Pseudonymize your documents, and include the values in your AI deliverablesPseudonymize your documents consistently, and restore (de-pseudonymize) the original values in your AI assistants' outputs
Pseudonymize your documents, and include the values in your AI deliverablesPseudonymize your documents consistently, and restore (de-pseudonymize) the original values in your AI assistants' outputs
Between James Carter and Sophia Brown
residing at 14 Elm Street, London.
Consulting services for Acme Logistics Ltd., french regulated company, with registered office at 27 Voltaire Avenue, Manchester.
Signed on 04/12/2026.
Supported By
Service agreement.pdf
Reading file…
Import your documents from over 100 supported formats, and get a pseudonymized or anonymized version in just a few seconds.
Send the pseudonymized corpus to your AI assistant without changing your usual workflow, while maintaining professional confidentiality regarding your documents.
Summary note - Ruling 26-06-2026.docx
Restoring…
Restore the original value of each pseudonym within the files generated by the AI or its responses, and continue your work.
Our technology relies solely on in-house, open-source building blocks, offering full independence from third-party services and eliminating shadow IT.
Beyond being a central argument in our value proposition, these guarantees of independence and sovereignty form the very foundation of our SaaS Terms of Service.
Anonymize or pseudonymize your users' documents and prompts before they reach the LLM layer; and ensure that their data is processed in a fundamentally sovereign and secure manner. API pricing starting at €72 per month, with pay-as-you-go or prepaid options and scalability up to 1,000 requests per minute.
██████ ██████ █████████ ███████████ █████ █████ █████ ██████ █████ ▒▒██████ ██████ ███▒▒▒▒▒███ ▒▒███▒▒▒▒▒███ ▒▒███ ▒▒███ ▒▒███ ▒▒██████ ▒▒███ ▒███▒█████▒███ ▒███ ▒███ ▒███ ▒███ ▒███ ▒███ ▒███ ▒███▒███ ▒███ ▒███▒▒███ ▒███ ▒███████████ ▒██████████ ▒███ ▒███ ▒███ ▒███▒▒███▒███ ▒███ ▒▒▒ ▒███ ▒███▒▒▒▒▒███ ▒███▒▒▒▒▒███ ▒▒███ ███ ▒███ ▒███ ▒▒██████ ▒███ ▒███ ▒███ ▒███ ▒███ ▒███ ▒▒▒█████▒ ▒███ ▒███ ▒▒█████ █████ █████ █████ █████ █████ █████ ▒▒███ █████ █████ ▒▒█████ ▒▒▒▒▒ ▒▒▒▒▒ ▒▒▒▒▒ ▒▒▒▒▒ ▒▒▒▒▒ ▒▒▒▒▒ ▒▒▒ ▒▒▒▒▒ ▒▒▒▒▒ ▒▒▒▒▒ ▞▀▖ ▐ ▚▄ ▌ ▌▞▀▘▜▀ ▞▀▖▛▚▀▖▞▀▘ ▖ ▌▚▄▌▝▀▖▐ ▖▛▀ ▌▐ ▌▝▀▖ ▝▀ ▗▄▘▀▀ ▀ ▝▀▘▘▝ ▘▀▀
curl -X POST "https://api.marvin-systems.com/v1/lab/sessions" \-H "Authorization: Bearer <API_KEY>" \-F "files[]=@Contrat_250519.pdf" \-F 'selectedEntities=["PR","MAIL","CIE","CID","ADR"]'{"id": "s9a4xn8b34vwd8f","mode": "pseudonymization","ocrMethod": "markdown","selectedEntities": ["PR", "MAIL", "CIE", "CID", "ADR"],"pseudonymsCount": 0}
Marvin Systems automatically understands the context of each document and combines a suite of cutting-edge technical components to maximize the accuracy and versatility of data detection.
Marvin Systems automatically understands the context of each document and combines a suite of cutting-edge technical components to maximize the accuracy and versatility of data detection.
An intuitive editor allows you to efficiently adjust the data to be pseudonymized and verify the complete absence of inaccuracies before export: add fields, zones, and remove superfluous detections. For uncompromising confidentiality, the human stays in the loop.
Redaction completely purges the detected data, rather than simply placing a rectangle over it. Then, so-called ‘dynamic’ pseudonymization groups syntactic variations of the same entity under a single label, thereby preserving the consistency of the processed documents.
Who we are and what we do.
Terms of use and technical integration.
Service access and support.