AI Act · Art. 50
AI policy and transparency
Coach Experience Careers uses artificial intelligence to help you build your profile and to enable semantic search between recruiters and football professionals. This page explains what that AI does, what it does not do, and what control you keep as a user.
Active disclosure version: 1.0.0
1. What AI does on the platform
AI guides the conversational onboarding, generates a professional summary of your profile, and creates embeddings so recruiters can run natural-language semantic searches.
2. What AI does not do
No automated selection
The platform does not select candidates or decide hires. It only ranks profiles by semantic similarity to the recruiter query.
No personal evaluation
The visible percentage does not measure human quality, potential, or job suitability. It only expresses semantic proximity between a search and the self-declared profile content.
3. Regulatory classification
Provider assessment
The system is documented as Limited Risk under Art. 6(4) of Regulation (EU) 2024/1689. The main applicable obligation is transparency under Art. 50.
LIMITED_RISK
4. Technical AI flow
- Gemini guides the onboarding chat and structures your answers into profile data.
- The platform generates an ai_summary to make your experience readable and reusable.
- The summary and selected key fields feed a 768-dimensional semantic embedding.
- When a recruiter searches for talent, the query is vectorized and compared against published embeddings.
5. Data analyzed by AI
- Professional summary generated from your answers.
- Experience, competition level and availability declared during onboarding.
- Career history, last club and professional track record that you include in your profile.
- Languages and proficiency level when provided in your profile.
- Nationality and professional context declared by you.
6. Transparency and user control
Advance notice
Before using the AI onboarding we show a specific disclosure and store the accepted version.
Percentage explained
The visible percentage is presented as semantic similarity and is always clarified as not being a professional evaluation.
Real opt-out
You can disable your visibility in semantic search. If you do, your profile stops being returned by the SQL matching query.
Public page
This public policy summarizes the regulatory classification, analyzed data and competent authorities.
7. Identified risks
The main identified risks are implicit bias in the base model, undocumented technical changes, and possible future regulatory reclassification. To mitigate them, the project keeps a living compliance document that must be updated whenever models, embeddings, scoring, or semantic search logic change.
8. Documentation record
The active AI disclosure version is 1.0.0.
The project compliance master document lives at docs/ai-act-compliance.md and is updated whenever models, embeddings, processed data, scoring UI, or transparency copy change.
9. Competent authorities