Search our integrations below.
Private AI provides industry-leading redaction and pseudonymization of PII in unstructured data (99%+) across 50 different entities in 7 languages (English, French, German, Italian, Spanish, Portuguese, and Korean). Better yet, Private AI provides their models in Docker containers that run directly within clients’ environments, and never share your data (including Private AI) for maximum security. Private AI models leverage the latest advancements in transformer architecture to provide de-identification of PII based on context, which enables extremely high accuracy out-of-the-box.
Private AI acts as a standalone service within your Inbenta workflows. Simply set up Private AI Docker container within your infrastructure and then use a POST request to call Private AI REST API and pass your data through their models, and select which entities you’d like to redact. Private AI models will return the de-identified text along with a library of the removed PII. For more information check out Private AI docs here.
Private AI is often used to de-identify datasets that contain customer information so they can safely share them with data science, ML teams, or 3rd parties. Or simply to provide quick and easy regulatory compliance for any work or storage being done with those datasets.
Example Use Case:
Private AI's pseudonymization feature is particularly helpful for ML teams - it enables them to replace all PII with artificially generated replacement terms (ie. “Doug went to Winnipeg at 12pm” becomes “Jorge went to Taipei at 15h”), which in turns allows those datasets to be used for model training without negatively impacting model performance.