Deep Dive
Engineering field manuals: architecture, the data pipeline, OCR, classification, the AI profiler, and DLP.
Platform map
The shape of the platform on one page: Classify, Verify, Delete, the service map, and links into each field manual.
Essentials
A condensed field manual covering the conceptual model, pipeline, OCR, profiler, classification, and technology stack.
Platform Architecture
How the ~42-service estate fits together: app tier, storage, gateway, tenancy, and delivery.
Data Pipeline
The seven-stage journey from a connected source to an enforced policy.
OCR Pipeline
Five passes that turn a raw page into clean text and document-level signals.
OCR Function
How the ocr service turns a REQUEST_OCR document into text and identity signals, and which data-index fields it writes.
Classification
How findings become tags, document classes, and policies — then a keep, reclassify, or delete decision.
AI Profiler
NER, phrase, and dependency lenses: the detection engine behind every finding.
DLP Service
Block, quarantine, or alert on personal data in Microsoft 365 before it spreads.
Management Tool
How operations deploys, upgrades, and watches the customer fleet.
Privacy & Security Model
How a local, Mistral-based 8B model trained on synthetic data classifies documents inside the Privacy & Security Model Harness.