Platform · Detection Engine
Reliable data protection starts with knowing what a file, image, or prompt actually contains. The GuardStance Detection Engine combines exact data match, document fingerprinting, checksum validation, classifiers, and OCR to eliminate false positives.
Book a demoOne paragraph of a pricing model leaves.
Sensitive data detection is not just a keyword search. The format of the information, the matching method, and the policy conditions all influence whether an event is relevant.
GuardStance lets security teams combine multiple detection signals, thresholds, and recipient rules in a single policy—powering both classification and real-time DLP enforcement.
Your databases and reference documents are indexed inside your network.
Each policy picks the methods and thresholds it needs.
Content is checked as it moves, including images and archives.
Every event says which rule matched and why.
Records from your CRM, HR or core banking extracts, not patterns.
Whole or partial copies of reference documents, down to one paragraph.
Cards, national IDs, IBANs and passports, with checksum validation.
Source code, financial statements, contracts, CVs and engineering drawings.
Text in images, scans and screenshots is read and checked.
For example “more than 50 records” or “fingerprint plus external recipient”.
Because a catch rate measured on someone else’s data says little about yours. The methods are tuned in your environment during the proof of concept, and you see the results on your own files. We publish our methods, not a catch-rate figure.
No. Exact-match and fingerprint indexes are built and kept with your policy server. In on-premises editions they never leave your network, and the console shows which sources were indexed and when.
Yes. A policy can require a fingerprint match and an external recipient, more than a set number of records, or a classifier result plus a label. Each policy has its own combination, so one setting never has to fit every kind of data.
