Products › Conversational AI Avatar Interview
The same interview,
the fortieth time.
A composed assessor that never drifts from the script
An avatar conducts the round, and it asks the fortieth candidate exactly what it asked the first. No fatigue, no rapport bias, no half-remembered follow-up — the script is the script, and every answer meets the same rubric.
Runs in the browser. Nothing for the candidate to install.
The same eight questions, asked identically, forty times over.
Your best interviewer is worse
at four in the afternoon.
Interviewer variance is the largest uncontrolled variable in most hiring processes, and nobody measures it. An avatar removes it by construction: the questions, the order, the wording and the manner are identical for candidate one and candidate forty.
No fatigue drift. The fortieth round of the week is delivered exactly like the first — same questions, same order, same wording, same pacing.
No rapport bias. A candidate who happens to remind the interviewer of themselves gets no head start, because there is nobody there to warm to.
Identical manner for everyone. A composed, even presentation, so a nervous candidate is not reading tension in the interviewer’s face and adjusting to it.
Adaptive, but bounded. A thin answer draws a follow-up — and the follow-up comes from your question set rather than being improvised on the spot.
The same rubric as the other two rounds. An avatar score, a recorded score and a live score are directly comparable, so the modality is a logistics choice rather than a scoring one.
“What did you change first?”
Candidates are told they are being interviewed by an avatar — before the round starts
One script. Every candidate.
Five stages, and the middle three are identical for everybody. That is the entire point.
What the Avatar round does
What “runs the same script” actually means
Not just the same list of questions. The same wording, the same order, the same pacing, and the same manner of asking — because a question delivered warmly and a question delivered flatly are not the same question. A structured interview only produces comparable results if the delivery is comparable too, and that is the part a human interviewer cannot hold constant across forty rounds in a week.
Bounded follow-ups
A thin answer still gets probed — but the probe is selected from your own question set rather than invented in the moment. That keeps the round adaptive without reintroducing the variance the avatar exists to remove. You can see exactly which follow-ups are available for each question before a single candidate takes it.
The check before the round, and the disclosure
The candidate runs a short check on camera, microphone and connection, and identity and document collection is handled in chat first. They are also told, before the round begins, that the interviewer is an avatar. Nobody should discover that halfway through a job interview.
Scoring, and the evidence underneath it
Every answer is scored against your rubric by a model tier reserved for scoring, validated against a fixed schema rather than accepted as free text. Each score links back to the transcript excerpt and the moment of video it came from, so a rating can be traced to something the candidate actually said.
Body language and sentiment
Posture, eye contact, gestures and composure are turned into structured, reviewable signal. Worth noting that an avatar round changes the baseline here — candidates behave differently when there is no human face watching — so treat it as one input among several rather than a verdict.
Pipelines, leaderboards and evaluation
Candidates move through pipelines with weighted scorecards and leaderboards. Hiring managers rate the same recorded round rather than a summary of it, and transcripts and analytics are retained for reporting on the process itself.
Consistency is the claim.
So is the disclosure.
An avatar interview is a stronger claim than a recorded one, so it carries a heavier obligation.
A human makes the call
It ranks and evidences. It does not reject anyone. The hire, the pass and the pipeline stage are decisions a named person takes, on the record.
One AI gateway, no loose ends
Every model call passes through a single choke point with schema validation, caching, per-tenant rate limits and spend caps. There is no second path to a provider.
Media behind signed links
Recordings sit in private storage reached only through short-lived signed URLs, with lifecycle rules that expire media on a schedule.
And plainly. An avatar is not a person, candidates are told so before they start, and some of them will not want to be interviewed this way — which is a legitimate preference and a reason to offer the 2-Way round as well. Consistency is not the same as fairness: an identical script applied to everybody is a genuine improvement on interviewer variance, but a badly written script is now badly written for all forty candidates instead of some of them. We publish no accuracy figure for the scoring. And an avatar round is a screening instrument — it is not a substitute for meeting somebody before you hire them.
Frequently asked questions
Does the candidate need to install anything to take the interview?
No. The interview runs in the browser, so there is nothing for the candidate to install.
Are candidates told they are being interviewed by an AI avatar, and what happens before the round starts?
Yes, candidates are told before the round begins that the interviewer is an avatar. Before that, they run a short check on camera, microphone and connection, and identity and document collection is handled in chat first.
How are answers scored, and can I see why a candidate got a particular score?
Every answer is scored against your rubric by a model tier reserved for scoring, and the output is validated against a fixed schema rather than accepted as free text. Each score links back to the transcript excerpt and the moment of video it came from, so you can trace a rating to what the candidate actually said.
Can the avatar ask follow-up questions, or does it only read a fixed list?
It probes thin answers with follow-ups, but each probe is selected from your own question set rather than invented in the moment, so the round stays adaptive without reintroducing interviewer variance. You can see exactly which follow-ups are available for each question before any candidate takes it.
Can I compare avatar interview scores with scores from a recorded video interview or a live two-way interview?
Yes. All three rounds use the same rubric, so an avatar score, a recorded score and a live score are directly comparable, and choosing the format becomes a logistics decision rather than a scoring one.
Can an avatar interview replace meeting candidates in person?
No. The avatar round is a screening instrument, not a substitute for meeting somebody before you hire them, and TalbotIQ publishes no accuracy figure for the scoring.
Remove the interviewer variance.
Book a 30-minute demo. Bring a question set and we will run it twice, so you can watch it come out identical.
Malaysia-based team · Response within one business day · No obligation