Security, Privacy & AI Transparency

Version 1.0 March 2026 vetti.app hello@vetti.app

This document describes how Vetti handles candidate data, what technical and procedural safeguards are in place, and how the AI scoring system operates. It is intended for recruitment agencies, their clients, and any party conducting due diligence on Vetti's data practices.

1. Infrastructure & Data Residency

Vetti is hosted on Railway.app, which operates on infrastructure provided by Google Cloud Platform (GCP). All data is stored and processed within the European Union.

Cloud Provider

Railway.app — infrastructure backed by GCP data centres in the EU region (Frankfurt / Belgium).

Data Residency

All database records, screening sessions, and candidate evaluations are stored exclusively within EU borders. No data is transferred to third-party servers outside the EU for storage.

Database

PostgreSQL, managed via Supabase. User accounts and session metadata are stored with standard access controls — each user's data is accessible only to that user's authenticated session.

2. Encryption & Access Control

Data at Rest

Database volumes are encrypted at rest by Railway's underlying infrastructure (GCP), which uses AES-256 by default.

Data in Transit

All communication between users, the Vetti application, and third-party APIs is encrypted via HTTPS (TLS). Connections to the application are HTTPS-only; unencrypted HTTP is not accepted.

Authentication

User authentication is managed by Supabase Auth, which handles password hashing, session tokens, and secure cookie management. Email-based login with session expiry is enforced.

3. GDPR Compliance

Vetti processes candidate CV data on behalf of recruitment agencies (data processors). The following rights are implemented directly in the application interface.

Right to Erasure (Article 17)

A permanent deletion function is available in the application. When invoked, all records associated with a screening session — including CV content, AI evaluations, and scores — are permanently removed from the database. This action is irreversible.

Data Minimisation & Retention

Vetti enforces a 90-day automatic deletion policy on inactive screening sessions. Sessions not accessed within 90 days are purged from the database without manual intervention.

Data Export (Recruiter-Side)

Recruiters can export screening results to Excel and download a full audit report (PDF/text) for each session. These exports contain the AI evaluations and decision log for that session and are intended to support subject access requests or client reporting.

Purpose Limitation

Candidate CV data is used solely for the purpose of generating a structured evaluation score. Data is not shared with third parties for any secondary purpose, including advertising or model training.

Note on Data Controller responsibilities: Vetti operates as a data processor. The recruitment agency using Vetti is the data controller and is responsible for obtaining appropriate consent or legitimate interest basis for processing candidate CVs, and for responding to subject access requests from candidates.

4. EU AI Act Alignment (Article 13 & 14)

AI-assisted recruitment screening is classified as a high-risk use case under the EU AI Act. Vetti is designed to meet the transparency and human oversight requirements that apply to high-risk AI systems.

Transparency & Explainability

Every candidate evaluation includes a score breakdown showing points awarded across five categories: CV structure & ATS compatibility, content quality, keywords & skills, presentation, and job relevance. The weighting for each category is disclosed to the recruiter. Scores are not a black box.

Human Oversight (Article 14)

Vetti produces ranked shortlists and advisory scores only. No candidate is automatically rejected or progressed. All hiring decisions — including which candidates to interview, reject, or flag — require explicit action by a human recruiter. The AI is a decision-support tool, not a decision-maker.

Audit Trail

Each screening session generates a downloadable audit report containing the scoring methodology, full candidate decision log, session statistics, and a sign-off section for the responsible recruiter. This report is designed to support regulatory review or client governance requirements.

Bias Safeguards

The AI evaluates CV content only — structure, language quality, relevance to the job description, and keywords. It does not receive or evaluate candidate names, photographs, age, nationality, gender, or any other protected characteristic. Protected characteristics are not scoring inputs.

5. Sub-Processor Disclosure

Vetti uses the following third-party services to deliver its functionality. Each sub-processor is selected for its EU compliance posture.

Provider Purpose Data Shared Compliance
Anthropic AI inference (CV evaluation) CV text content only, no personal identifiers required SOC 2 Type II, zero data retention API*
Supabase Authentication & database User account data, session records SOC 2, GDPR compliant, EU hosting available
Railway Application hosting Application runtime only GCP infrastructure, EU region
SendGrid Transactional email (Pro plan) Recipient email addresses SOC 2, GDPR compliant
* Anthropic's API operates under a zero data retention policy for API customers. CV content submitted for evaluation is processed in memory and is not stored, logged, or used for model training by Anthropic. This is contractually enforced via Anthropic's API terms of service.

6. Contact & Data Requests

For data protection enquiries, subject access requests, or questions about this document, contact:

Data Protection Contact

hello@vetti.app
Vetti — vetti.app

We aim to respond to all data protection enquiries within 5 business days.