DocSignal - Document Authenticity Signal Explorer

DocSignal — Document Authenticity Signal Explorer

A transparent document verification experience that turns black-box document checks into visible, explainable authenticity signals.

Verification Workflow

Upload, classify, score, and explain document authenticity in one flow

DocSignal accepts common identity and financial documents, runs multiple verification heuristics, extracts key fields, and surfaces an interpretable confidence score instead of hiding the reasoning behind a pass or fail.

Overview

DocSignal is a transparent document verification platform designed to analyze uploaded documents and expose the signals behind an authenticity decision. Rather than returning a single opaque verdict, the experience breaks down the checks performed on the file, classifies the document type, extracts key information, and surfaces explainable reasoning for the overall score.

What The Product Does

Accepts uploads for documents such as passports, driver's licenses, utility bills, bank statements, and paystubs
Supports drag-and-drop or click-to-upload flows for PDF, JPG, and PNG files up to 10 MB
Combines multiple verification heuristics, including rule-based logic and LLM-assisted checks
Classifies document type and extracts key fields needed for downstream review
Returns authenticity signals, an overall confidence score, and a human-readable explanation of the result

Implementation Details

Built with Next.js 14 App Router, TypeScript, and Tailwind CSS
Uses pdf-parse for PDF text extraction and tesseract.js v5 for OCR on image documents
Runs a modular pipeline for classification, field extraction, signal generation, scoring, and explanation
Produces 6 authenticity signals: text quality, consistency checks, template/layout hints, metadata anomalies, tamper heuristics, and field completeness
Maps the weighted score to a recommended action such as Auto-accept, Needs review, or Reject
Includes JSON export and an /eval page backed by bundled fixtures for sanity testing

Why It Matters

The core idea is to make document verification more inspectable. Instead of treating authenticity assessment as a black box, DocSignal shows the evidence and reasoning behind the score, making the workflow more useful for product teams, risk teams, and operations reviewers who need clarity rather than just a binary decision.

Design Decisions

Stateless by design: files are processed in-memory only, with no database or long-term storage
Heuristic-first pipeline keeps the default experience deterministic and inexpensive to run
Optional LLM insights sit behind a feature flag rather than being required for the core workflow
Server-side validation enforces file-type and size limits before analysis

Interface Notes

Single-screen workflow centered on upload and analysis
Designed to make document support and file constraints immediately visible
Presentation emphasizes trust, legibility, and interpretability over opaque automation

Role and Focus

Role: Solo builder across product framing, trust-oriented UX, and implementation.

Tech Stack: Next.js 14, TypeScript, Tailwind CSS, pdf-parse, tesseract.js, Vitest, Vercel.

Category: AI / ML, document verification, explainable scoring, risk tooling.

Positioning: Think "Stripe Radar for documents, but transparent."

Thumbnail Alt Text

DocSignal interface showing a document upload panel for verification with supported identity and financial document types and a trust-focused analysis workflow.

Moraldeep Sachdeo