Built on psychometrics, not personality clichés.
Every question, weight, and recommendation is grounded in measurement theory and tuned against real answers. Here’s what runs under the hood.
Adaptive item selection
Instead of a fixed list, the engine picks the next question that will tell it the most about you right now, the same idea behind modern computer-adaptive testing.
Bayesian inference
Each answer updates a probability distribution over fields and careers in real time. Nothing is decided from a single answer; belief accumulates as the evidence does.
Psychometric calibration
Items are weighted and continuously reviewed using discrimination and difficulty signals, so weak or ambiguous questions get rewritten rather than trusted.
Confidence before commitment
The system reports how sure it is, and refuses to force a result when your signal is genuinely mixed, an honest explorer outcome instead of a false match.
Labour-market grounding
Careers, growth signals, and salary ranges are mapped to the Saudi market and the Vision 2030 economy, not a generic Western dataset.
Bilingual by design
Calibrated independently in Arabic and English, so meaning, not translation, drives every question and result.
Mentora’s results are decision support, not a verdict. They’re grounded in your actual behaviour and reported with a confidence level, never dressed up as certainty we don’t have.