We will not tell you our assessments are unbiased
That is not provable, and a vendor who claims it should not be believed on anything else. What we will tell you is exactly what is structured, what is evidenced, what a person decides, and what we have not measured — so your compliance or equity function can form its own view instead of taking ours.
Already a customer? Sign in to your dashboard
What is structurally fair, and why that is the only kind of fair worth claiming
The scorer never sees who it is scoring
Grading works from a transcript labelled only by role, never by name. No name, no email, no photo, and none of the inference a name can carry about someone's background, gender or origin reaches the thing producing the score.
The standard is fixed before anyone applies
Competencies, questions and the rating scale are set when the assessment is published, not tuned once results start arriving. Nobody can move the bar after seeing who cleared it.
Identical treatment across the whole cohort
Every candidate for a role receives the same questions in the same order and is scored against the same rubric, whether they applied in week one or week six, and whichever member of your team is looking.
Work samples, not psychological inference
The scored instruments are job-relevant: a case study built from your own advert, situational judgement items, knowledge questions, and a structured interview on competencies you named. They assess demonstrated behaviour and knowledge, not personality traits.
The optional profile never ranks anyone
A work-style profile is available and is deliberately unscored. It produces no number, contributes nothing to any ranking or shortlist, and can be left switched off entirely.
Every score carries its reasoning
Each competency score points back to what the candidate actually wrote or said. There is no unexplained overall figure, and no score derived from a signal the candidate was never asked about.
Written for the people who have to sign it off
Talent acquisition asks whether it works. Compliance, equity and procurement ask whether it can be defended. Those are different questions and they deserve different documents.
You can reconstruct any decision
For any candidate, you can show what was asked, what they answered, how each competency scored and why. That record is the thing an internal review, a candidate query or a regulator actually asks to see.
Consistency is the part manual sifting cannot promise
Human screening drifts with fatigue, order and reviewer. A fixed rubric applied identically is not a claim about intent, it is a property of the process, and it is checkable.
The limits are written down, not discovered later
What we have not tested is stated in the same document as what we have. A limitation you find in procurement is a problem; a limitation we hand you up front is a scoping conversation.
One document your team can forward as-is
The workflow handbook is written for a compliance and equity readership rather than for a buyer, so it can go straight to the people who need it without translation or a sales call attached.
How to check any of this rather than take our word for it
Every step here is something you can do before committing to anything.
Read the Handbook
Request the workflow handbook and give it to your compliance and equity teams. It covers the workflow end to end, the basis for scoring, and data handling, including the sub-processor detail this page deliberately leaves out.
Take It Yourself
Sit the assessment as a candidate on your own role, then read the scored report it produces. The fastest way to judge whether the evidence would satisfy a panel is to look at it.
Interrogate a Score
Pick any competency on that report and trace it back to the answer it came from. If a score cannot be explained from the record, that is the objection to raise before you buy, not after.
Plan the Impact Analysis
Decide up front how you will monitor outcomes across groups once real candidates run through it. We would rather agree that as a deliverable of your first cohort than leave it unowned.
What we do not claim
This section exists because its absence elsewhere is the reason these documents are usually worthless.
We have not run an adverse-impact study
We have not analysed differential outcomes across race, gender or any other group, and we will not imply otherwise. Doing it properly needs real cohort data from a real employer. We propose it as a joint deliverable of a first intake rather than a claim made in advance.
AI-graded sections are highly reproducible, not identical
Objectively-scored sections give the same answer the same score, guaranteed. For sections graded by AI the honest word is highly reproducible: the rubric, the order and the inputs are fixed, but we will not promise bit-identical output, because we cannot.
Every decision is yours, including the decision to automate one
Progression and rejection are your team's calls. If you would rather apply a cut-off automatically, rule-based automatic rejection is available as an option you switch on and set the threshold for — the rule is still a human decision, in the same way a knockout screening question is. Worth governing and reviewing as one.
Our validation evidence is ours, not yours
Scoring has been checked against human-scored reference material, and we will share what that showed. It is evidence that the method behaves sensibly. It is not validation on your population, your roles or your market, and it should not be presented internally as though it were.
Explore Related Solutions
Discover more ways Kaairo can help your organization hire and develop talent.
Frequently Asked Questions
Does AI make the hiring decision?
No — your team does. Assessments produce scores and the evidence behind them; your team reviews, compares and decides. Organisations that want a cut-off applied automatically can switch on rule-based automatic rejection and set the threshold themselves. That is still a human decision, in the same way a knockout screening question is: you define the rule, in advance, for everyone. It is off unless you turn it on, and if you do turn it on it is worth governing and reviewing like any other selection rule.
Have you tested your assessments for adverse impact?
No, and we say so rather than imply otherwise. An adverse-impact analysis compares outcomes across groups and needs a real cohort from a real employer to be meaningful. We hold no such dataset. Where an employer is running an intake with us and already collects the demographic data its own reporting obligations require, we would rather agree the analysis as a joint deliverable of that first cohort than make a claim we cannot support.
Can the same answer receive a different score?
For objectively-scored sections, no: the same answer always receives the same score. For sections graded by AI, results are highly reproducible rather than bit-identical, and we prefer the accurate word to the reassuring one. The rubric is fixed before anyone applies, every candidate faces the same questions in the same order, and every score carries the reasoning it was based on, so a result can always be examined rather than taken on trust.
Does the system know the candidate's name, gender or background when scoring?
No. Grading works from a transcript labelled only by role. The candidate's name, email and contact details are not part of what the scorer receives, and no demographic attribute is collected for or used in scoring. Your team of course sees candidate identity in the hiring interface; the scoring step does not.
Where is candidate data stored, and for how long?
Candidate and assessment data is stored in India, encrypted at rest, with some processing carried out by sub-processors outside India. The specific sub-processors, locations and purposes are set out in the workflow handbook so your team can run its own transfer assessment. Retention is set to your policy: tell us the period your organisation works to and we configure your account to it and record it in the agreement. Absent an instruction, assessment records are kept for the life of the account.
Are you compliant with our data protection law?
We will not answer that with a yes, because compliance is a property of your processing rather than of our product, and the applicable regime differs by market. What we will do is give you the facts your own assessment needs: where data sits, who processes it, what is collected, how long it is kept, and how deletion works. Our own privacy framework is written to India's data protection regime; we do not currently offer data residency in other regions, and we will not describe that as equivalence with a law we are not built to.
What can a candidate ask for?
Candidates can ask what was collected about them and request deletion, either through the organisation that invited them or directly from us. Assessment records belong to the organisation running the hiring process, so a deletion request is handled with them rather than around them.
Ask for the workflow handbook
Tell us who needs to review it — compliance, employment equity, privacy, procurement — and we will send the document written for them.