans, checks noise and halftone patterns for pasted or retouched areas.
04Text Consistency
STATISTICAL
Compares writing patterns across sections as supporting evidence, never as proof of authorship.
05Findings Report
PAGES · REGIONS
Lists each finding by page and region with the measurement and its limits.
Control plane / CUSTODY LOG
Case Ledger
Hashes the source file and each extracted page image.
Hover or tap a stage to inspect
[ 03 / Core systems ]
Core sensors
01 / STRUCTURE01 / 04
FILESTRUCTURE
PDF and office files keep a record of how they were built and edited.
OBJECTS · REVISIONS · EMBEDDED FILES
Read the file, not just the page.
MEASURES
Object tree, incremental updates, producer fields and embedded content.
REPORTS
Evidence that content was added or replaced after the first save.
LIMITS
Flattened or re-exported files lose history, which the report notes.
02 / TYPOGRAPHY02 / 04
TYPOGRAPHYSENSORS
Retyped text rarely matches the original font, weight and spacing exactly.
FONTS · GLYPHS · KERNING
Compare like with like.
MEASURES
Font usage, glyph shapes and character spacing across lines.
REPORTS
Regions whose typography differs from the surrounding text.
LIMITS
Mixed-font templates can look inconsistent legitimately, so each flag is reviewed in context.
03 / SCAN03 / 04
SCAN & PRINTTRACES
Scanned documents keep noise and halftone patterns that pasted areas often lack.
NOISE · HALFTONE · EDGES
The paper remembers.
MEASURES
Noise and halftone behaviour, edge sharpness and compression blocks.
REPORTS
Areas that look pasted, cleaned or retouched.
LIMITS
Low-quality scans reduce sensitivity, and the report states this clearly.
04 / TEXT04 / 04
TEXTCONSISTENCY
Statistical comparison of writing across sections, used as a lead and never as a conclusion.
STYLE · REPETITION · TIMELINE
Supporting evidence only.
MEASURES
Vocabulary, phrasing and structure patterns across sections.
REPORTS
Sections that differ markedly, with the measured differences.
LIMITS
This cannot identify an author, and machine-generated text detection is treated as unreliable on its own.
[ 04 / Reference patterns ]
Where it applies
FINANCIAL
Payslips, statements and IDs
Screening of submitted documents for edits before a lending decision.
INVESTIGATION
Contested contracts and letters
Structure and typography evidence for the examiner's notes.
GOVERNMENT
Certificates and permits
Checks on digital and scanned records against expected structure.
[ 05 / Delivery & deployment ]
From evidence to report
Findings are listed by page and region, and exported with the original hashes. Documents stay inside your environment.
Forensics for documents and text. We examine file structure, typography and scan traces for signs of editing, and treat statistical text analysis as supporting evidence, never as proof of who wrote something.
What changes compared with the current approach?+
A changed figure or retyped line looks right to the eye, and the editing history is buried inside the file. Structure, typography and scan sensors surface edits by page and region, each with the measurement and its limits.
Can it run on-premise or air-gapped?+
Yes. Analysis runs on your own hardware, and the platform can be installed with no connectivity. Reports are exported as signed PDF and machine-readable JSON.
How do I start?+
Request a briefing by form, WhatsApp (+60 17-773 7302) or email (kumar@spearcompute.com). We scope a pilot on samples you provide.