AI is used at specific, clearly defined stages of the platform, each with its own level of control.
Our scientific approach
Science as the foundation. AI as a controlled accelerator.
Technology alone cannot make an HR decision reliable. It must be underpinned by robust methods, clearly defined criteria, and results that users can understand.
Our psychometric assessments are based on validated methods, explicit criteria, and interpretable results. AI enriches competency frameworks and supports analysis, without replacing HR decision-making.
What we assess
Key Predict brings together several complementary dimensions:
- professional personality;
- cognitive abilities;
- motivations and values;
- behavioural and cognitive competencies;
- potential for development;
- suitability for a role, team or context.
Measured, contextualised
Iinterpretable competencies
Scientific assessments
Our assessments measure the aptitudes, personality, motivations, and competencies that influence professional success. Each tool is built on a rigorous methodology and has undergone scientific validation studies.
A structured job context
Results are compared against the competencies expected for a role, a function, or a project. This context is based on more than 1,200 frameworks validated by our psychologists, which users can consult and adjust.
Interpretable matching
Clear indicators highlight matches, differentiating competencies, gaps, and points worth exploring further in interviews or development plans. The job context is based on structured, validated frameworks. It is not freely inferred by generative AI.
Where AI is used in Key Predict
Based on a job description or an HR project, AI can suggest relevant roles, functions, and competencies. These suggestions provide a starting point. Users can amend the reference roles and job families, add or remove competencies, adjust their level of importance, or switch to a fully manual approach. This assistance is optional. A model can be created without using AI at all, by selecting a standard role, a job family, a combination of both, or by choosing competencies directly.
AI also helps enrich and update the competency models associated with our standard roles and job positions, using structured sources, international occupational frameworks such as O*NET and ESCO, and Key Predict’s scientific expertise. This work is carried out upstream. The optimised profiles are reviewed and validated by our psychologists before they are made available in the platform. It does not involve freely interpreting an individual candidate at the point of decision.
Talent Brain, our natural language processing technology (NLP, not generative AI), is integrated into Smart Interview, our structured, pre-recorded video interview solution. It analyses the transcript of the candidate’s answers and identifies the framework competencies expressed in them, linking each detection to the passages of speech that explain it. The results remain interpretable and complement other assessment methods; they are never used as a single criterion and do not feed into the matching score calculation. Talent Brain does not analyse image, voice, or emotions.
The matching score is not generated by generative AI. It is based on deterministic calculation rules: the results obtained for each competency are compared against the competencies expected and weighted according to their level of importance. Each result can be reviewed competency by competency. HR professionals therefore retain full visibility over the criteria used and the elements behind the score.
AI that supports decision-making,
without replacing it
The more AI is involved in a sensitive HR decision, the more its role must be understandable, controllable, and documented.
At Key Predict:
AI-generated suggestions remain editable;
a fully manual build of the model remains available at all times;
results are presented as aids to analysis;
the score never constitutes, on its own, a recruitment or mobility decision;
results must be considered alongside context, the interview, and other available information;
the final decision remains the responsibility of the HR professional.
What our AI does not do
- It does not make decisions on the recruiter’s behalf.
- It does not “guess” a candidate’s personality from unstructured content.
- It does not intervene in the calculation of test scores or matching scores.
- It does not perform emotion recognition in the workplace.
- It does not turn a suggestion into automatic truth.
- It never removes the need for human supervision.
Our approach to the AI Act
The EU AI Act regulates AI systems used in employment, recruitment, and the assessment of individuals. Key Predict addresses this through constant human oversight, transparent and traceable documentation, rigorous risk and bias management, and data security certified to ISO/IEC 27001:2022 (ISMS).
Frequently
asked
questions
asked
questions
No. Key Predict assessments are, first and foremost, psychometric assessments, not AI-generated evaluations. They are based on validated psychological models, statistical methods and scoring algorithms designed to measure personality, abilities, motivations or competencies.
Artificial intelligence may be used in certain complementary features of the platform, for example to structure a competency framework or analyse certain signals. However, it does not replace the scientific construction of the assessments, their scoring rules, or their psychometric validation.
Psychometric assessments provide a structured way of measuring personality, abilities, motivations and certain competencies. AI helps to structure requirements, enrich models or analyse certain content. The Predictive Model then links assessment results to the competencies expected in a specific context to produce interpretable matching indicators.
No. The results provided by Key Predict are decision-support tools. They help structure criteria, compare competencies and prepare points to explore further. The final decision remains the responsibility of the HR professional.
Yes. The suggested occupations, functions, competencies and levels of importance can all be adjusted. Users can also build their framework manually.
No. Talent Brain analyses the content of speech using natural language processing techniques. It does not perform emotion recognition based on facial expressions, voice or biometric data.
Informing candidates is the responsibility of the employer running the process, and our journeys are designed to support this: presentation of the assessments, consent, and the option of a results debrief. We recommend stating that an assessment support tool is being used and that the final decision remains human.
Key Predict is carrying out a structured programme to prepare for the requirements applicable to AI systems used in employment: governance, documentation, risk and bias management, traceability, user information, and human oversight. This programme evolves alongside the texts, standards, and guidance published by European institutions.