Questionnaires
The test is a record,
and so is the judgement.
You write the questionnaire once — sections, questions, right answers, weights — publish it as a version, and send it as a link that needs no login. What comes back is graded: choice questions by the system, open answers in a queue a person signs off.
A questionnaire has versions, and submissions keep theirs.
Editing a published questionnaire creates the next version, and the answers already collected stay pinned to the version they answered. Six months later a score still means what it meant on the day, because the questions behind it have not moved.
Version 1 and version 2 side by side, each with its own submissions.
Choice questions come back scored. Open ones queue for a person.
Single and multiple choice come back scored, with partial credit where you allow it, and per-skill weights build the final figure. Open answers land in a grading queue — with an AI proposal where you want one — and nothing becomes a result until a person signs it.
The grading queue: what the system scored, what the AI proposes, what a person still has to sign.
The candidate clicks a link. That is all.
An invitation is a link with an expiry — no account, no password. Answers save as the candidate goes, and an invitation about to lapse warns the person who sent it.
The invitation as the candidate sees it: the questionnaire, and nothing else.
Recruiting is where it works first.
A position attaches its screening questions and its skills tests from this same engine, so what a candidate answered sits on their application next to the interviews. The engine itself is neutral: a survey or an onboarding check is the same mechanism pointed elsewhere.
A position's pipeline with a skills test on one stage, scored on the candidate's file.
Everything it does
- Question and test builderSections with three question types, points and skills; start from templates and publish with tracked versions.
- AI drafting and gradingAI drafts from a brief and proposes reasoned scores; a person decides, tracked in an oversight dossier.
- Campaigns and surveysTargeted distribution with an open window, HR-confidential mode and results aggregated per question.
- Personal invitations and public linkPersonal links with resend, extend and revoke, extra time for accessibility, or a public link and QR.
- Respondent transparency and privacyMandatory privacy notice, upfront transparency, no-account access, timed retention, a read access log.
- Grading and review queueChoices graded automatically, blind or rubric review, a company-wide queue, an answer-by-answer report.
- Exam integrity and rankingRanking with a pass line, integrity signals, invalidated compromised attempts, server-enforced time limit.
- Tests linked to applications · with the Recruiting (ATS) moduleThe questionnaire outcome shows directly on the candidate's file.
Questions we actually get asked.
- Can we see partial scores while grading is still open?
- No, deliberately. A result appears when grading is complete, so a half-graded test never circulates as if it were a judgement.
- Does the AI decide the grade?
- It proposes. Every submission that needs judgement passes through a person who signs it, and blind review — graders who cannot see each other's marks until they commit their own — is there where you want it.
- Which question types exist?
- Single choice, multiple choice and open answer, in sections, with weights per skill. There is no code-exercise and no file-upload question type, and we would rather say so here than let you discover it.
Works with
- RecruitingPositions, candidates and interviews, from the careers page to the signed contract.
- AttendanceClock-ins that become a daily timesheet, reconciled against the hours logged on job orders.
- Shift planningPlan people by period against the jobs they work on, and see the load before it becomes overtime.
Tell us how you work today.
There is nothing to prepare. Half an hour on how your company runs — who tracks the hours, how an invoice comes together, when you find out whether a commessa made money — and we tell you which modules would fix it and what a quote looks like.