ViblyEmployers
Read the original answers, add a note or evidence, ask for a second read, and move an applicant to another group. Nobody is rejected by the software.
How the matching works
What the AI reads, and what it does not.
When someone applies, a model reads what they submitted and compares it with what the employer wrote. It gives a first read. A person decides.
| The model reads | The model never receives |
|---|---|
| The applicant's answers to the employer's questions | A name |
| The note they chose to write | An email address or a phone number |
| What they declared: pay, level, years of experience, start date, place | Profile sections the person switched off |
| The profile sections they left on | Anyone else's application |
| On a second read: the employer's notes, follow-up answers and interview answers | Anything from another employer's listings |
What it gives back.
For each requirement the employer wrote, a result: met, not met, or unknown. A result of met or not met has to point to the applicant's own words. If it cannot, it is recorded as unknown.
Unknown is a different thing from a no. It means the application did not say enough to tell. The employer sees the difference, and an unknown is not treated as a failure.
The label comes from the score by fixed cut-offs. 80 and above is a Match, 60 to 79 is a Maybe, and below 60 is a No match. A must-have counts for more than a good-to-have, and a clearly stated disqualifier lowers the score.
The employer reads the same answers the model read, can disagree with any result, and moves anyone to any group. Every move is logged, and the first read is kept.
What people can do about a result
Applicants
Add to your profile and your note before you apply. If you think a read was wrong or unfair, tell the employer, and write to the grievance officer on the page linked above.
Fairness, as far as a rule can promise it
Only the requirements count
The model is told to judge against the requirements that were listed, and not to use or infer names, age, gender, religion, caste, disability or anything that stands in for them. It is told not to reward writing style or the prestige of a school or employer beyond what a requirement asks.
Employers have rules too
Employers agree not to use a first read to discriminate unlawfully. Anyone can report a listing, a company or a conversation, and a moderator reads it.
Can the model get it wrong?
Yes. It can misread an answer, miss a detail, or be wrong about a skill that was never written down. That is why the evidence is shown beside each result and why a person decides.
Does my profile change my result?
The sections you leave on are part of what is read, so a fuller profile gives the read more to go on. Sections you switch off are not sent.
Is my application used to train a model?
Our providers process your text only to give us the result. Read the Privacy policy for the details.
Does the AI talk to me?
On the home page, yes: it helps you find roles or draft a listing, and it only discusses Vibly. For an application, it only reads. An optional interview is the one other place, and only if an employer invites you and you accept.