Contextual AI classification of unstructured data, with anomaly detection on how it moves

Cognni Ltd. was founded in Tel Aviv in 2019 by Guy Eisdorfer and Ami Marueli. The company works on a problem that has resisted automation for years: most sensitive material in an enterprise is unstructured, and the classification methods available to protect it either depend on pattern matching that misses meaning or on employees labelling their own files, which nobody does consistently. Cognni's models apply prior knowledge about document types to infer what a piece of content is, so a board paper, a term sheet, and a severance letter are recognized as distinct categories of business risk rather than as generic text.

Classification is only the first half. Once assets are labelled, the platform tracks how they circulate across mail, collaboration tools, cloud storage, and AI assistants such as Microsoft Copilot, and treats departures from normal flow as the detection signal. That yields insider-risk alerts grounded in the value of the information involved rather than in volume alone, and it gives governance teams a view of which categories of data are reaching generative AI systems. For incident response, the same index supports scoping work: investigators can reconstruct what was accessed, in what order, and how far it spread. Customers and partners include Microsoft, Check Point, SoftwareOne, Softchoice, and Accenture.