Trust, privacy, and compliance

How the Privacy Vault protects sensitive data

The Privacy Vault is a core component of Unless’s architecture designed to protect sensitive personal data from the moment it enters the platform. It acts as a robust data-protection layer that ensures personally identifiable information (PII) such as names, emails, phone numbers, and account numbers are never exposed in their raw form to AI models or external parties. Instead, these sensitive identifiers are tokenized—replaced with safe placeholders—right at the platform’s boundary. This tokenization process means that the generative AI models only ever receive anonymized tokens, never the actual personal data.

how the Privacy Vault works

The Privacy Vault implements a comprehensive set of twelve privacy measures that work together to safeguard data. These include PII filtering, tokenization, data minimization, prompt anonymization, and strict EU data residency controls. When customer data enters Unless, the Vault immediately detects and filters out personal identifiers. If the AI agent needs to use personal context—such as addressing a customer by name—the Vault re-attaches the real identifier only within a controlled environment, never exposing it externally.

Filtering happens twice: once at ingestion when data first arrives on the platform, and again at runtime before any call to a third-party generative model. This two-step filtering ensures that no raw personal data leaves the platform or reaches the AI model providers. The tokenization process replaces sensitive values with tokens that only the Privacy Vault can map back to the original data, maintaining strict control over re-identification.

data residency and subprocessors

All personal data and the Privacy Vault itself reside exclusively within the European Union or European Economic Area. This design eliminates the need for adequacy decisions or transfer impact assessments because data never leaves the EU. Unless uses EU-based cloud providers such as AWS, Microsoft Azure, and Google Cloud, all operating within EU regions. Sub-processors involved in the platform’s operation are also EU entities, and their full list is available in your Data Processing Agreement (DPA). Importantly, business tools used for Unless’s own administration may be outside the EU but never handle end-user personal data.

privacy by design and compliance

The Privacy Vault is not just a feature but a foundational element of Unless’s privacy-by-design approach. It supports compliance with GDPR and the EU AI Act by ensuring that personal data is handled lawfully and transparently. The Vault enables configurable PII filtering and whitelisting, so you can control which fields the AI model can see in plain text. It also supports consent management, allowing you to require explicit user consent for tracking and personalization.

Additionally, the platform provides built-in workflows to process data subject requests such as access, correction, and erasure. When a data erasure request is fulfilled, tokens related to the deleted data are orphaned automatically, preventing any reuse of old references.

summary

The Privacy Vault protects sensitive data by filtering and tokenizing personal identifiers at the platform’s edge, ensuring that raw personal data never reaches AI models or leaves the EU. It enforces strict data residency, supports consent management, and enables compliance with GDPR and the EU AI Act. This comprehensive approach allows you to deploy AI-powered agents confidently, knowing that personal data is safeguarded throughout its lifecycle.