Skip to content
GUIDSE
GUIDSE Platform · 02 / DATA PROTECTION

Control what the model is allowed to see.

Privacy Wall applies configured masking rules to sensitive fields before information reaches an external AI provider, while keeping authorized workflows able to use the resulting output.

Map a critical processPlatform overview
The protection layer

Protect sensitive fields before the AI step.

Documents may contain client names, bank details, internal prices and other confidential data. Privacy Wall is designed to identify and mask fields according to defined rules, so a process can use AI without sending every original value to the model.

Privacy is a processing decision, not just a policy statement.

Illustrative field treatment

The task remains. The sensitive value does not need to travel with it.

Original process input
Invoice reference: INV-2048
Client: Example Company
Contact: Alex Example
IBAN: BE00 0000 0000 0000
Privacy WallConfigured masking rules
AI-facing input after masking
Invoice reference: INV-2048
Client: [CLIENT_01]
Contact: [PERSON_01]
IBAN: [BANK_ACCOUNT_01]

Fictional example showing a configured masking policy. Actual handling depends on policies, supported fields and integration configuration.

The protection sequence

Policy first. Provider second.

  1. 01

    Identify

    Define fields and patterns that need protection, including rule matching by field name or path.

  2. 02

    Mask

    Replace sensitive values before an AI-provider request where the configured policy applies.

  3. 03

    Resolve

    Retain protected values in a secure vault and restore only where an authorized step requires them.

  4. 04

    Audit

    Connect access decisions to the platform’s broader governance and execution evidence.

Rules, vault and control

Masking needs a policy and an authorized path back.

Protection is only credible when a team can specify what is sensitive and where a result may be re-associated with the original data. Privacy Wall separates the AI-visible representation from values retained for authorized use.

  • Field-level masking rules
  • Secure handling of original values
  • Controlled restoration for authorized steps
  • Provider-independent policy design
Questions

What to know about Privacy Wall.

Does this mean an AI provider never receives any business data?

No. A provider can receive the non-masked content needed for a particular task. Privacy Wall is intended to keep fields covered by configured masking policies out of that request.

Does using Privacy Wall automatically make a process GDPR compliant?

No. Compliance depends on the complete processing context, lawful basis, configuration, contracts, access controls and organizational practices.

THE NEXT CONVERSATION

Bring one real process. We will map it with you.

See where this layer belongs in your operational workflow, and what still needs a person.

Map a critical processNext layer: Orchestration