Screen a candidate with AI

AI screening reads a candidate's application against the job they applied to and writes up an assessment: an overall score, a short summary, what stands out in their favour, what gives you pause, and a recommendation with the reasoning behind it.

It runs on one candidate at a time and is always tied to the job they applied to. It doesn't move anyone through your pipeline and it doesn't change anyone's status. It gives you a read so you can decide faster.

• An overall score: a single rating out of 100 that you can compare across candidates.
• Category scores: separate ratings for skills match, experience, education, culture fit and communication, so a strong overall score doesn't hide a weak spot.
• Summary, strengths and concerns: a short written assessment in plain language, rather than just numbers.
• A recommendation: a suggested call with its rationale spelled out, so you can judge whether you agree with the reasoning.

Screening is a paid action, so a few things need to be in place first:

• A paid plan: AI screening is a paid feature, so it needs a plan that includes it. If yours doesn't, you'll be told as soon as you try to run one.
• Tokens: each screening costs 4 tokens. If your balance is short you're told before anything is charged.
• A candidate on the application: screening reads a real applicant's details, so an application with no candidate attached can't be screened.

Open the job, open the candidate you want to screen, and go to the Evaluations tab. Click the AI screening button and confirm the token cost. The button is only there while the candidate hasn't been screened yet, so if you can't find it, the results are probably already further down the page.

Screening takes a moment to run. When it finishes, the results appear on the same tab as an AI screening card.

Rather than clicking through candidates one at a time, you can have a pipeline stage screen them for you. Open your pipeline, click into the stage where screening should happen, and go to the Automations tab. Add the Run AI screening automation and set it to fire when a candidate enters the stage.

• Pick a trigger: entering the stage is the usual choice, so anyone you move in gets screened.
• Choose the screening language: so the write-up comes back in the language you work in.
• Leave the automation enabled: a disabled automation stays configured but never fires.

A finished screening shows up as a card on the candidate's Evaluations tab, with the overall score, the category breakdown, the summary, the strengths, the concerns and the recommendation.

Treat it as a second opinion rather than a verdict. The score reflects what the candidate wrote in their application, so a thin application can undersell even a strong candidate. The write-up is there to be argued with.

A finished screening is the prerequisite for both of the other AI features. Once it's done you can have AI draft follow-up questions for that candidate, or draft a rejection letter that reflects why they weren't a fit. Neither is available until the screening has completed, because both are written from its conclusions.