Live Demo
Enter a prompt and validate it against the EU AI Act. Results appear below.
Recent validations
Loading…
Solution & Performance
Core compliance checks and PII stripping run on a dedicated Rust engine. Deterministic rule-sets deliver <50 ms decisions on the fast path, with PII stripping in the microsecond range so the privacy layer adds no perceptible delay.
P99 < 1 ms · Zero-allocation · Rule-based
When nuance is required, the same pipeline routes to an LLM for deep analysis. Model-agnostic: use Gemini, Claude, or local models via LLM_PROVIDER. Intelligent routing minimizes LLM calls without sacrificing accuracy—validated against 36+ EU AI Act scenarios.
2–4 s when needed · ALLOW/DENY/WARN · Article refs
Workflow
graph TD
User((User/App)) -->|Prompt| Gatekeeper[API Gatekeeper]
Gatekeeper -->|Step 1| PII[PII Stripper - Rust Engine]
PII -->|Masked Prompt| Logic{Decision Logic}
Logic -->|Deterministic| Rust[Fast-path Compliance - Rust]
Logic -->|Nuanced| LLM[Deep Analysis - LLM]
Rust -->|ALLOW/DENY| Result
LLM -->|ALLOW/DENY/WARN| Result
Result --> Audit[Audit Log - GDPR Proof]
Result --> User Enforcement vs Surveillance
ComplianceCode.eu acts as an essential gatekeeper, ensuring adherence to the EU AI Act without infringing on privacy or data sovereignty. We don't watch your users; we protect your data.
Proactive Protection
Block prohibited practices before they reach your AI stack. Real-time guardrails, not after-the-fact monitoring.
Data Privacy
PII stripped before processing. Only masked prompts and audit metadata are stored—GDPR-proof by design.
Legal Adherence
EU AI Act Article 5 and high-risk provisions. Clear ALLOW/DENY/WARNING with article references for compliance teams.
Live API Preview
See how a /gatekeeper request and response look.
POST /api/v1/gatekeeper
Content-Type: application/json
x-api-key: <your-api-key>
{
"prompt": "Summarize this contract for our legal team.",
"context": { "department": "legal" }
}{
"decision": "ALLOW",
"reason": "No prohibited practices identified.",
"article_ref": null,
"audit_id": "550e8400-e29b-41d4-a716-446655440000"
}