AI Transparency

Orvensa includes AI-assisted candidate evaluation features to help recruiters review applications more efficiently and consistently.

The AI functionality is designed as decision support. It does not replace human judgment and is not intended to make final hiring decisions autonomously.

The recruiting organization decides whether AI-assisted screening is enabled in its workspace and remains responsible for how the platform is used in its recruitment process.

When enabled by the recruiting organization, Orvensa may use AI to:

• analyze CV-derived text against job-related context,
• generate structured candidate summaries,
• identify apparent strengths and risks from the provided material,
• generate interview-question suggestions,
• produce score-related outputs, fit indicators, and confidence-related values where configured,
• format extracted CV content into a cleaner review view for users.

Orvensa does not position its AI as an autonomous hiring decision-maker.

• It does not make the final hiring decision.
• It does not autonomously accept or reject a candidate as the final legal outcome.
• It does not remove the need for a human reviewer.
• It is not intended to infer special category data for recruitment purposes.
• It is not intended to be used as a sole basis for a legally significant employment decision.

The AI-assisted evaluation flow may use the following categories of input:

Orvensa applies a minimization step before AI submission by redacting obvious direct identifiers such as email addresses and phone-like numbers from CV text. This is a minimization control, not a guarantee of full anonymization or de-identification.

At the current implementation stage, extracted raw CV text may be stored within evaluation metadata in the platform for workflow and display purposes. For that reason, Orvensa does not describe its current architecture as one in which CV text is never persisted.

Depending on the configuration, Orvensa may produce outputs such as:

• summary text,
• strengths,
• risks,
• interview questions,
• score and subscore outputs,
• fit bucket or similar categorization,
• confidence-related values,
• formatted explanation and review payloads for user visibility.

These outputs are generated from the material provided to the system and should be treated as advisory. They may reflect missing, incomplete, ambiguous, or poorly extracted input data.

Orvensa is built around a human-in-the-loop model.

• The platform supports human review as the decision-making layer.
• Final decision fields are recorded separately from model outputs.
• Evaluations can be contested and reopened for human review.
• The platform can require human review by default.
• A recruiting organization can disable AI screening or AI evaluations at organization level.

This separation is intended to reduce over-reliance on automated outputs and to keep accountability with the human decision-maker.

Orvensa includes safeguards relevant to AI-assisted recruitment workflows:

• organization-level kill switches for AI screening and AI evaluations,
• lawful-basis and privacy-notice checks before AI processing,
• blocking of AI processing where an applicant has objected to automation,
• blocking of AI processing where processing is restricted,
• contestability and manual-review escalation,
• audit logging of relevant actions,
• retention settings and deletion routines for applicant records,
• role-based access controls and organization scoping.

AI-assisted recruitment tools have limitations and should be used with care.

• Outputs may be influenced by incomplete or low-quality source documents.
• Formatting issues, extraction errors, and ambiguous CV content can affect results.
• Model outputs may be imperfect, overconfident, or incomplete.
• AI output quality may vary across roles, sectors, and CV styles.
• Recruiters should not treat generated content as a substitute for professional judgment.

Recruiting organizations should ensure appropriate reviewer training, fair-hiring procedures, and internal oversight when using AI-assisted tools in recruitment.

Orvensa includes product features intended to support controller-led compliance with applicant rights, including:

• access/export workflows,
• rectification workflows,
• erasure workflows,
• restriction of processing,
• objection to automation,
• human review and contestation of evaluation outcomes.

In most recruitment scenarios, the recruiting organization is the controller and remains the primary contact point for candidate rights requests and complaints relating to the hiring process.

Kobenski BV provides the technical platform and related safeguards. The recruiting organization remains responsible for:

• choosing whether and how to use AI-assisted features,
• ensuring a valid legal basis for applicant-data processing,
• providing a compliant candidate privacy notice,
• ensuring human oversight in actual hiring decisions,
• using the platform in a lawful, fair, and non-discriminatory manner.

Orvensa should be understood as a support tool within a broader recruitment governance process, not as a replacement for human, legal, or compliance judgment.