In accordance with the AI Act (EU) 2024/1689 — Art. 13 and Art. 14
Informational translation — the Romanian version prevails
This is an informational translation provided for convenience. The Romanian version of this document is the sole legally binding version. In case of any discrepancy between the Romanian and English versions, the Romanian version shall prevail.
This document is governed by Romanian law. Any disputes arising from or related to it are subject to the exclusive jurisdiction of the Romanian courts.
AI Transparency
We explain how artificial intelligence works within the JobGrade platform: what data we use, how we make decisions and what rights you have.
Last updated: 8 April 2026
1.What is JobGrade and how we use Artificial Intelligence
JobGrade is a job evaluation and professional development platform, operated by Psihobusiness Consulting SRL (CIF: RO15790994).
The platform uses Large Language Models, provided by Anthropic (the Claude model), to assist evaluation and guidance processes.
B2B — for companies
We assist the job evaluation process: we structure information, comparatively analyse roles and propose scores based on objective criteria. The final decision always rests with the human evaluation committee.
B2C — for individuals
We guide personal and professional development dialogues through specialised agents. Each agent adapts the conversation to the user's profile and needs.
What our AI does NOT do
- Does not make autonomous decisions regarding hiring, dismissal or remuneration
- Does not replace human specialists (psychologists, counsellors, legal advisors)
- Does not operate without human oversight
2.How AI works in the evaluation process (B2B)
Job evaluation in JobGrade is based on 6 objective criteria. These assess the role, not the person holding it — ensuring gender neutrality and compliance with EU Directive 2023/970 on pay transparency.
Rolul AI-ului
- Structures information from job descriptions
- Analyses and compares positions based on criteria
- Proposes initial scores — as a starting point
- Generates comparative reports
Rolul oamenilor
- The evaluation committee validates each score
- Minimum 3 consensus stages before finalisation
- Any evaluation can be challenged and re-evaluated
3.How AI works in personal development (B2C)
The B2C module offers guided dialogues with specialised AI agents. Each agent has a precise role and operates within the boundaries of a validated methodology.
What AI does
- Adapts the conversation to the user's profile and pace
- Asks calibrated questions that aid reflection and the process of self-discovery
- Offers perspectives based on validated psychological frameworks
- Generates progress reports
What AI does NOT do
- Does not diagnose — it is not a clinical tool
- Does not treat — it does not provide therapy
- Does not manipulate — guides through questions, not directives
SafetyMonitor — automatic protection
The platform includes a monitoring system with 4 alert levels that protects the user in sensitive situations.
AI operates in the rational and reflective domain. For aspects relating to deep lived experience, the platform recommends qualified human counsellors.
4.Human oversight (Art. 14 AI Act)
Echipa de supraveghere
- Specialised staff accredited by the College of Psychologists of Romania (CPR)
- Specialisations: occupational, transport and services psychology
- Psychoanalytic training (relevant to B2C interactions)
- Certificate of independent practice — grants the right to practise as a psychologist independently
In B2B processes
- Specialised staff supervise the evaluation methodology
- Evaluation committees are composed exclusively of people
- AI-ul propune, oamenii decid
- Complete audit log for each session
In B2C processes
- SafetyMonitor automatically escalates to specialised staff when distress is detected
- Specialised staff can intervene at any time
- Sensitive reports are validated by specialised staff
Principiul fundamental: No decision with an impact on a person is made exclusively by AI. There is always a human who validates, approves or intervenes.
5.Data we use
Surse de date
- Information provided directly by the user: questionnaire responses, dialogue content, uploaded documents (CV, job descriptions)
- Browsing and interaction data within the platform (for experience improvement)
What we do NOT do with the data
- We do not train AI models with client data. We use pre-trained models (Claude, provided by Anthropic). Data is processed, not used for training.
- We do not collect data from external sources without the explicit consent of the user
- We do not sell or share data with third parties for marketing purposes
Protection measures
- B2C: mandatory pseudonym — real identity is not accessible to AI agents
- Two layers of separation — identity data and interaction data stored in separate databases
- Encryption in transit (TLS) and at rest
Retention periods
- Date evaluare B2B: durata contractului + 5 ani
- B2C interaction data: 3 years from last activity
- Inactive B2C account: notification at 24 months, deletion at 36
- Date audit: 10 ani
6.Drepturile tale
Drepturi specifice AI Act
The right to know you are interacting with an AI
All agents on the platform clearly identify themselves as AI assistants. We do not simulate human interaction.
The right to request an explanation
You may ask at any time for an explanation of how the AI reached a recommendation or proposed score.
Dreptul de a contesta o evaluare
Every decision is reviewed by a human team. AI does not have the final word.
Right to human intervention
At any point during your interaction with the platform, you may request to speak directly with a specialist from our team.
Drepturi GDPR (Art. 15-17 RGPD)
Dreptul de acces
Request a copy of all data we hold about you
Dreptul la rectificare
Correct any inaccurate data
Right to erasure
Request the deletion of your account and all associated data
Dreptul la portabilitate
Export your data in a structured format
How to exercise your rights
- Export date: from your account, section "My Data" (or via API:
GET /api/v1/b2c/my-data) - Account deletion: from settings, section "Delete account" (or via API:
DELETE /api/v1/b2c/account). Grace period: 30 days. - Other requests: contact@jobgrade.ro
7.AI limitations
We believe in transparency. That is why we openly tell you what our AI cannot do.
AI can be wrong
Recommendations and proposed scores are informed suggestions, not certainties. That is why every process includes human validation.
AI is not a substitute for a specialist
It is not a psychologist, therapist, legal adviser, or medical professional. Where a situation calls for professional expertise, we recommend consulting an accredited specialist.
AI has no lived experience
It guides from accumulated knowledge, not personal experience. Knowledge without experience has its limits — and we acknowledge them.
AI works from the information it receives
The quality of recommendations depends on the quality and completeness of the information provided.
AI-ul nu poate garanta rezultate
The platform offers tools and guidance. Outcomes depend on how they are used.
In crisis situations
If you are in a difficult situation or need immediate help:
- Hopeline: 0800 801 200 (gratuit, 24/7)
- Telefonul Sufletului: 116 123 (gratuit)
- Emergencies: 112
8.How to contact us
Psihobusiness Consulting SRL
- contact@jobgrade.ro
- CIF: RO15790994
Data Protection Officer (DPO)
Autoritate GDPR (ANSPDCP)
This page is updated whenever changes occur in the way AI operates on the JobGrade platform. Document drafted in accordance with the requirements of Regulation (EU) 2024/1689 on Artificial Intelligence (AI Act), Art. 13 (Transparency and provision of information) and Art. 14 (Human oversight).