Native anonymization,
reversible and frictionless
Your documents are full of information that should never travel in clear text to a language model.
narratheque automatically replaces sensitive elements with neutral markers before any storage, while the real values remain protected by AES-256 encryption. The principle is simple: anonymized when you write, restored when you have the right to read.
Confidentiality
that never holds back your teams
Protected on write, readable on read
Rather than permanently hiding your information, narratheque reverses the usual logic. Names, emails, phone numbers or addresses are detected and then replaced with neutral references at the moment of storage. Authorized people, in turn, retrieve the original data effortlessly, with no manipulation.
The result: your requests draw on your knowledge base, the web or external models (ChatGPT, Gemini, Anthropic, Mistral…) without ever exposing a single critical piece of information to them.
Real data minimization,
without changing your work habits.
How does it work?
Three steps invisible to you, decisive for your data.
From detection to restitution, narratheque protects information without slowing down your teams.
Three identities for a single data item
Each piece of sensitive information comes in three complementary forms, to reconcile usability and protection.
A neutral reference
A non-reversible fingerprint
The encrypted value
Compliance without the manual chore
Automatic detection → substitution before storage → controlled restitution.
You stay compliant, without spending your days on it.
Manual anonymization
- Re-read each document line by line
- Redact by hand, at the risk of omissions
- An irreversible anonymization
With narratheque
- Automatic detection of sensitive data
- Substitution before storage
- Reversible and controlled restitution
When AI sees everything, your secrets leak.
Without anonymization, every request sends your sensitive information in clear text. Native anonymization cuts the risk off at the source.
Your documents go to the AI with their personal data in clear text.
A name, an email, a number… and the sensitive information ends up exposed to the model.
Every day, without anonymization:
- you re-read each document by hand
- you risk omissions and non-compliance
- you give up using AI on your sensitive files
👉 The result: exposed data, and restricted usage.
4% of revenue in the event of a GDPR breach
Fines can reach €20M or 4% of global revenue. Anonymization reduces the risk at the source.
0 sensitive data sent to AI
Models receive only neutral references. Your original information never leaves your encrypted vault.
3 identities per data item
Neutral reference, non-reversible fingerprint and AES-256 encrypted value: usability and protection combined.
Compliant, encrypted, reversible. Frictionless.
Native anonymization makes you compliant while preserving the real use of your data.
Our other security features
Sovereign hosting
Access control
AES-256 encryption
Your frequently asked questions
This section is only here for SEO purposes; otherwise, you could simply ask your questions to our ChatBot connected to our solution’s knowledge base…
Aucune question sélectionnée.
Have specific needs? Let’s talk about your project.
narratheque.io is the consumer-facing version of the KDBCore by Jolifish Europe application core. It lets you build dedicated environments to meet specific confidentiality, isolation or hosting requirements. Feel free to book a meeting to tell us about your project.
