Get in touch
Home OverviewChat Interview Features Trust All productsEight products, one data layer
Get in touch See how detection works

Products › Chat Interview

Blind assessment. And it knows
when a model wrote it.

Behavioural text interviews, assessed blind

Behavioural questions answered in text, assessed with no name, no face and no accent attached — and with AI-generated answers detected as they arrive rather than discovered three rounds later.

Book a demo See how detection works

Any device, any connection. Nothing to install and nothing to schedule.

MCQsA declared answer set, so the score is arithmetic.Timed Q&AA visible clock per question. Recall, not research.Conversational Chat InterviewOpen-ended, with follow-ups that go deeper.

Closed answers, constrained answers, open answers — three instruments with one rubric behind them, so a knowledge score and a conversation score sit on the same axis.

Everyone has the same model open
in the next tab.

A written screening question stopped being a signal the moment a language model could answer it perfectly. So this round detects that as the answer arrives, and keeps the two things a text round is uniquely good at: blindness, and thinking time.

  • AI-generated answers are flagged in real time — while the round is running, not after somebody notices three candidates wrote the same suspiciously polished paragraph.
  • Blind by construction. No name, no photograph, no accent, no appearance, no age cue. Just the answer against the rubric, which is the strongest structural bias control of the three rounds.
  • Written communication as the actual skill. For support, documentation, legal, marketing and any remote-first role, how somebody writes under a prompt is the job rather than a proxy for it.
  • Thinking time is a feature, not a loophole. A candidate who freezes on camera and writes clearly is a candidate you would otherwise never have seen.
  • Knowledge checks in the same round. Multiple-choice items drawn from your own question sets sit alongside the behavioural questions, so a factual screen does not need a second tool.
Round in progress — support engineer
QDescribe a time you gave a customer an answer they would not like.
A“A merchant’s payouts were held by our risk system. I couldn’t release them, so I told them that plainly and got them a written timeline the same afternoon…”Scored
QHow do you decide what to escalate?
A“Escalation should be guided by a holistic assessment of impact, urgency and stakeholder alignment…”
no specifics · register shift from the previous answerFlagged
Blind: no name, no photo, no accent on this screen · a flag goes to the reviewer, never to a rejection

No scheduling. No camera. No excuses.

Five stages, and the lightest round of the three. Identity is settled first, so everything after it is about the answers.

Invite sent
automatic
Identity first
in chat
They answer
any device
Flagged if needed
real time
You decide

What the Chat Interview does

Blind by construction, not by policy+
The reviewer sees the answers against the rubric. No name, no photograph, no accent, no appearance, no age cue — because none of them are on the screen to begin with. That is a structural bias control rather than a training reminder, and it is the reason a text round is worth running even when you could run a video one.
Detecting an answer a model wrote+
AI-generated answers are detected in real time, while the round is running. The flag goes to the reviewer with its reasoning attached — and it is a flag, not a rejection. Nobody is removed from a pipeline by a detector; a person looks and decides. That distinction matters, because a false positive here costs somebody a job.
Identity and documents, collected first+
Before the questions begin, identity and document collection is handled in the chat itself. Everything after that is about the answers, and you are never reviewing somebody whose paperwork turns out to be missing at the offer stage.
Question sets, adaptive follow-ups and knowledge checks+
Reusable question sets grouped into sections, with multiple-choice items mixed in where a factual screen is what you need. The interview adapts as it goes, so a thin answer draws a follow-up. Question generation runs in the background, so a new role does not wait on somebody writing a bank from scratch.
Scoring, with the excerpt attached+
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 to the answer it came from — and because this is the same rubric the video and voice rounds use, a chat score and a live-round score sit on one axis.
Pipelines, leaderboards and evaluation+
Candidates move through pipelines with weighted scorecards and leaderboards, so evaluation lives in one place instead of five inboxes. Hiring managers assess the same answers rather than a summary of them, and transcripts and analytics are retained for reporting on the process itself.

A flag is not
a verdict

Detection is the most consequential thing on this page, so it is the one with the most restraint built around it.

A human makes the call

It ranks, evidences and flags. It does not reject anyone. A flagged answer is surfaced with its reason for a person to judge, never auto-removed from a pipeline.

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.

Blind is the default state

Identifying detail is absent from the assessment screen rather than hidden behind a toggle somebody can switch off on a busy afternoon.

And plainly, about detection. No detector is perfect, and we publish no accuracy figure for this one — a percentage from somebody else’s dataset tells you nothing about your applicants. That is exactly why a flag is shown with its reasoning and never acts on its own: the cost of a false positive is a person losing a job they were qualified for. Treat it as a prompt to read the answer properly, not as a result. A text round also cannot assess spoken fluency or presence — that is what the Voice and Video rounds are for.

Voice InterviewWhen speaking is the skill — the same rubric, spoken and transcribed.Video InterviewWhen you need the round back, and the non-verbal signal with it.

Frequently asked questions

Does the candidate need to install an app or book a time slot to take the interview?+
No. It runs on any device over any connection, with nothing to install and nothing to schedule. There is no camera either.
If the interview is blind, how is the candidate's identity verified?+
Identity and document collection is handled in the chat itself before the questions begin. Everything after that is about the answers, so you are not reviewing someone whose paperwork turns out to be missing at the offer stage.
Will the AI-writing detector automatically reject a candidate, and how accurate is it?+
No. A flag is shown to the reviewer with its reasoning attached and never acts on its own; a person reads the answer and decides. TalbotIQ states that no detector is perfect and publishes no accuracy figure for this one, on the basis that a percentage from someone else's dataset says nothing about your applicants and a false positive can cost a qualified person a job.
How are the written answers scored?+
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 answer it came from, and the rubric is the same one used by the video and voice rounds, so a chat score and a live-round score sit on one axis.
Can I add multiple-choice knowledge questions to the same text interview?+
Yes. Reusable question sets are grouped into sections, with multiple-choice items from your own question sets mixed in where you need a factual screen, so it does not need a second tool. The interview also adapts as it goes, so a thin answer draws a follow-up question.
When is a text interview the wrong choice?+
When you need to assess spoken fluency or presence, which a text round cannot do. That is what the Voice and Video rounds are for: the Voice Interview uses the same rubric, spoken and transcribed, and the Video Interview is for when you need the round itself back, along with the non-verbal signal.

Published 9 September 2026 · Updated 9 September 2026 · Written by the TalbotIQ team

Ready to see it?

Tell us the role and what you are actually screening for, and we will run a text round against it — including a deliberately model-written answer, so you can see what happens.

hello@talbotiq.com
0 / 250
Request demo

Screen for the writing.

Book a 30-minute demo. We will run a chat round live and slip a model-written answer into it.

Book a demo Talk to us

Malaysia-based team · Response within one business day · No obligation