How the guide is made

The guide never gives a candidate a score, a rating, a pass or fail, or a place in a ranking. It does not compare candidates with each other, decide who should progress, or recommend who to hire. Candidates are listed in the order they submitted, nothing else.

Every decision is made by a person at the hiring company, who reads the candidate's own answers.

When you create an assessment, the questions are written for the role, the workplace and the experience level you choose. Every candidate for that role gets the same questions, in the same order, with the same time limit.

Under each answer, the guide lists the points the candidate made, in their own words and in the order they made them. It does not add points they did not make, judge how good the answer is, or fill gaps.

Very short answers (for example "I don't know") and unanswered questions get no summary: the guide simply shows what they wrote, or that the question was not answered, and why (for example, the time ran out).

Each question comes with two or three follow-up questions for the interview. Some are core questions you can ask every candidate; others build on something the candidate actually wrote ("You mentioned…"). They are never chosen because an answer seemed weak.

"Listen out for" points under each follow-up help with note-taking during the interview. They are not a checklist, a marking scheme or a pass mark.

An AI model writes the questions, the summaries and the suggested follow-ups, following fixed instructions that forbid scoring, rating or ranking. It reads only what the candidate typed: never their name, photo, email address or anything else about them.

AI can still get things wrong, for example missing a point or phrasing a summary loosely. That is why the candidate's full answer is always shown above the summary, and the guide reminds you to check anything important with the candidate directly.

No tool is free of bias. We reduce it by keeping the AI's job narrow (summarising and suggesting questions, never judging), giving every candidate for a role the same questions, and never using personal details. We cannot promise the summaries are perfect, so the decision always rests with a person who has read the answers.

If part of an answer is pasted in rather than typed, the guide shows a flag so the interviewer can ask the candidate about it. A flag is a fact about how the text arrived, not a judgement: drafting elsewhere or using dictation software can cause one. Flags never change the guide's summary or questions.

The workplaces and job titles you can choose from combine everyday UK and Irish job titles with occupations from ESCO, the European Union's official classification of skills, competences and occupations. ESCO also supplies the alternative titles that make search work: searching for "bellman", for example, finds Porter.

Contains ESCO data © European Union, used under the Creative Commons Attribution 4.0 licence (CC BY 4.0). Titles have been adapted, and some lists have been shortened. The European Commission does not endorse this service. More about ESCO: esco.ec.europa.eu