How it works
in four steps.

AI Framework Discovery

Connector: Git

Clones your repository, generates a Software Bill of Materials (SBOM), and cross-references every package against a curated risk map of AI/ML frameworks.

• Detects TensorFlow, PyTorch, scikit-learn, LangChain, OpenAI SDK
• Maps findings to EU AI Act Article 6 risk classification
• Auto-creates AI System Cards for undocumented AI usage

PII & Special Category Data Discovery

Connector: PostgreSQL

Samples text columns from your database and runs local NLP models to detect personal data plus GDPR Article 9 special category data (health, religion, politics) via GLiNER zero-shot NER. All analysis runs on our European infrastructure.

• Finds names, emails, phone numbers, national IDs, addresses
• Detects health data, religious beliefs, political opinions, trade union membership via GLiNER
• Redacts all samples before storage (raw PII never persisted)
• Auto-populates GDPR Record of Processing Activities

Multi-Source PII & Secrets

Connectors: S3, MySQL, MongoDB, Redis, Slack, Google Drive

Scans multiple data sources for hardcoded secrets, API keys, and personal data patterns using pattern-matched detection.

• Detects AWS keys, tokens, passwords, credit card numbers
• Covers S3 buckets, databases, Redis, Slack, Google Drive
• Pattern-matched severity scoring (critical/high/medium)

Cloud Security Posture

Connectors: AWS, GCP, Azure

Evaluates your cloud configuration against CIS benchmarks and DORA ICT resilience requirements.

• Checks IAM policies, encryption, logging, network exposure
• Maps findings to DORA Articles 6-15 and ISO 27001 Annex A
• Flags misconfigurations that could cause compliance gaps

Model Artifact Security

Connector: AI System Artifact

Scans AI model files for serialization attacks and unsafe patterns that could compromise model integrity.

• Scans pickle, H5, SavedModel, PyTorch, ONNX formats
• Detects model-serialization attacks (EU AI Act AI Act Art. 15 robustness)
• Comprehensive coverage of model file vulnerabilities

LLM Red-Teaming

Connectors: LLM Endpoint, OpenAI, Anthropic

Probes your language models with adversarial attacks and runs compliance benchmarks.

• Tests for jailbreaks, prompt injection, encoding bypasses
• Runs truthfulness and harmful content benchmarks
• No external AI provider is ever called during testing

MDM Endpoint Audit

Connectors: Microsoft Intune, Jamf Pro, Kandji

Polls your MDM API to verify employee device compliance, disk encryption, OS patching, and passcode policy. Read-only, no agents installed.

• Checks BitLocker/FileVault disk encryption status
• Verifies OS version is current and patched
• Confirms passcode/screen lock policy is enforced
• Privacy-first: reads MDM policies via API only, never installs agents on devices

AI Act Compliance Benchmark

Connectors: LLM Endpoint, OpenAI, Anthropic

Runs a curated subset of ETH Zurich's COMPL-AI benchmark suite against your chat LLM endpoint, testing cyberattack resilience, fairness, toxicity, and disclosure-of-AI tasks.

• Tests cyberattack resilience and adversarial robustness
• Measures fairness and bias across protected attributes
• Evaluates toxicity and harmful content generation
• Verifies AI disclosure and transparency compliance

EU-Specific PII Patterns

Connector: Filesystem (pipeline-compatible)

Supplementary regex scan for EU-country-specific identifiers that general-purpose scanners often miss, BSN (Netherlands), Codice Fiscale (Italy), NIR (Italy), IBAN, and more.

• Detects BSN, Codice Fiscale, NIR, and other national IDs
• Covers IBAN patterns across EU countries
• Pattern-matched severity scoring (critical/high/medium)

Vulnerability Scanner

Connectors: Git, Docker Registry

General-purpose vulnerability scanner via Trivy. Checks git repositories and container registries for unpatched CVEs, misconfigurations, and secret leaks.

• Scans for unpatched CVEs in dependencies and containers
• Detects misconfigurations and secret leaks
• Flags critical and high-severity vulnerabilities per DORA Article 9 patching SLAs

Identity & Access Audit

Connectors: Google Workspace, GitHub, Microsoft 365

Polls your Identity Provider to verify MFA enforcement, detect dormant accounts, and ensure former employee access is revoked within SLAs.

• Verifies MFA enforcement across all accounts
• Detects dormant and orphaned accounts
• Checks former employee access revocation timelines
• Maps to ISO 27001 A.9.2.6 and DORA Article 9

Shadow AI Detection

When our scanner detects AI frameworks in a repository you registered as a standard IT system, it automatically promotes the system to an AI System and sends you an alert. This catches undocumented AI usage before regulators do.