A voice AI interviewer asks candidates spoken questions, listens to their answers and follows up before giving the hiring team a report. The conversation may happen through a phone call or a browser.
To choose one, listen to a full interview for a role you hire. A smooth greeting tells you little about what happens when a candidate gives a vague answer, corrects a mistake or asks a question of their own.
Start With the Role and the Conversation
Decide what the first round needs to establish. For a warehouse role, it may cover shift availability and relevant experience. For customer support, it may include explaining a policy or handling a complaint. A developer may need to write and run code as well as talk through it.
Fabric supports conversations, customer and sales roleplays, case studies and technical interviews. Choose the format from the work you need to assess. A voice conversation can be one part of a broader task.
Listen for Useful Follow-Up Questions
A candidate may say, “I improved our response time.” The interviewer should ask what changed, what the candidate did and how they knew it helped.
For an operations role, a useful exchange might be:
Interviewer: Tell me about a delayed order you helped resolve.
Candidate: I spoke to the warehouse and kept the customer updated.
Follow-up: What did you need to check with the warehouse before promising a delivery date?
Candidate: Whether the item was ready and whether the carrier could collect it that day.
The follow-up gives the reviewer evidence about the candidate’s process. It should stay close to the role and the answer, without turning a short first round into an unrelated conversation.
During a demo, try an incomplete answer and a correction. Check whether the interviewer follows the new information or repeats the same question.
Check Pauses, Interruptions and Candidate Questions
Candidates may pause to think, ask for a question to be repeated or speak before the AI finishes. Listen to how the system handles each case. A rushed turn can cut off a useful answer; a long delay can leave the candidate unsure whether the call is still active.
Candidates also ask about the job. Give the interviewer approved information about the role, pay range, location and next steps. Ask the vendor what happens when the answer is unknown. The system should have a clear way to refer that question to the recruiting team.
Greenhouse Voice AI, for example, describes dynamic follow-ups and answers drawn from information approved by the team. Use that kind of concrete behaviour as a buying check, whichever product you consider.
Test the Languages Your Candidates Use
Test with the languages and accents in your applicant pool. Check whether the transcript preserves the candidate’s meaning and whether follow-ups stay relevant after a misunderstanding or language change.
Fabric supports 86 languages globally, including switching between supported languages in the same interview. Teams can read results in their chosen language, with English as the default.
Keep language requirements tied to the job. For a customer-facing role, clear communication may be a skill you need to assess. For a technical role, assess the technical answer separately so an unfamiliar accent does not become a substitute for judging the work.
Read the Evidence Behind the Score
Take one result from the demo and trace it back to the answer. If a candidate receives a low score for problem-solving, the reviewer should be able to see which response led to it.
| What to inspect | What it should help you decide |
|---|---|
| Recording and transcript | What was asked and what the candidate answered |
| Skill assessment | Which role criteria the answer meets or misses |
| Cheating flags | Whether possible outside help needs review |
| Next-round questions | Which gaps a person should explore |
Fabric provides the recording, transcript and assessment. Its cheating detection flags signs of outside help for the hiring team to review. Check those flags alongside the answers before moving someone forward.
Avoid judging candidates from vague labels such as “confident” or “good personality.” Ask what the candidate actually demonstrated against the role’s criteria.
Check the Candidate Journey and ATS Handover
A phone call and a browser interview need different instructions. Tell candidates how to start, what device they need, how long to allow and who to contact for help. Fabric can be used on a phone or computer; use a computer for coding tasks.
Fabric works on slow internet connections and supports resuming after an interruption. Confirm the support process for the format you will use, then test how incomplete and completed sessions appear to reviewers.
Check your ATS integration with a sample candidate. Recruiters should know where to find the report and who is responsible for the next step. For calling products and their channel differences, use the AI phone-screening comparison.
Choose From the Conversation and the Review
Fabric is our recommendation when you need a responsive voice conversation alongside roleplays, technical work and other assessment formats. Its lively voices, quick responses and few interruptions let candidates develop an answer rather than rush through a script. Test a full session for your role, then let a hiring manager review the result. The choice should be clear from how well the interview explores the work and how easy it is to check the assessment.
Hear a Fabric Interview for Your Role
Try the conversation, follow-up questions and reviewer report with a role you are hiring for.
Frequently Asked Questions
Can a voice AI interviewer ask follow-up questions?
Conversational products can ask follow-ups based on the answer. Check this in a full demo, including an unclear answer and a candidate correction.
Do candidates need a phone call to use voice AI?
No. Some interviews use browser audio or video; others use telephone calls. Check the supported format and send candidates the right instructions.
Can voice interviews assess written communication?
Use a written task when writing is part of the job. Fabric supports chat-based roleplays as well as spoken interviews, so candidates can show how they would respond in the channel the role uses.