calhire
Every applicant arrives assessed and AI-interviewed.See how

AI interviewer

Meet the AI interviewer.The first interview, done before anyone applies.

A short, text-based interview that adapts each question to the last answer, scores every transcript the same way, and never sees who it is talking to.

Open to talent, no booking
24/7
Nothing watches or listens to talent
Text
Personal details sent to the model
0
Of the composite score, by default
35%

AI interview, backend engineering

Text only, untimed

  1. Interviewer: Tell me about a time a cache made things worse rather than better. What happened?
  2. Talent: Our payments screen cached balances. When data changed faster than the cache expired, people saw stale totals.
  3. Interviewer: Good. How would you find out that was happening in production?

What the interviewer is given

  • The role
  • The skills test score
  • This conversation

What it never receives

  • Name
  • Email
  • Photo
  • Work history

Every transcript is scored on

  • Communication
  • Depth
  • Reasoning
  • Professionalism
A sample. In a real interview every question is written live from the role, the test score and the answers so far.

What changes when the first interview runs itself

  • Interviews on talent’s time

    Talent starts whenever they are ready, day or night. Nobody books a slot on either side.

  • Your team starts at the shortlist

    Applying needs a finished vetted assessment, so every applicant has already been interviewed and scored.

  • The same criteria, every time

    Every transcript is scored on the same criteria, whether it is the first of the day or the thousandth.

  • Blind by design

    The model never receives a name, email or work history. Companies see the interview’s scores, never the conversation.

  • In the talent’s own words

    Paste is blocked during the interview and leaving the page is recorded. Anything unusual goes to a person.

  • A record you can explain

    Every AI-assisted score is logged with what it was based on, and AI never decides who moves forward.

How it works

From the talent's first answer to your ranked shortlist, in three steps.

  1. A timed skills test comes first. The AI interview follows straight after, whenever the talent is ready, from any time zone.

It reads what talent says about the work. Never how they look or sound.

Scoring faces and voices had a bad decade, and deserved it: appearance and speech carry disability, ethnicity, age and accent. Text puts a stammer, a strong accent or a webcam in a shared kitchen on the same footing as everyone else.

  • No facial analysis

    No camera, no recorded likeness.

  • No voice or tone scoring

    Nothing listens to how an answer sounds.

  • Re-readable answers

    Written reasoning can be checked by a second person.

Talent can practise first

A private AI mock interview, with feedback, that no company ever sees. Free talent get a daily allowance, so nobody walks into the real one cold.

Join as talent

What it does not do

Where the AI interviewer stops, and your team takes over.

It does not decide
It scores. Your team decides who moves forward, and AI never does.
It is not a video interview
Live video rounds are run by your team. No model watches or listens to them.
One interview, every role
The interview is part of the vetted assessment, so it counts for every role the talent applies to for 90 days. Each job also has its own short written test, generated from the role.
Writing still matters
Text favours fluent writers. For roles where written communication is genuinely irrelevant, weigh the interview score accordingly.
An integrity record is a signal, not a verdict
Leaving the page can be innocent, so a record of it is evidence for a person to weigh, never a decision.

The method in detailHow scores are built and monitored

Questions about the AI interviewer

Is the AI interviewer really available 24/7?
Yes. Talent starts it whenever they choose, straight after the skills test. There is no booking and no interviewer on the other side, so it runs at 3am as readily as at 3pm.
Does the AI decide who gets hired?
No. It scores answers and contributes one weighted part of the composite score. Your team reviews the ranked shortlist and decides who moves forward.
What does the model see?
The role, the talent’s test score and the conversation so far. Never a name, email, photo or work history: identity is held separately by CalHire’s identity broker and is not available to the interview.
Do companies see the transcript?
Not of the AI interview: companies see its scores, and the conversation stays between the talent and CalHire. A job’s own short written test is different: that company sees the written answers, with identifying details removed.
Is it a voice or video interview?
No. It is text only, by design. Live video rounds later in a process are run by your own team, and no model watches or listens to them.
Can we write our own interview questions?
Not today. The interview is part of the vetted assessment, so one interview counts for every role a talent applies to. Each job also has its own short written test, generated from the role.
Can talent use AI to write their answers?
Paste is blocked during the interview, and leaving the page is recorded for a person to review. Each question also follows up on the last answer, asking how and why.
Which model runs it?
Model choice is configuration, not code, so a model can be changed without a release. What never changes is the boundary: whichever model runs, it receives no personal data.

Start with applicants who are already interviewed.

Post a role on the free plan and review anonymous talent, ranked by what they can actually do.