Cognni

Cognni, founded in Tel Aviv in 2019, builds an AI-driven data intelligence platform aimed at a persistent problem in data protection programs: most sensitive data is never accurately classified in the first place, which makes every downstream DLP, access control, or compliance control built on top of that classification unreliable. Cognni’s platform autonomously scans, maps, and classifies data across databases, files, and cloud services, using an approach the company describes as mimicking how a human would infer the meaning and sensitivity of a document from context rather than relying purely on keyword or regex pattern matching.

The product line has expanded from data-at-rest discovery and classification into data-in-motion classification, data subject search (relevant to GDPR/CCPA subject access requests), and data security posture dashboards, plus an “Autonomous Learning Classifier” the company launched to reduce the manual training typically required to tune classification models to an organization’s specific document types. Cognni integrates with Microsoft Purview Information Protection and other DLP tooling rather than replacing them, positioning itself as the classification layer that feeds better labels into whatever enforcement tools a customer already has.

Cognni’s membership in the Microsoft Intelligent Security Association (MISA) and its direct Purview integration are independently checkable signals that a major platform vendor considers its technology interoperable and credible enough to build toward. The company has raised a Series C round (announced November 2023) with investors including Microsoft for Startups and Springtide Ventures, with total disclosed funding in the high single-digit millions to around $18 million depending on the source — a modest capital base for a Series C-labeled company, worth noting rather than taking the round label at face value.

For CISOs and data governance teams, Cognni’s value proposition is fixing the classification foundation that most DLP and access-control programs quietly get wrong, rather than adding another enforcement layer. Its evidence base leans on integration credibility (Microsoft Purview, MISA membership) more than named customer case studies or independent classification-accuracy benchmarks, which buyers should weigh accordingly.