AI candidate screening
What AI candidate screening is
AI candidate screening is software running the first pass. What it looks at — a résumé or an answer — decides what you get back.
Two things share the name
They are not the same job, and the difference decides what you get back.
The résumé
Reading what people wrote
Software matches a document to keywords. Fast, cheap, and shallow — it ranks how someone writes about themselves.
It can’t tell you whether they can do the work.
The interview
Listening to what people say
Software asks the questions and listens to the answers. Slower per candidate, and much closer to evidence.
This is what Kira-AI does. Voice only, no camera.
A score with no evidence is noise
A number on its own tells you nothing you can act on.
If you can’t see why it landed there, you open the résumé anyway. The automation has added a step.
So every criterion carries the line it came from: timestamped, word for word, out of the transcript.
When the evidence is too thin to call, Kira-AI withholds the band instead of guessing one.
Screening for a client rather than for yourself? The same quotes are what agency recruiting software has to hand over.
How screening runs
You describe the job
A job description is enough. Kira-AI drafts five to eight questions tied to criteria, each with Weak, OK and Strong anchors.
You send one link
One link covers the whole role. It goes in the ad, in an email, or on your careers page.
Kira-AI asks and listens
Voice only, no camera, in 13 languages. When an answer is thin she asks about it rather than moving on.
You compare and decide
Open scorecards one at a time, or put up to ten side by side and read them criterion by criterion.
What lands on your desk
One per interview, with the whole conversation underneath.
Strong, Mixed or Weak
A read on the interview as a whole, so you know which ones to open first.
Scored 1 to 5
Each criterion is scored against the Weak, OK and Strong anchors you wrote, with a sentence saying why.
Met, Not met, Not confirmed
Up to three per role. Not confirmed is its own answer: the interview never produced the evidence either way.
The full transcript sits underneath, searchable, with markers where nobody spoke. Your own rating and notes go on the same page.
Working out which tool to buy? Our roundup of candidate screening software covers the rest of the field.
What it can’t tell you
Screening narrows a list. It doesn’t decide, and it doesn’t know everything.
- Whether a claim is true. Kira-AI records what was said, not what checks out.
- Skills you can’t hear. A work sample still belongs later in the process.
- Whether someone fits your team. That is a conversation with a person.
- Who to reject. Kira-AI never rejects anyone, and every scorecard is read-only.
Best fit: a first round with more applicants than your team can call this week.
Screening questions
Software runs the first pass instead of a person. It asks everyone who applied the same questions. Answers are recorded and scored against a written standard, so you read evidence rather than CVs.
No. Kira-AI never rejects anyone. Every scorecard is read-only and says so in print. It narrows the list; you decide who moves forward, and the evidence is there to argue with.
We do not claim it removes bias, and we hold no bias audit. What we can say: the question set is the same for everyone. Every score cites a quote, and blind review hides names.
Plans include 100 or 500 interviews a month. Two hundred people can answer the same evening. At the cap, new interviews stop rather than bill.
Not today. You read scorecards in Kira-AI. It does connect to ChatGPT and Claude, so you can ask either one about a role without opening the app.
Open a link and talk. No app, no account, no camera, no slot to book. They answer when it suits them, in any of 13 languages, and Kira-AI says she’s AI in her opening line.
See it on a real vacancy
Free trial, 25 interviews, no card. $149 a month after that.