What Is an AI Interviewer? The Four Types You Might Actually Face

Software that conducts some or all of a job interview without a human present. In 2026 there are four distinct types. The prep strategy for each is different - knowing which one you are facing changes everything.

Platform-to-Type Mapping (April 2026)

Type 1: Text Chat Interview

The most accessible format. You type answers to behavioural questions in a chat-like interface. No camera, no microphone, no video. The AI evaluates your written text using Natural Language Processing (NLP), analysing sentence structure, word choice, response coherence, and behavioural indicators.

Representative platforms: Sapia.ai (formerly PredictiveHire), Mya (StepStone), and some Paradox/Olivia configurations. Sapia is the most widely deployed standalone text-chat interviewer in 2026, used heavily in Australian retail, hospitality, and airline sectors (Qantas, Woolworths) and expanding into US markets.

What it scores: Sapia explicitly does not use visual or auditory signals. It evaluates written language only: behavioural content, sentence completeness, specific examples, linguistic markers of the competencies it is measuring. Sapia also explicitly scans for AI-generated content via linguistic fingerprinting. Typing your ChatGPT response is flagged.

Format: Typically 5-7 behavioural questions. Sapia is untimed (you can take as long as you need per response), which benefits candidates with dyslexia or reading processing differences. Other text-chat platforms may impose time limits per question.

Prep implication: STAR-structured written responses of 150-300 words per question, with specific examples and quantified outcomes. Avoid generic language. Write in your own voice; generated text is detectable and scores lower on specificity regardless of detection.

Type 2: One-Way Recorded Video Interview

The dominant format in 2026. You record your answers to pre-recorded prompts (questions appear on screen and often play as audio). No human or AI is on the other end during your recording. You answer into your camera within a time limit, and the recording is submitted for algorithmic analysis followed by human review.

Representative platforms: HireVue, Willo, myInterview, VidCruiter, Modern Hire / Harver, Spark Hire, Talview. HireVue is the largest US player and is used across graduate schemes at Unilever, Goldman Sachs, JP Morgan, Hilton, and Intel, among many others.

What it scores (using HireVue as the archetype): Language content (keywords, completeness, rubric match), speech structure (sentence shape, discourse markers, specificity), pacing and speaking rate, and rubric alignment using Behaviorally Anchored Rating Scales (BARS). HireVue removed facial expression analysis in 2021. See the scoring mechanics page for the full breakdown.

Format: Typically 4-6 questions, 2-3 minutes per answer. Most platforms allow one re-take per question. Candidates are tier-banded (Top/Middle/Bottom) before human review on HireVue; recruiters see the scorecard alongside the video.

Prep implication: Tech setup matters (camera at eye level, lighting, audio). STAR structure is essential. Body language matters for the human reviewer who watches after the AI scores. See tech setup and body language for AI.

Type 3: Live AI Avatar Interview

The emerging 2026 format. A video-call-style experience where an AI-generated avatar conducts the interview in real time, asking questions and reacting to your answers. Unlike one-way recorded video, there is something on the other side - an AI that adapts follow-up questions based on your responses.

Current state: Uncommon in mainstream hiring. Technology providers (Tavus, Synthesia-backed tools, some HireVue experimental features) are developing these systems. A small number of employers in tech and high-volume retail are piloting live AI avatar interviews in 2026. Most candidates will not encounter this format until 2027 or later in mainstream corporate hiring.

Prep implication: Similar to one-way video for content. The real-time, adaptive nature adds conversational pressure. If you receive an invite that describes a “live AI interview” or “AI conversation”, treat it like a structured video interview with the body language elements of a live human interview.

Type 4: Assessment Hybrid

The video interview is one module within a multi-stage assessment. Common combinations: recorded video plus situational judgement test (SJT), video plus personality questionnaire, video plus cognitive ability test, video plus technical assessment (CodeSignal, HackerRank).

Representative platforms: Harver (formerly Modern Hire) volume-hiring suite, Sapia chat combined with SJT modules, Pymetrics game-based assessments combined with HireVue video, CodeSignal or HackerRank technical assessments with video intro.

Prep implication: The video component is one score among several. Do not over-invest in video preparation at the expense of the SJT or other modules. Read your invite email carefully to understand the full assessment structure.

What AI Interviewers Are Not

Clarity matters here because several different things use the label “AI” in the hiring process:

• xATS screening (Applicant Tracking System): keyword filtering of CVs before human review. Not an AI interviewer - this happens before you are invited to interview.
• xAutomated scheduling chatbots (Paradox/Olivia, Mya in some configurations): these handle scheduling, answer basic FAQs, and route you to the next stage. They are not conducting a scored interview.
• xSkills tests (CodeSignal, HackerRank, Vervoe): automated tests of specific technical or functional skills. Not a behavioural interview, not scored the same way.
• xReference check tools: automated digital reference checks are not AI interviews.

Brief Timeline (Key Moments)

Frequently Asked Questions