Veritas Lab – Technical Documentation

Comprehensive technical documentation for the Veritas Lab document intelligence platform.

Overview

Veritas Lab is a document intelligence system that leverages advanced AI to Extract, Analyze, and Verify information from unstructured documents. Built specifically to operate within the Canadian sovereign cloud, Veritas Lab enables compliance teams, auditors, and legal professionals to automate document review while maintaining strict data residency and regulatory standards.

AI Architecture: Extract • Analyze • Verify

Phase	Capabilities	Technologies Used
Extract	OCR, key field detection, named entities	AWS Textract, spaCy 3.6, en_core_web_sm
Analyze	Clause classification, entity linking, summarization	Legal-BERT, HuggingFace Transformers, OpenAI GPT-4, Claude, spaCy pipelines
Verify	Correlation checks, anomaly detection, insight scoring	scikit-learn, vector DB (Weaviate), custom logic rules, metadata validation

Data Flow

graph TD; A["Upload to S3"] --> B["AWS Textract OCR"]; B --> C["spaCy Entity Recognition"]; C --> D["HuggingFace Model or GPT-4"]; D --> E["Clause Classification / Summary"]; E --> F["Weaviate or Vector Scoring"]; F --> G["Anomaly / Insight Verification"]; G --> H["Overlay + JSON + Compliance Output"];

Document processing flow from upload to final output

Pricing Tiers

Gold

Basic extraction capabilities

Textract + spaCy
~85% Accuracy
$0.08–0.10 per document
Basic extraction, dates, parties, NER

Platinum

Advanced analysis capabilities

Gold + Legal-BERT / GPT
~92% Accuracy
$0.15–0.25 per document
Contract analysis, clause detection

Diamond

Complete intelligence suite

Full Stack + LLM + Vector + Anomaly Scan
~97% Accuracy
$0.30–0.50 per document
Full insight reports, risk & compliance