Fortifiers Logo
Fortifiers
Back to Documentation

Extraction Models

How Fortifiers uses AI to extract data from documents.

Hybrid Extraction Engine

Fortifiers uses a hybrid approach combining multiple AI technologies to achieve industry-leading accuracy. We dynamically select the best model for each field in a document.

1. OCR

Optical Character Recognition

Converts images and scanned documents into machine-readable text.

  • Tesseract + Google Cloud Vision
  • 98.5% accuracy on printed text

2. LLM Extraction

Large Language Models

Understands document structure, context, and implied information.

  • Google Gemini 2.0 Flash + Fine-tuning
  • Trained on 10M+ documents

3. Pattern Matching

Regex & Rules

Deterministic extraction for structured fields like dates, amounts, and IDs.

  • Zero hallucination rate
  • Validates LLM outputs

Extraction Pipeline

  1. 1

    Document Upload

    File is received and classified by type.

  2. 2

    OCR Processing

    Text is extracted from images/PDFs.

  3. 3

    Layout Analysis

    Document structure is identified (tables, headers).

  4. 4

    LLM Extraction

    AI model extracts key fields based on document type.

  5. 5

    Validation & Review

    Data is validated against rules; low confidence flagged.

Confidence Scores

90-100%High

Auto-approved. No human review needed.

70-89%Medium

Flagged for quick human review.

< 70%Low

Requires manual verification.

Custom Model Training

Enterprise customers can fine-tune extraction models on their specific document formats.

  • Provide 50-100 sample documents
  • 2-3 week training turnaround