Skip to the content
AbilityRating

For developers

AbilityRating is being prepared for launch; developers can try it first. Early sittings start in the first months of 2027, for developers in Europe, the United Kingdom and the United States at every level of seniority: employed, freelance or between jobs. They are anonymous: no account, no name and no email address. Taking part has no effect on any job application; no employer sees your results.

The sitting is built for development work as it is now: AI tools allowed, no trivia and no need to perform while someone watches. What matters is how you understand a brief, review a plausible answer, direct an AI agent and explain your judgment.

What taking part involves

What you get

What the early sittings check

Before any employer relies on a score, it must be shown to be consistent and fair. The first sittings check whether results are reliable and whether independent readings of the open answers agree; later ones check whether any group of people is disadvantaged. Results from early sittings are used only to check and improve AbilityRating.

Questions developers ask

Q: Do I need to prepare?
A: No. The problems are about everyday development judgment, not syntax or algorithms.

Q: Which language will the code be in?
A: You choose from several mainstream languages before the sitting. The early sittings check that the choice does not change the difficulty.

Q: Can I use AI tools?
A: Yes, as at work. One problem asks you to direct an AI agent; what counts is your judgment about what to ask for and what to accept.

Q: Will an employer see this?
A: No. Results from early sittings are never shared with an employer and create no profile that follows you.

Q: Can I stop halfway?
A: Yes. You can close the sitting at any time; an unfinished sitting is not used. At the end you get a code that deletes your answers if you change your mind within 7 days; after that they cannot be traced to anyone.

Q: Can I take it more than once?
A: Please don’t. Each sitting counts as a different person, so a second one would distort the results the early sittings exist to check.

Q: How are the open answers scored?
A: AI models read each answer against a scoring guide written for its problem; a person reviews every case where the readings disagree. You see the reasons for every score.

Q: What if I disagree with my result?
A: Write to us within 7 days with the code shown at the end. A person reviews it; before launch that is exactly the feedback we need.

Q: What do I need?
A: A computer with a current browser and a stable internet connection.

Your data

We do not ask for your name, your email address, a CV, a LinkedIn profile or the name of your employer, and you do not create an account. The sitting stores your answers and how long each step took, under a random code that is linked to nothing else. Questions about your background for the fairness analysis come only after the assessment; they are voluntary and anonymous like your answers. If an open answer mentions a name, an employer or anything else that could identify someone, we remove it before the answers are analysed. The notice shown at the start of the sitting explains the rest; the privacy page covers this website.

How to take part

The sittings open in the first months of 2027. The link will be on this page: there is nothing to sign up for and nothing to wait for by email.