How Penroll handles hiring data and AI

How data moves through Penroll

1. A hiring team creates a role, and candidates submit contact details, answers, and a CV.
2. Job, account, and candidate records are stored in Penroll’s Supabase project in the EU region. CVs and offer documents use its object storage.
3. Role requirements, CV text, or offer details are sent to the listed AI provider when a customer requests a draft or ranking. The result returns to the Penroll workspace.
4. Other providers support application hosting, security, billing, and transactional email. Their current purposes and data categories are listed in the subprocessor register.

EU-hosted customer records

Penroll’s primary database and object storage for customer and candidate records are hosted in Supabase’s EU region in Frankfurt. Some subprocessors operate outside the EU. The DPA describes Standard Contractual Clauses, prompt redaction, and data minimisation for those transfers; this page does not claim that every processing step occurs only inside the EU.

Application and CV retention

Unsuccessful-candidate application data is retained for up to 24 months. If a candidate explicitly opts in to future-role consideration, the period can extend to 4 years. Each application has a deletion deadline that the scheduled cleanup checks. Candidates can request earlier erasure through the erasure form.

Sub-processors

Penroll uses a limited set of providers for database and storage, application hosting, AI inference, billing, transactional email, and edge security. The maintained subprocessor register names each provider and states its purpose, data category, and processing location.

AI processing and training

Penroll currently uses Anthropic for job-description generation, CV ranking, screening questions, and offer-letter drafts. Under the business API arrangement described in the Privacy Policy and subprocessor register, those inputs are not used to train the provider’s models. The provider still processes the prompt to return the requested output, so Penroll applies data-minimisation and transfer safeguards.

A person makes the hiring decision

CV ranking produces evidence, scores, and Interview, Maybe, or Pass recommendations. Penroll does not automatically reject a candidate from that recommendation. The hiring team reviews the source CV, can override the recommendation, runs interviews, and makes the final decision.

What Penroll does not decide

AI output is a draft or recommendation, not legal advice or an employment decision. Penroll does not replace human interviews, reference or background checks, payroll, onboarding, local employment advice, or the employer’s responsibility for a hire. It manages an inbound hiring workflow; it does not act as a recruitment agency or run outbound candidate search.