Your DataStays Yours.Security by Design.
MiruIQ is built for sensitive, regulated, and high-trust document workflows. Security is not an add-on; it is enforced through explicit data lifecycles, controlled processing, and clear ownership boundaries. Run it self-hosted on your own infrastructure (on-premise deployment with local AI models) or on Swiss data hosting operated by MiruIQ.
Deployment Models & Security Guarantees
MiruIQ offers two deployment models. Both provide the same functional capabilities and pipeline logic, but differ in infrastructure ownership, trust boundaries, and operational responsibility.
As-a-Service (Managed)
MiruIQ operates and manages the platform on behalf of the customer.
Self-Hosted / On-Prem
MiruIQ is deployed entirely inside the customer's environment: on-premise deployment where no data leaves your infrastructure.
Core Security Principles
Whether Swiss-hosted or as on-premise document processing on your own hardware, MiruIQ enforces the same architectural security principles. This architecture makes GDPR and revDSG compliance a property of the system rather than a promise.
Processing, Not Storage
MiruIQ processes documents to produce structured data and validation results. It does not act as a document repository unless explicitly configured.
Explicit Control
Every form of retention, encryption, output, or reprocessing is explicitly defined by the user. Nothing is retained implicitly.
Least Persistence
Documents and intermediate data exist only as long as required to complete the configured pipeline logic.
Explicit Data Lifecycle
Documents are processed, not stored.
When a document enters a MiruIQ pipeline, it is treated as a processing unit, not as persistent content.
By default, documents flow through classifiers, extractors, and verification modules in memory and are removed immediately once their role in the pipeline is complete.
However, certain pipeline steps explicitly require temporary state. MiruIQ supports this in a controlled, transparent, and time-bound manner.
In Transit
Stateless Processing
The document is actively processed and exists only for the duration of execution.
Temporarily Retained
Stateful Processing
Extracted data or intermediate results are retained only when a module requires group-level context.
Delivered
Output Completion
Files, extracted data, or validation results are written to configured sinks.
Terminated
Automatic Deletion
Once processing or validation is complete, all intermediate state is immediately deleted.
Stateless by Default
Most pipelines remain fully stateless. Documents flow through the pipeline in memory and are discarded immediately after processing.
Stateful Only When Needed
State is introduced only when a processor explicitly requires it, and always in a controlled, transparent, time-bound manner.
State Only When Required
Example: Cross-Document Validation
Cross-document validation is a prime example of stateful processing in MiruIQ. Some workflows require verifying consistency across a set of related documents that together form a single business case.
How Grouping Works
Validation Behavior
Once a document group is complete:
Controlled State Retention
Minimal Retention
State is retained only when a processor explicitly requires it. Most pipelines remain fully stateless with no intermediate data persisted.
Immediate Deletion
As soon as state has served its purpose, such as completing cross-document validation, it is immediately and permanently deleted.
Always Encrypted
Any temporarily retained state is encrypted at rest using pipeline-specific keys. State is never stored in plaintext.
Encryption & Secure Data Handling
Encryption for processing, state, and delivery, with clear ownership boundaries.
MiruIQ applies encryption at multiple stages of a pipeline depending on the lifecycle of the data.
A clear distinction is made between internal encryption (which protects data while it is processed) and output encryption (which protects data after it leaves the pipeline). These follow different ownership and control models.
Internal Encryption (Pipeline & Temporary State)
Internal encryption protects:
Internal keys are used only inside MiruIQ and are never shared with downstream systems.
Automatic Pipeline Encryption (Default)
One dedicated encryption key per pipeline. Created and rotated automatically. Never exposed to users.
Customer-Managed Keys (Inside MiruIQ)
Keys created and managed within MiruIQ. Controlled lifecycles and scoped usage.
External KMS Integration
Keys fully owned and managed by the customer. Revocation intentionally stops dependent pipeline steps.
Output Encryption (Data Leaving the Pipeline)
When data leaves MiruIQ (as files, extracted data, or validation results), encryption follows a different model.
Output encryption supports both MiruIQ-managed secrets and external KMS integration. Choose the model that fits your security requirements.
Encrypt specific sensitive fields (SSN, account numbers, PII) while leaving other data readable. Ideal for partial data protection where some fields need selective access control.
Encrypt the entire output file or data payload. Ideal for high-security scenarios where all extracted data should be protected in transit and at rest.
Customer-Provided Keys (Secrets in MiruIQ)
Store encryption keys securely within MiruIQ as managed secrets. Keys are used for output encryption but never exposed in logs or API responses.
External KMS Integration
Integrate with your existing Key Management Service (AWS KMS, Azure Key Vault, HashiCorp Vault). Keys remain fully under your control and are never stored in MiruIQ.
Backup, Retention, Reprocessing
Optional pipeline backups
MiruIQ supports explicit, user-configured backup retention to enable controlled reprocessing.
Retention Strategies
Examples include:
Retention applies only to backup snapshots, not live pipeline state.
Reprocessing Behavior
When reprocessing is triggered:
Reprocessing is deterministic and auditable.
Designed for Auditability & Control
MiruIQ enables automation without sacrificing control, trust, or transparency.
GDPR / DSGVO (EU 2016/679)
Full compliance – privacy-by-design & privacy-by-default
EU ePrivacy Directive
Communications metadata & tracking protection (national implementations)
EU AI Act (alignment)
Transparency, auditability, logging, and human-in-the-loop support
EU Data Act (alignment)
Data portability, customer data ownership, and avoidance of vendor lock-in
EU Data Governance Act (principles)
Purpose limitation, data sovereignty, and controlled data sharing
NIS2 Directive (security alignment)
Risk management, access control, and incident-response readiness
Security Questions
The three questions every regulated buyer asks about on-premise document processing.
Yes. MiruIQ deploys on your own hardware including the AI models: local LLMs on dedicated GPUs, no cloud dependency, air-gapped if required. On-premise operation is MiruIQ's founding constraint, not a feature added later.
Yes. MiruIQ runs as a managed service on Swiss infrastructure: data residency in Switzerland, GDPR and revDSG compliance by architecture, suitable for banking- and insurance-grade workloads.
Choose a platform where the models run where your data lives. MiruIQ uses local AI models on Swiss or on-premise infrastructure. Documents, embeddings and extraction results never touch a US hyperscaler, which makes GDPR-compliant document AI a system property rather than a contractual promise.
Security Without Compromise
Build document automation workflows with confidence. Your data stays protected at every step.