How Kif works
How Kif matches candidates and roles, and what a match means.
Kif checks eligibility, scores fit in both directions, and combines those scores with a confidence adjustment. A match explains the available information. It is not a verified credential, a hiring decision, or a guarantee of performance.
01 / Confirmed requirements come first.
Matching requires an active candidate profile and a published role. The engine checks confirmed requirements such as work mode, employment type, and relevant location or authorization conditions. Some missing information blocks eligibility; other missing information is recorded as uncertainty. Missing information does not silently become proof of a match.
These checks can exclude a pair from matching. That is different from an AI making a hiring decision. The requirements need to reflect the real job, and people remain responsible for their decisions.
02 / Each side gets a separate fit score.
These are the current top-level weights in Kif's scoring implementation, reviewed on 16 October 2026. They describe the product's rules, not independently validated probabilities of a successful hire. Individual role competencies also have their own confirmed weights.
| Hiring team's fit | Weight | Candidate's fit | Weight |
|---|---|---|---|
| Competency fit | 45% | Career direction | 30% |
| Skill fit | 20% | Compensation | 25% |
| Evidence strength | 15% | Work mode | 15% |
| Domain context | 10% | Growth scope | 15% |
| Preference alignment | 10% | Candidate preferences | 15% |
03 / A weak side cannot disappear into a strong average.
Kif combines the two directional scores using a harmonic mean, then adjusts for confidence. This makes an uneven match less attractive than a simple average would suggest.
Illustrative arithmetic: scores of 90 and 30 have a harmonic mean of 45 before the confidence adjustment, rather than an arithmetic average of 60. This example explains the formula; it is not a real candidate's score.
For unknown scoring components, the engine uses a neutral value and reduces confidence. Explanations distinguish strengths, gaps, and unknowns so you can see what still needs a conversation.
04 / Evidence has a source and a limit.
A claim in a CV is not automatically an independently verified fact. Work samples and references can add context, but they do not justify treating every statement as verified. Ask what the person did, what supports it, and what remains uncertain.
AI helps extract and organize information. The score is computed separately from those conversations. The same inputs under the same scoring version produce the same score, but deterministic does not mean unbiased or infallible.
05 / People decide whether to connect.
Both sides must express interest before an introduction. A high score does not promise an interview, and a lower score should not be read as a judgment of someone's worth or overall ability.
If an explanation depends on incorrect profile information, correct the underlying details. Treat the match as a starting point for questions about the work, not a substitute for an interview or due diligence.
Common questions
Is a score of 80 an 80% chance of getting hired?
No. It is a fit score under Kif's current rules, not a calibrated probability of an interview, offer, or successful hire.
Is every profile claim independently verified?
No. Review the source and verification context of the evidence. Candidate-provided information and independently supported information are not interchangeable.
Your next step
Make your experience easier to understand.
Review your experience and preferences. Decide which introductions are right for you.