How our AI matching works

1. 1

   Suggestions, not decisions

   AI only ranks likely fits; every decision is taken by a person.

   Read the full clause: Suggestions, not decisions

   MatchPRO never hires, rejects, or filters anyone out automatically. The AI only ranks likely fits and shows them to both sides. Every consequential step — revealing identity, starting a chat, making an offer — is taken by a person, and identity is disclosed only with the candidate’s explicit consent.

2. 2

   How a match is computed

   Filters for occupation and work mode, closeness of meaning, and a factor breakdown — in any language.

   Read the full clause: How a match is computed

   Matching is a hybrid of three layers: deterministic compatibility filters (occupation, work mode, relocation), semantic similarity between the meaning of a profile and a vacancy (vector embeddings — so a “line cook” can match a “chef de partie” posting), and a transparent factor breakdown computed from skills, experience, salary and location. Across languages the work is done by the occupation and skill identifiers, which are the same concept whatever language it is written in — a nurse’s profile in Greek and a nursing vacancy in English share an identifier even where their wording shares nothing.

3. 3

   Why you see a match (explainability)

   Each match shows its factors as percentages with a short reason; there is no hidden score.

   Read the full clause: Why you see a match (explainability)

   Every match shows its factor breakdown: how well skills overlap, whether experience fits, how salary expectations align, and whether locations are compatible — each as a percentage with a short explanation. What you see is exactly what the system computed; there is no hidden score.

4. 4

   What data the AI uses — and never sees

   Professional data only; contact details are removed before any text reaches a model.

   Read the full clause: What data the AI uses — and never sees

   The AI processes professional data only: skills, occupations, experience, industry, salary expectation band, location and work-mode preferences, and the text of your CV or vacancy. Before any text reaches an AI model, contact details are removed — email, phone, links and @handles. Names and employer names are not stripped: the AI provider receives the rest of your CV or vacancy text as written, to build your profile — other candidates and employers only ever see a profile with contact details removed. The platform does not collect or infer ethnicity, religion, health, age or gender.

5. 5

   Paid features never change your score

   Seeker Boost raises visibility in the catalogue, never a match score.

   Read the full clause: Paid features never change your score

   Seeker Boost raises a profile’s visibility position in the catalog — it never changes a match score or factor. Employers cannot pay to rank higher in matching either.

6. 6

   Your controls

   Export or erase your data at any time; a person reviews a match you dispute.

   Read the full clause: Your controls

   You consent to AI processing at sign-up (the consent version is recorded). You can export everything we hold about you (Settings → Privacy) and delete your account with full data erasure at any time. If you believe a match or a factor reading is wrong, contact support — disputes are reviewed by a human.

7. 7

   Compliance posture

   High-risk under the EU AI Act: explanations and human review are in place, the formal documentation is in progress.

   Read the full clause: Compliance posture

   Recruitment AI is regulated as high-risk under the EU AI Act. In place today: an explanation for every match shown to you, and human review of any match you contest. The formal technical documentation and the EU database registration are still being prepared. Pay ranges on vacancies follow the EU Pay Transparency Directive, and we never ask candidates about their salary history.

Last updated: 2026-06-09